Google Cloud Next

What’s new in data agents: Supercharging your AI workflows

Mon, 15 Jun 2026 17:00:00 +0000

The rise of AI agents is fundamentally disrupting applications and analytical systems. Generic AI platforms don't usually have access to the context stored within enterprise databases. This is because traditional data architectures often lack context for agents across the data estate, which can lead to agents being inaccurate. They’re also prone to security gaps due to a lack of granular access controls.

Google’s Agentic Data Cloud is an AI-native system of action that includes both operational and analytical systems. By infusing AI across the entire stack — from custom silicon to frontier Gemini models — we provide a deterministic, template-driven developer framework that allows agents to ground their reasoning in real-time enterprise data with near-100% accuracy, as well as unified governance.

Today, we’re making it easier to develop agents, with a whole host of new data agents and tools: for business analysts within Conversational Analytics; for data scientists, engineers, and database admins with a series of Google-built Data Agents that provide greater automation and intelligence; and finally, for developers, with Data Agent tools that help you better integrate with today’s open agentic ecosystem.

1. Conversational Analytics

To support developers building agents using natural language, we’re announcing expanded support for Conversational Analytics across Data Cloud.

Conversational Analytics in BigQuery integrates a sophisticated AI reasoning engine directly into BigQuery Studio, helping data and business teams go beyond writing manual SQL, leveraging business context to ground answers using multimodal synthesis and deep-dive research. Agentic workflows, in preview for select customers, automate root-cause analysis, and schedule actions — turning enterprise data into proactive, actionable intelligence.

Create agents for faster data insights with Conversational Analytics in BigQuery

Conversational Analytics in Lakehouse, now in preview, extends the Lakehouse unified infrastructure, so users can query distributed data lakes across AWS, Azure, and Google Cloud using natural language. This makes it possible to combine insights across cloud platforms without moving a single byte of data.

Conversational Analytics in AlloyDB, Spanner, and Cloud SQL, now in preview, supports out-of-the-box conversational AI, making data accessible for everyone. AlloyDB, Spanner, and Cloud SQL users can start natural-language conversations with their databases to gain visibility into their real-time operational data and capture analytical insights.

Use Conversational Analytics to get answers from your operational data

Looker Embedded Conversational Analytics, now generally available, allows you to embed agents directly into your custom applications and internal workflows via a low-code iframe implementation, making it easier to ship production-ready, conversational AI within any application. Additionally, with the Conversational Analytics API in Looker, you can create multi-turn conversational workflows that offer AI-powered recommendations, while also verifying and explaining the underlying SQL query. We are also significantly upgrading Looker’s core Conversational Analytics agent, which is already GA, with superior reasoning and semantic grounding, helping to eliminate ambiguity.

Embed agents directly into your applications for conversational AI

2. New data agents

To help data professionals move from reactive data management to proactive intelligence, and business analysts better interact with their dashboards, we’re announcing a new set of data agents that bring automation, intelligence, and natural language capabilities into their daily workflows.

Data Engineering Agent, now generally available, automates the heavy lifting of building and maintaining data pipelines. It transforms natural language requirements into optimized SQL or Python code for BigQuery and Dataflow, while proactively identifying and fixing pipeline breaks. By suggesting schema improvements and partitioning strategies, it ensures your data foundation is scalable, reliable, and performance-tuned without manual trial and error.

Data Science Agent accelerates the path from raw data to production-ready models. It assists data scientists by suggesting relevant features, generating boilerplate notebook code, and automating the technical documentation process.

Database Observability Agent, in preview with select Cloud SQL, AlloyDB, Spanner, and Bigtable customers, proactively monitors database performance and continuously identifies potential issues before they escalate. It then delivers intelligent recommendations and multi-turn remediation workflows for fast, comprehensive troubleshooting and optimization. It provides performance analytics for the entire database fleet, helping you quickly identify performance optimization opportunities across databases.
Database Onboarding Agent, in preview with select customers, takes the guesswork out of database selection and deployment. By evaluating your stated requirements — from simple use case descriptions, to complex enterprise needs — it recommends the best Google Cloud database and guides you through provisioning.
Looker Dashboard Agent, now in preview, enables conversational interaction with data within dashboards. Users can ask natural language questions and receive context-aware answers within the dashboard. This feature also provides AI-generated summaries that highlight key takeaways and insights from the dashboard.
Conversational Analytics in Gemini Enterprise, now in preview for Looker, BigQuery, and Lakehouse, brings governed intelligence built by data practitioners directly to business leaders. It serves as a "front door" to the Google Data Cloud, allowing business users to consume agents built in BigQuery, Looker, or Lakehouse without needing to access technical consoles. By publishing these agents from Google Data to Gemini Enterprise, organizations provide a single, grounded interface for precision data exploration and immediate answers to the business users.

Deep Research Agent, now in preview, uses the Knowledge Catalog to solve high-stakes, multi-layered business problems. It moves beyond simple search to build comprehensive research plans that synthesize intelligence from internal documents, BigQuery tables, and the public web. The result is a detailed report with dynamic visualizations and verifiable citations, that respect enterprise privacy and user permissions all the while.

3. Tools for data agents

Open-source standards for agentic development provide developers building AI applications and custom agents with a unified framework to access data and tools consistently and securely. Today, we are announcing the following tools to help ground your agentic development initiatives:

Data Agent Kit: now in preview, provides a standardized suite of skills and tools directly within preferred developer environments (IDE/CLI), empowering data practitioners to discover, transform, and action data at scale using the prescriptive guidance from the Agentic Data Cloud capabilities.
Managed MCP Servers for Databases, now generally available for AlloyDB, Spanner, Cloud SQL, Bigtable, and Firestore, fully manages the infrastructure required to connect AI models securely to your data, so you don’t have to host, secure, or scale MCP servers yourself. Now, developers can provide their agents with up-to-date context from across our database portfolio, so that your AI models can reason and act upon your most up-to-date enterprise data.
Managed MCP Server for Looker, now in preview, allows any MCP client or agent platform to query Looker's semantic models, extending governed BI insights across third-party applications.

Access Looker semantic models through Managed MCP Server

MCP Toolbox for Databases 1.0, now generally available, has achieved a major stability milestone, giving you the confidence to build production applications. We also overhauled the documentation, making the platform significantly more approachable for both human developers and autonomous agents.
QueryData for Cloud SQL, AlloyDB, and Spanner, now in preview, turns natural language questions into database queries. It’s built natively into these databases, and provides near-100% accuracy for natural language to SQL conversions through metadata, query examples, and evals.
Universal Commerce Protocol (UCP) Analytics powered by BigQuery, now in preview, enables merchants and developers to stream real-time events from UCP directly into BigQuery (see sample). This integration provides out-of-the-box observability for agentic commerce, allowing teams to monitor conversion funnels, track automated checkout performance, and identify system errors. By standardizing these metrics within BigQuery, businesses can bridge the gap between AI-driven transactions and existing business intelligence workflows.

Details on how to access the new agents and tools can be found from each of the documentation links on this page. Data agents are also available through Gemini Enterprise and the Google Cloud console.

What we announced in streaming AI at Next ‘26

Mon, 18 May 2026 16:00:00 +0000

Every device, user, and microservice generates data. Ingesting this data, extracting meaning and insights, and driving business decisions in real time has the potential to deliver transformational business value.The rise of agentic AI represents an opportunity for users to overcome the challenges inherent in real-time analytics. But while agentic AI has the potential to accelerate adoption, users face a new set of challenges with effectively leveraging real-time data:

Real-time context is hard to implement. Teams will choose to incorporate data from batch-oriented approaches, like periodic database syncs and scheduled refreshes. Agents have to either rely on stale data or require memory-intensive context windows. This “context lag” makes them ineffective for real-time agentic tasks like fraud detection, dynamic e-commerce recommendations, or autonomous supply chain adjustments.
Real-time systems are inflexible. Agentic tools lack the modularity to adapt to customer-specific requirements, forcing organizations to make difficult architectural choices. Data practitioners need a platform to meet them where they are, where they are free to make the tradeoff between latency, accuracy, and cost.

Google Cloud provides a tightly integrated, unified streaming data platform that delivers both fully managed, Google Cloud-native services, as well as open-source-compatible services, and that come together to power large-scale AI training and inference. The platform is comprised of five key services:

Pub/Sub: Highly reliable, serverless, and fully managed service for messaging and event streaming that’s integrated with BigQuery, Dataflow, and Cloud Storage. Pub/Sub is utilized by organizations like Anthropic.
Dataflow: A serverless engine for batch, streaming, and now agentic AI. Leading enterprise organizations like Palo Alto Networks use Dataflow, as do Google services like Waymo and Google Maps. For instance, Waymo cars use Dataflow to help it “see” the world, plan their routes, and predict obstacles. Before a car hits the actual pavement, it “drives” millions of miles in a simulator, with Dataflow generating training datasets and validating the models that are used for autonomous driving.
Managed Service for Apache Kafka: The fully managed way to run the popular open source streaming storage and data integration system on Google Cloud that’s highly reliable, secure, and cost efficient. Across the largest enterprises and startups, Apache Kafka serves as a staging location for critical training data and real time updates to AI agent context.
BigQuery: A unified platform for real-time ingestion and analysis. The Storage Write API provides high-throughput streaming into BigQuery and Lakehouse for Apache Iceberg tables with exactly-once delivery semantics and stream-level transactions. Additionally, BigQuery continuous queries enable real-time AI inference directly within the data pipeline by calling generative functions like AI.GENERATE_TEXT, allowing for immediate insights as data is ingested.
Bigtable: Google’s NoSQL real-time database for processing streaming data from Pub/Sub and Dataflow automatically using continuous materialized views, delivering results in seconds that are ready for low-latency serving using Bigtable’s in-memory tier.

Moving from insight to autonomous action

At Google Cloud Next, we announced a set of streaming AI capabilities to the Agentic Data Cloud, providing autonomous agents with instant context and enabling real-time actions, helping organizations feed real-time context to their AI agents.

For instance, imagine a supply chain agent that doesn't just monitor IoT data, but autonomously reroutes a shipment around bad weather, confirms new delivery windows with the receiving warehouse, and updates the customer's portal — all before a human supervisor is even aware of the problem. Consider a financial services agent that identifies a fraudulent transaction pattern, instantly freezes the account, communicates with the customer via their preferred channel, and initiates a new card shipment — all within seconds of the suspicious activity. Whether you’re creating embeddings on streaming data to power search, or building a sophisticated multi-agent fraud detection system, these new capabilities add powerful tools to your toolbox.

Let’s take a closer look at these new capabilities.

New streaming AI capabilities

At Next ‘26, we launched tightly integrated capabilities to our platform across three key areas:

1. Providing real-time, enriched context for agents

1.1. Pub/Sub AI Inference SMT (GA): You can now run inference on messages streamed through Pub/Sub. Data practitioners can choose any models available on Gemini Enterprise Agent Platform. Pub/Sub makes the inference call and appends the result to each message before sending it downstream, bringing Pub/Sub’s simplicity together with the Gemini Enterprise’s fully managed tools.

1.2. Pub/Sub Bigtable subscriptions (Preview): Stream Pub/Sub data directly to Bigtable. Pub/Sub Bigtable subscriptions directly materialize event data from a Pub/Sub topic into a Bigtable table, eliminating the need for custom pipelines and dramatically simplifying your streaming architecture. For instance, you can easily ingest vector embeddings into Bigtable to power semantic search workloads.

1.3. BigQuery continuous queries stateful data processing (Preview): BigQuery continuous queries can now perform complex correlations between multiple data streams using JOINs and calculate metrics over consistent time intervals with tumbling window aggregations. This enables sophisticated analysis, such as calculating 30-minute averages or correlating events across different streams, directly as data is ingested into BigQuery. Furthermore, you can integrate AI directly into your data pipelines by calling generative functions like AI.GENERATE_TEXT, as well as materialize continuous query SQL results into BigQuery tables or export them to operational sinks like Bigtable, Spanner, and Pub/Sub for real-time reverse ETL.

2. Direct agents to manage your resources

2.1. Model Context Protocol (MCP) support for Pub/Sub, Managed service for Apache Kafka, Bigtable and BigQuery (GA): Your agents can manage Pub/Sub,Managed service for Apache Kafka services, and BigQuery using fully managed MCP endpoints. Agents can also publish messages to Pub/Sub.

2.2. ADK integration (GA): Your agents can interact with your real-time data stored in Pub/Sub, Bigtable, BigQuery, or other Google Cloud services using pre-built ADK integrations. Developers can build agents acting on real-time context without having to implement complex configurations or plumbing.

3. Combine multi-agent systems with your data processing

3.1. Event-driven autonomous agents: As agents become core to our workflows, real-time data pipelines must evolve to incorporate them directly into the stream. We have enabled this capability by treating agentic logic as a first-class citizen within the Dataflow pipeline. You can now incorporate your agent code using the Agent Development Kit (ADK) and deploy it as a specialized node using the RunInference transform and the new ADKAgentModelHandler. Key advantages of this approach include:

- Massive scalability: Leverage Dataflow’s architecture to process high velocity events upstream and keep hundreds of agents sessions active simultaneously, each driven by specific incoming events.
- Pre-processing power: Dataflow handles the heavy lifting of complex data enrichment, delivering a “ready-to-act” context directly to the agent so it can focus on reasoning.

3.2. Dataflow Unified embeddings Sinks: We are introducing unified embedding generation directly within the data stream to eliminate “context lag”. You can now transform incoming data into high-dimensional vectors at low latency using Dataflow. These real-time embeddings are then seamlessly materialized into our expanded suite of high-throughput vector sinks, which now includes Cloud Spanner (featuring its new built-in vector search) and AlloyDB, providing you with an up to date vector database for semantic search needs as well as for your autonomous agents making RAG calls with an instantly searchable and perfectly synchronized long-term memory. This feature works with both remote and local models, for example Gemma.

As we continue to build out the platform, customers can expect to see even tighter integrations and more powerful capabilities. We look forward to seeing what you build with these new capabilities.

Meet the latest Database Center, now with Gemini-powered fleet intelligence

Mon, 11 May 2026 19:00:00 +0000

Managing a modern database fleet is both a scale and cognitive problem. As database estates grow, the effort required to monitor, troubleshoot, and optimize them often outpaces teams’ capacity, who find themselves fighting database issues in isolation, buried under a mountain of fragmented signals.

We designed Database Center as a single pane of glass that provides fleet-wide visibility across all Google Cloud managed database services. At Google Cloud Next ’26, we announced an AI-native manageability interface for Database Center powered by Gemini that is designed to reason across your entire Data Cloud. In this new era of database operations, Gemini acts as an expert teammate, replacing manual scripts and error-prone workflows with AI-driven observability. In this blog, we showcase several key innovations in Database Center that you can take advantage of today.

Gemini-powered enhancements

Fleet-level intelligence with Gemini-powered analysis: Go from reactive firefighting to proactive, fleet-wide AI analysis. Gemini correlates performance shifts across your estate, highlighting patterns and providing actionable insights with an option for detailed investigations for diagnosis and remediation. These features are in preview with select customers.

Generative views (coming soon): Database Center now offers a hyper-personalized, dynamic interface driven by natural language, moving beyond standard dashboards to surface only the most relevant insights. Users will also be able to iteratively update these views.

Integrating into developer workflows: Database Center APIs are now public and integrated with the Model Context Protocol (MCP), bringing fleet management directly to tools like VS Code and Gemini CLI, as well as enabling custom third-party dashboards.

Gemini-powered chat: A new conversational interface uses natural language to interact with the entire database estate, across services like Cloud SQL, Spanner, or Bigtable, for fleet-wide exploratory questions and contextual troubleshooting, including triggering Investigations for root-cause analysis and remediation.

Gemini-backed recommendation validation (coming soon): Trust is vital for AI recommendations. New Gemini-based recommendation validations are available for specific performance optimizations. Users can trigger a Testing agent to simulate impacts on latency, IOPS, or throughput before applying changes like new indexes or machine upgrades, enabling confident, automated optimization.

Other platform enhancements

There’s more to Database Center than just Gemini. Here are the other enhancements we’ve made to the platform.

BigQuery inventory and data affiliation: BigQuery reservations and datasets inventory are now integrated into Database Center. Reservations and datasets inventory provide a single unified view across Google Cloud Databases and BigQuery. Data affiliation, meanwhile,maps data flows between transactional databases and BigQuery, helping surface hidden dependencies for faster troubleshooting.

Fleet-wide slow query analysis: Instead of hunting through logs and multiple pages, Database Center now centralizes and explains slow queries across the entire fleet, helping you prioritize your remediation efforts with AI-assisted troubleshooting. You can sort query patterns across the organization based on CPU execution time, number of instances, average rows returns, etc., and investigate the most impacted queries first.

Observability for top resources: Instantly identify the top 10 resources by critical metrics (CPU, IOPS, latency, etc.) to jumpstart investigations before they impact users.

Intelligent maintenance policies: For Cloud SQL and AlloyDB, Database Center now provides a unified, intelligent view of fleet maintenance status and compliance across all resources. You can also receive maintenance window suggestions based on your unique usage patterns, preventing downtime during peak business hours.

Reporting: Configure Database Center to generate natural language summaries of fleet health and inventory, delivered directly to your inbox so stakeholders stay informed without ever needing to log into a console.

What are Database Center customers saying

“Database Center gives our teams a comprehensive view of our Google Cloud database fleet and enables proactive risk management across security, performance, and capacity. Some of our product teams are already integrating it into their daily standups to improve monitoring and response. With Database Center’s APIs and MCP tools, we will be able to embed real-time fleet health directly into application team workflows — combining Google’s signals with our internal context like team ownership and application mapping to make insights truly actionable. This reduces context switching, accelerates recovery times, and strengthens proactive ownership across our engineering teams. We’re excited to see how the product continues to evolve.” - Bogdan Capatina, Technical Expert in Database Technologies, Ford Motor Company

Get started with Database Center

The future of database management isn't just unified — it's intelligent. With these new Database Center features and capabilities, you can:

Reduce operational overhead: Eliminate management silos and the need for expensive third-party observability software.
Enjoy faster mean time to resolution (MTTR): Resolve cross-domain issues in minutes rather than hours through Gemini-led correlation.
Scale with confidence: End-to-end lineage monitoring and automated health checks minimize blind spots, so that your AI and apps are powered by fresh, reliable data.

You can access Database Center from Google Cloud managed database services console for Cloud SQL, AlloyDB, Spanner, Bigtable, Firestore and Memorystore. Database Center is enabled by default for users with the necessary IAM permissions at the desired hierarchy. It is available at no cost, although certain premium features, including Gemini-backed fleet performance/ inventory insights, cost recommenders and natural language chat require Gemini Cloud Assist. Advanced security and compliance monitoring requires a Google Security Command Central (SCC) subscription.

Get started with Database Center today:

New Bigtable in-memory tier for sub-millisecond read latency

Thu, 07 May 2026 16:00:00 +0000

In the high-stakes world of digital infrastructure, speed isn't just a metric — it’s currency. At Google Cloud Next ‘26 we announced the Bigtable in-memory tier, a breakthrough for our fully managed cloud database service that delivers:

Sub-millisecond read latency for time-sensitive data
~10x higher point read throughput per dollar, dramatically reducing TCO
Hotspot resistance, supporting up to 120,000 queries per second on a single row without breaking a sweat.

For more information see Bigtable performance documentation. Now, let's look at the impact Bigtable in-memory tier can have on your workload performance and operational processes.

The cache-miss nightmare: A familiar story

Imagine it’s 2:00 AM. Your promotional campaign just went viral, and traffic is spiking. Your database architecture, meanwhile, is a house of cards: a primary database struggling to keep up and a separate caching layer acting as a shield.

Suddenly, you're hit with a hot key problem: everyone is trying to access the same viral content. Your cache node saturates. You’re forced to upgrade to larger nodes or add read replicas. You and your team are exhausted. Not only are you managing two different systems, maintaining a complex cache-aside logic (and praying the data in the cache stays in sync with the database), but you also need to respond to the actual incident. To do so, you overprovision CPU to handle the peak, and add more RAM so that everything fits in memory, as well as to avoid cache-aside complexity. Now you’re paying premium prices for warm data that doesn't actually need to be in memory. And while your hypothetical throughput-per-dollar looks great on paper, 90% of your resources sit idle most of the time.

Enter Bigtable’s in-memory tier

The Bigtable in-memory tier ends this cycle. By bringing data tiering across RAM, SSD, and HDD into a single, unified service with a hybrid storage architecture, we've removed the middleman.

The result? You get the raw throughput and speed of a cache with the durability and scale that Bigtable was designed for. When that viral spike hits, Bigtable automatically moves hot rows into memory to handle the load. No CPU spikes, no performance degradation. If the traffic grows, so does your Bigtable cluster by giving you more in-memory read capacity. You no longer need to overpay for idle RAM or cache nodes; Bigtable intelligently manages your data, keeping only the hot data in memory and ensuring data consistency between in-memory tier and SSD storage.

The TCO benefits are tangible, but maybe the most important part is the peace of mind that comes with it — and that’s priceless.

A peek behind the curtain

Almost every database server uses memory to give CPU fast access to latency-sensitive, frequently accessed data such as indexes and Bloom filters. You might be wondering, what makes this announcement different?

The secret lies in Remote Direct Memory Access (RDMA), a high-performance networking technology that allows computers to transfer data directly from one machine's memory to another without involving either the system's operating system or CPU. Our architecture uses RDMA to provide a high-speed, direct path to server memory, and as a result, throughput and latency of in-memory tier isn’t bound by server CPU, translating to impressive benefits. Much like Data Boost enables direct disk access for heavy workloads such as ML training, RDMA provides high-speed, direct memory access for real-time processing.

Imagine you’re running a popular social media site where 98% of users have fewer than 250 followers, while your most popular users have over 100 million. 60% of users post less than once a week, and the top 10% of users generate 80% of the content. And while a typical post receives 500 impressions, popular ones receive tens of millions.

To efficiently address this use case you will want data tiering that will likely look like this:

Memory: Content from profiles of users with large followings
SSD: Recent content, active user profiles
HDD: Older content, inactive user profiles

Luckily this is very easy to accomplish in Bigtable. Simply enable in-memory for your cluster and use a memory-enabled application profile when issuing your database requests to automatically manage the hot data lifecycle. You can also set an age-based policy to tier cold data to infrequent access. With this setup, when a piece of content is read, it gets promoted to the memory tier from persistent storage and stays there until it is evicted to make room for more recently read items. It is hands-free; even if a post from 5 years ago makes a viral comeback all of a sudden, you don’t have to worry about it.

But let’s say you want more fine-grained control of what you cache. You have a list of popular content creators and want to limit memory usage to only that small subset of their posts. Simply route the traffic for those users via the memory-enabled app profile, and for the rest of the content use an app profile that isn’t memory-enabled.

The cache-miss nightmare, revisited

Let’s rewind and replay our cache-miss scenario, but with Bigtable’s in-memory tier enabled. It’s 2:00 AM Sunday morning. Your promotional campaign just went viral, and traffic is spiking, you need to serve an additional 80K reads per second for the next hour. You don’t get paged. You wake up at 11 AM to the sound of birds chirping and enjoy a peaceful breakfast. It’s a beautiful day. The only sign that traffic spiked between 2-3 AM is that your bill shows an extra $0.40 charge.

Power laws govern distribution of requests for applications in a wide range of industries so scenarios like this are not limited to social media. For example, stock exchanges trade several thousands of securities but the top 30 most active stocks typically represent more than 40% of the total daily trading volume. At the same time, the most recent data points (last trade, ask/bid price) are requested frequently with an expectation of low latency responses, while historical data is accessed much less frequently, and has a rather forgiving latency budget. Let’s break down this example into Bigtable data tiers:

Memory: Most recent price of securities for most sought out stocks
SSD: Recent history, aggregated metrics (hourly, daily, monthly etc.)
HDD: Older data, raw events like individual trades

The list of possible use cases for this capability is long. Automated trading systems access latest prices from memory, while retail investors build their candlestick charts from data on SSD, and quants access historical data on HDD using Data Boost to backtest their models. All in one database, without interfering with each other. You can replace financial time series with telemetry data, sensor networks, digital twins and the story wouldn’t be much different.

Nor does using Bigtable’s in-memory tier interfere with other enterprise features like high availability, scaling, auditing, governance, and access controls, which typically introduce significant overhead. Achieving sub-millisecond latency despite these enterprise requirements is extremely impressive. By optimizing our clients and network, we’ve also successfully reduced p50 SSD latencies to below 2 milliseconds.

Get started with Bigtable Enterprise Plus

The Bigtable in-memory tier is available exclusively as part of the new Bigtable Enterprise Plus edition, which offers many additional features and is designed for organizations that demand the highest levels of performance, and management efficiency.

Elevate your stack to Bigtable Enterprise Plus and in-memory capabilities today so you can stop managing infrastructure and start building the future!

Learn more

Learn more about Bigtable Enterprise Plus edition and its capabilities beyond the in-memory tier. Try it out by heading over to Google Cloud console and creating new clusters upgrading existing ones.
If you’re new to Bigtable, you can now experience Google’s pioneering NoSQL database with the new Bigtable Free Trial. Get a dedicated Enterprise Edition node, 500GB storage, and a guided tour of Bigtable.
For more detailed information on getting started, technical specifications, and regional availability, visit the official Bigtable product page.

260 things we announced at Google Cloud Next '26 – a recap

Fri, 24 Apr 2026 19:00:00 +0000

Google Cloud Next ‘26 took place this week in Las Vegas, and the energy was incredible as we welcomed over 32,000 leaders, developers, and partners to explore the Agentic Era with us.

Across three keynotes, 25 spotlights, 700+ breakout sessions, 490+ sponsorships, our Showcase & Expo tech playground, and innumerable impromptu networking and collaboration moments, you could really feel the excitement as organizations move from pilot to production at a scale the world has never seen.

During our opening keynote and throughout the week, we shared our blueprint for the agentic enterprise and all the latest innovations for businesses to meet this moment. To name just a few:

- Gemini Enterprise Agent Platform, our new, comprehensive platform to build, scale, govern, and optimize agents
- Gemini Enterprise app brings the best of Google AI to every employee, for every workflow and includes new features that let every employee build with AI
- Google Cloud’s AI Infrastructure offers the industry’s widest variety of compute options and new innovations at every layer of the AI Hypercomputer, including our new 8th generation TPUs
- Agentic Defense is delivered via a cybersecurity platform that combines Google’s Threat Intelligence and Security Operations with Wiz’s Cloud and AI Security Platform to detect, prevent, and respond to threats
- Agentic Data Cloud introduces a new, AI-native architecture that allows your data to be utilized at the speed and scale required by agentic AI
- Agentic Taskforce redefines engagement and productivity with new innovations across our Gemini Enterprise for Customer Experience and Workspace products
- And so much more

That was just the tip of the iceberg from a jam-packed week of programming. So, we’ve pulled together a comprehensive summary of all 260 announcements we made across our products, partners, and customers at Next ‘26 below.

Product news

AI models, platforms and agents

As you move toward an autonomous enterprise, where agents act with the same independence and reliability as members of your team, you need a new agentic foundation that can sustain your trust. Here’s how we do it:

1. Gemini Enterprise Agent Platform: Build, scale, govern, and optimize agents. As the evolution of Vertex AI, it offers the model selection, model building, and agent building capabilities that customers love, plus new features for agent integration, DevOps, orchestration, and security.

Within Agent Platform, we announced many new features that help you move from managing individual AI tasks to delegating business outcomes with total confidence:

2. Agent Development Kit (ADK): Unlocks more powerful reasoning by organizing agents into a network of sub-agents. This new, graph-based framework allows you to define clear, reliable logic for how agents work together to solve complex problems.

3. Agent Studio: Go from building simple prompts to deploying complex agents in Agent Studio. Once you're ready for deep customization, export your logic directly into ADK to continue development in a full-code environment.

4. Agent Runtime: Delivers sub-second cold starts and allows you to provision new agents in seconds.

5. Agent Sandbox: Now available for everyone, provides a hardened environment to safely execute model-generated code and perform computer-use tasks like browser-based automation without risk to your host systems. Read more here.

6. Agent Memory Bank: Personalizes interactions by dynamically generating and curating long-term memories from conversations. Using new Memory Profiles, agents can recall high-accuracy details with low latency, so as not to lose context. Read more here.

7. Agent Sessions: Stores and manages history. With Custom Session IDs, you can use your own unique identifiers to track sessions and map them directly to your internal database and CRM records. Read more here.

8. Agent Identity: Assigns every agent a verifiable identity in the form of a unique cryptographic ID. This creates a clear, auditable trail for every action an agent takes, mapped back to defined authorization policies. Read more here.

9. Agent Registry: Maintains a central library of approved tools. It indexes every internal agent, tool, and skill, simplifying discovery and ensuring only governed, approved assets are available to your users. Read more here.

10. Agent Gateway: Manages your agent fleet from one control point, providing secure, unified connectivity between agents and tools across any environment, while enforcing consistent security policies and Model Armor protections to safeguard against prompt injection and data leakage. Read more here.

11. Agent Anomaly Detection: Detects suspicious behavior in real time and uses statistical models and an LLM-as-a-judge framework to flag unusual reasoning. It works alongside Agent Threat Detection to provide visibility into malicious activity, such as reverse shells or connections to known bad IP addresses.

12. Agent Security dashboard: Uncovers vulnerabilities automatically. Powered by Security Command Center, it unifies threat detection and risk analysis, allowing teams to map relationships between agents and models, automate asset discovery, and scan for vulnerabilities in the underlying operating system and language packages. Read more here.

13. Agent Simulation: Simulates realistic conversations and tests agents against human-like synthetic user interactions and virtualized tools in a controlled environment. Agents are automatically scored based on task success and safety across multi-step conversations. Read more here.

14. Agent Evaluation: Tracks live performance to continuously score agents against live traffic using multi-turn autoraters that can evaluate the logic of an entire conversation, not just a single response. With turnkey dashboards and Agent Observability, you can visually trace complex reasoning to debug issues as they happen. Read more here.

15. Agent Optimizer: Automatically clusters real-world failures and suggests refined system instructions to improve accuracy. Read more here.

16. Long-running agents: Execute complex business processes that can work autonomously in secure cloud sandboxes, giving agents the ability to orchestrate business logic, write code to build custom tools, and complete multi-step work — without needing constant prompting. Read more here.

Then there’s Gemini Enterprise, the end-to-end system for the agentic era, which includes the Gemini Enterprise app and the following new capabilities:

17. Agent Designer: Uses the same no-code agent designer experience of Agent Platform and lets employees build sophisticated schedule- or trigger-based agents using any enterprise connector, and providing a virtual flowchart of your agent.

18. Inbox in Gemini Enterprise: Provides a central location to monitor, guide, and help manage all of your agent activity, including long-running agents. Notifications are categorized into actionable groups like "Needs your input," "Errors," and "Completed.”

19. Projects: Create a dedicated space where the agent’s memory is confined to the files and conversations your team adds, and connect it to data sources like Google Drive, NotebookLM, and Google Group Chats.

20. Skills: Create simple shortcuts using an “@” mention for repetitive tasks like applying brand guidelines, formatting a report, or accessing specific data.

21. Canvas: An interactive editor directly within Gemini Enterprise that lets teams easily create and edit Docs and Slides, and even export to Microsoft 365 files, within the same experience.

22. Enhanced Deep Research agent: Executes complex, multi-step research plans that autonomously synthesize information from the open web and internal company data.

23. Dynamic and interactive Agent UI experiences: Move beyond simple text chat with new support for the Agent-to-UI protocol (A2UI), enabling custom agents to dynamically generate rich, native UI components directly within the Gemini Enterprise app.

24. High-code agents: Allow your engineering teams and IT leaders to build sophisticated agents with full code and publish them directly into the Gemini Enterprise app for the business to use.

25. Data Insights agent for unified analysis: Uncovers deep patterns across your entire data landscape.

26. Integrations for Bring Your Own Model Context Protocol (BYO-MCP): Allows admins to securely connect the Gemini Enterprise app directly to your own custom or third-party business tools, internal workflows, and custom-built applications.

27. Cross-platform unified intelligence: Delivers unified intelligence across a fragmented data landscape where data lives in Google Workspace, Microsoft 365, or other business applications.

You can try the Gemini Enterprise App today by signing up here.

Our incredible AI ecosystem is also growing with Agent Gallery, where customers can browse a highly vetted set of agents built by our partners.

28. Agent Gallery: Provides access to new third-party agents from partners including Accenture, Adobe, Atlassian, Deloitte, Lovable, Oracle, Palo Alto Networks, Replit, S&P Global, Salesforce, ServiceNow, Workday, and more.

AI and compute infrastructure

We announced the eighth generation Tensor Processing Units, or TPUs, which consist of:

29. TPU 8t: The training powerhouse, specifically designed for high-throughput AI workloads, delivering nearly 3x higher compute performance than the previous generation to shrink training timelines for massive models. Learn all about TPU 8t here.

30. TPU 8i: A breakthrough system for inference and reinforcement learning (RL) workloads, engineered to deliver the ultra-low latency and up to 80% better performance-per-dollar advantage for agentic workflows and Mixture of Experts (MoE) models compared to the previous generation. Go deep on TPU 8i here.

31. Native PyTorch support for TPU, or TorchTPU: With TorchTPU, you can run models on TPUs as they are, with full support for native PyTorch features like Eager Mode. Now in preview with select customers.

Whether you need a fully disconnected deployment, or want an integrated, Google-managed environment on your own hardware, Google Distributed Cloud (GDC) is the foundation of our sovereign neocloud architecture. Here’s the latest:

32. NVIDIA Blackwell GPUs: NVIDIA Blackwell (NVIDIA HGX B200) and NVIDIA Blackwell Ultra platforms (NVIDIA HGX B300) GPUs accelerate AI performance, leveraging fifth-gen NVIDIA NVLink to deliver data-center scale bandwidth directly to your environment.

33. New machine families: In addition to the N2 and N3 machine families for general-purpose workloads, GDC now supports the new A4 machine family for inference, and the memory-optimized M2 and M3 machine families for workloads that require higher memory-to-vCPU ratios.

34. Enhanced storage: GDC now supports 6PB object storage per zone — 6x the previous storage capacity. In addition, it now offers 30 IOPS/GB per zone, a 10x performance boost.

We also announced several new instances and features across our Compute Engine family:

35. A5X with NVIDIA Vera Rubin platform: A5X will be powered by NVIDIA Vera Rubin NVL72 when it becomes available later this year, including concepts from the open-source Falcon networking protocol. Learn more here.

36. Google Axion N4A: Google’s custom Arm-based Axion CPU achieves up to 2x better price-performance than comparable current-generation x86-based VMs for cost-sensitive workloads, now GA. Read more here.

37. Google Axion N4A supports GKE Agent Sandbox. GKE Agent Sandbox lets agents safely execute untrusted code and tool calls, and running on Axion N4A instances provides up to 30% better price-performance than the next leading hyperscale cloud provider. Now GA. Read more here.

38. Google Axion C4A.metal: Our first Axion bare metal instance is designed to power Android development, automotive simulation, CI/CD pipelines, security workloads, and custom hypervisors. Now in preview. Read more here.

39. C4 instances support for Intel Xeon 6 (Granite Rapids): Achieve high-performance for AI workloads with Intel AMX and native FP16 support. C4 VMs are available with Intel Xeon 6 processors across all shapes. Read more here.

40. Flexible CUDs for more VM families: Flex CUDs are now available for memory-optimized (M1-M4) and HPC-optimized (H3, H4D) VM families, as well as Cloud Run. Learn more here.

41. C4N VMs: For high-volume network applications, C4N with Titanium adapters provides nearly 400 Gbps of VM-to-VM bandwidth, and C4N with Hyperdisk Extreme provides 25 GiB/s of block storage throughput and nearly 1M IOPS. Now in preview; sign-up here.

42. M4N instances: Designed for memory-intensive databases, M4N with Hyperdisk Extreme offers 26.57 GiB of RAM per vCPU. Sign-up for the preview here.

43. Z4D instances: Optimize I/O-intensive workloads such as SQL, NoSQL, and vector databases with up to 84 TiB of high-performance local SSD directly on the node. Z4D virtual machines and bare metal instances will be in preview soon.

44. Hyperdisk Balanced improvements: Hyperdisk Balanced High Availability now offers a 4x performance improvement for high-availability databases like SQL Server or PostgreSQL by dynamically routing full-disk performance to the active VM. Learn more here.

45. Hyperdisk ML performance improvements: With 2 TiB/s of aggregate throughput (up from 1.2 TiB/s), Hyperdisk ML helps eliminate AI storage bottlenecks, with more than 200x higher throughput per disk than competitive offerings. Learn more here.

46. Hyperdisk Exapools: For large-scale training, Hyperdisk Exapools offer the highest aggregate block storage performance and capacity per AI cluster of any hyperscaler. Now GA, learn more here.

47. Z4M: For distributed parallel file systems and large-scale AI/ML workloads, Z4M provides up to 168 TiB of local SSD, up to 400 Gbps of network bandwidth, and support for RDMA. Z4M VMs and bare metal instances are expected in preview in Q3 2026.

If you want help deploying agents in your environment, you can partner with our experts for a diagnostic of your data and infrastructure by signing up here.

Agentic Data Cloud

As AI shifts from reactive intelligence to proactive action, legacy data architectures are breaking down under the agentic load. Our Agentic Data Cloud closes the gap between thinking and doing, merging analytical history with transactional power in a real-time loop, so your agents can securely act on your behalf. The Agentic Data Cloud comprises many products and platforms, starting with:

48. Knowledge Catalog: A universal context engine that maps and infers business meaning across your data estate using aggregation, enrichment, and search to help agents execute tasks accurately. Read all about Knowledge Catalog here.

49. Deep Research Agent powered by the Knowledge Catalog: Synthesizes business data, internal docs, and web research to autonomously answer your hardest, most complex questions.

50. Zero-copy federation: Provides direct access to data and context across applications, operating systems, and AI platforms — including Palantir, Salesforce Data360, SAP, ServiceNow, and Workday. Read more here.

51. LookML Agent: Autonomously reads strategy documents, spreadsheets, and reports to generate business-ready semantics, ensuring AI and analysts use the exact same enterprise definitions.

52. BigQuery measures: Redefines data consistency by embedding programmatic business logic directly into the SQL engine, ensuring calculations are reusable, accurate, and governed.

53. Data Products: Packages data assets with built-in intent, SLAs, and governance constraints. These self-contained intelligence blocks ground agents reliably in production.

54. Reusable data quality rules: Scales consistent, high-quality data governance by combining reusable, expert-defined test logic across multiple assets to ensure trusted data.

55. Structured approval workflows: Validates new business terms and data products with structured workflows, ensuring changes are verified by domain experts before reaching AI.

56. Column-level lineage: Tracks exactly where your agent's truth originated. We’ve expanded end-to-end lineage tracking to include Looker, plus column-level lineage for BigQuery and Spark.

An AI-native, cross-cloud lakehouse is another foundation element of the Agentic Data Cloud. Here’s what’s new in ours:

57. Cross-cloud Lakehouse (formerly BigLake): Breaks down data silos with a borderless lakehouse. Powered by Iceberg REST Catalog, agents seamlessly access data across AWS, Azure, and a vast partner ecosystem.

58. Lakehouse catalog federation: Enables easy discovery, analysis, and zero-copy sharing of your data seamlessly across AWS Glue, Databricks, Snowflake, and SAP.

59. Lightning Engine for Apache Spark: Accelerates your agentic data science workloads with an engine that delivers up to 2x the price-performance over proprietary market alternatives. Read more here .

60. Lakehouse federation for AlloyDB: Enables live joins between your transactional data and massive historical analytical insights without moving data or building complex ETL pipelines.

61. Managed Iceberg storage & REST Catalog: Benefits from automatic management and multi-table transactions, plus read/write interoperability across BigQuery, Spark, and OSS engines.

62. Cross-cloud caching: This intelligent cache stores cross-cloud data on the first read to slash massive egress fees and accelerate follow-on queries for AWS and Azure data.

For practitioners, we announced several products with enhanced automation:

63. Google Cloud Data Agent Kit: A portable suite of skills, MCP tools, and plugins that turns environments like VS Code and Gemini CLI into native data workspaces that autonomously orchestrate outcomes. Read more here.

64. Data Science Agent: Simply state your goals in plain English to automatically execute complex plans for loading, cleaning, and visualizing data in BigQuery notebooks. Read more here.

65. Data Engineering Agent: Autonomously builds complex Dataform pipelines from scratch while enforcing governance rules and testing to keep bad data out of your production systems. Read more here.

66. Database onboarding and observability agents: Lets agents guide you through database provisioning, or proactively monitor health and execute automated troubleshooting across your database fleet. Read more here.

67. Conversational Analytics: Empowers teams to query complex datasets using natural language and publish custom analytical agents directly into Gemini Enterprise. Read more here.

We also announced new products and features across our database portfolio:

68. Spanner Omni: This edition brings Spanner’s globally consistent, multi-model database beyond Google Cloud. Run it anywhere — across multiple clouds, on-premises, or on your laptop. Read more here.

69. Spanner Columnar Engine: Accelerates analytical queries up to 200x on live operational data using vectorized execution, without impacting your production transactional workloads.

70. Managed remote MCP servers for databases: Securely manages the infrastructure to connect AI models directly to your operational data, eliminating the burden of hosting MCP servers.

71. Vibe coding integrations with AI Studio: Agent-led workflow that generates live applications from simple text prompts, easily connecting to trusted database services.

72. Tools for Data Agents: Modular building blocks that provide ready-to-use functions, like high-accuracy text-to-SQL, giving custom agents secure and direct database access.

73. Bigtable in-memory: A new hybrid storage architecture natively integrates frequently accessed data in-memory to deliver sub-millisecond read latency for real-time apps.

74. Memorystore for Valkey 9.0: Provides a fully managed migration path from self-managed Redis, complete with enterprise-grade security features like Access Control Lists (ACLs).

75. AlloyDB AI-powered search at scale: Scales enterprise search to 10 billion vectors using Google’s ScaNN index, offering up to 6x faster queries than standard PostgreSQL.

76. New AI functions for AlloyDB: Uses optimized modes for ai.if, ai.analyze_sentiment, and ai.summarize to embed Gemini’s powerful reasoning directly into database queries.

77. Firestore full-text and geospatial search: Google’s search technology is now natively integrated into Firestore, delivering highly relevant keyword and geospatial queries.

78. Move-to-managed migrations: Move your PostgreSQL databases to Cloud SQL or AlloyDB easily and securely with Database Migration Service, now featuring automated networking, support for consolidating multiple sources in parallel into a single destination, migrating tables without primary keys, and many more capabilities.

79. Database Center with BigQuery: Monitors your entire data estate — from transactional databases to analytical warehouses — from a single management plane powered by Gemini fleet analytics.

BigQuery, now more than ever, is the linchpin of modern, agentic data analytics, and includes many new capabilities:

80. Reverse ETL for BigQuery: A one-click solution to sync heavy analytical insights from your lakehouse back to AlloyDB or Spanner for low-latency, high-QPS serving to end users. Read more here.

81. BigQuery Graph: Maps entities, relationships, and business logic directly within your data platform to anchor AI agents in a governed reality for complex reasoning tasks.

82. BigQuery fluid scaling: Executes highly variable, agentic workloads with per-second billing. This premier autoscaling model instantly adjusts resources.

83. ObjectRef: Processes unstructured data alongside structured data using SQL and Python. Build rich, multimodal context on your Knowledge Catalog without moving data.

84. AI.PARSE_DOCUMENT: Part of BigQuery AI, simplifies complex document processing with a single SQL function that automates OCR, layout parsing, and chunking using Gemini’s layout parser.

85. TabularFM model: Part of BigQuery AI, brings high-quality, zero-shot regression and classification natively to BigQuery without requiring extensive feature selection or model management.

86. Optimized mode for SQL-first AI: Part of BigQuery AI, trains task-specific models on the fly for functions like AI.CLASSIFY and AI.IF, delivering up to a 230x reduction in token consumption and cost.

87. BigQuery-native Gemma embeddings: Part of BigQuery AI, generates high-quality vector embeddings at scale on standard CPUs directly within your data warehouse to simplify enterprise AI preparation.

88. Autonomous embedding generation: Part of BigQuery AI, fully manages pipelines for unstructured data, automatically keeping vector indexes in sync asynchronously as new data is ingested at BigQuery scale.

89. BigQuery hybrid search: Part of BigQuery AI, unifies semantic and full-text search into a single powerful function to deliver superior retrieval precision for RAG and complex exploration tasks.

90. Python UDF: Part of BigQuery AI, enrich, transform, or clean data using fully managed Python scalar functions that autoscale serverlessly to millions of rows directly in BigQuery.

91. Connected Sheets with TimesFM: Part of BigQuery AI, brings the scale of BigQuery to Google Sheets, featuring state-of-the-art zero-shot forecasting with the TimesFM model and robust anomaly detection.

92. Geospatial analytics datasets: Part of BigQuery AI, combines native Google Earth AI and roads management data directly with your enterprise data for deep, real-world agentic analysis without wrangling.

93. Proactive agentic workflows: With BigQuery, agents can go beyond basic Q&A to detect metric shifts, perform root-cause analysis, and deliver scheduled research briefings directly to your inbox.

94. BigQuery Agent Analytics plugins: Logs detailed interactions from frameworks like ADK and LangGraph directly to BigQuery to audit, troubleshoot, and optimize long-term agent performance.

95. BigQuery Studio productivity tools: Brings a contextually aware assistant, SQL and Visualization Cells, Files Explorer, and Git workflows directly into BigQuery.

96. Advanced runtime, small query, and history-based optimizations: These BigQuery optimizations have accelerated performance by 35% year-over-year while cutting processing costs.

97. New BigQuery workload management features: Granular cost controls arrive with reservation groups, dynamic assignments, and project-level slot controls, simplified by rules-based management.

98. Colab Data Apps: Transforms your notebook analyses into shareable, fully managed interactive Python applications that business teams can access directly from Data Studio.

With Looker Business Intelligence, you get not just agentic AI, but agentic BI. Here’s the latest:

99. Data Studio (formerly Looker Studio): Available in Free and Pro versions, this data visualization engine now also provides a unified interface to browse data sources, apps, and agents. Read more here.

100. Looker Dashboard Agents: Transforms static views into interactive hubs. Ask natural language questions within dashboards for instant summaries and context-aware, governed answers.

101. Agentic Workflows in Looker: Moves beyond reactive queries to proactive results. Agents monitor critical metrics for irregularities and identify hidden correlations so you can instantly address business shifts.

102. Embedded conversational experiences in Looker: Leverages Looker's API and SDK to integrate Gemini-powered, multi-turn conversational experiences into custom data applications and internal workflows.

103. Looker self-service explores: Blends personal files with enterprise data using an updated AI-Explore experience that automatically generates LookML, giving you the agility of a spreadsheet with enterprise governance.

104. Looker AI assistants: Simplifies BI with Visualization, Expression, and Insight Assistants. Use natural language to create charts, generate custom filters, and automatically summarize key trends.

105. Open BI and MCP: Empowers external AI agents to autonomously query and analyze data. We offer both open-source MCP via the MCP Toolbox and an all-new Google-managed MCP server that’s native to Looker for ease of management.

106. Continuous Integration: In Looker, automates SQL validation and content testing within your development workflow to identify potentially breaking changes before they reach production, for reliable model updates.

Application development, management and runtime platforms

Cloud Run, Google’s fully managed application platform, now supports:

107. Full-stack apps in Google AI Studio: Build and deploy apps vibe-coded in AI Studio, including a Firestore database and user authentication, to Cloud Run, now GA.

108. Integration with Gemini Enterprise Agent Platform: Through its integration with Cloud Run, Agent Platform helps you transition agents from experimental environments into fully managed, production-grade systems. Now in preview with select customers.

109. Fully managed remote MCP server: Enable easy deployments from developers and agents, now GA.

110. NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs: Run high-end inference on Cloud Run — up to 70B+ parameter models — without having to manage any underlying GPU infrastructure, and scaling to zero when the resource is not in use.

111. Billing caps: Define your maximum spend per month, after which point, Cloud Run resources are de-activated. Coming soon.

112. Cloud Run instances: Host long-running background jobs, in preview to select customers.

113. Cloud Run sandboxes: Spin up an ephemeral sandbox that’s strictly isolated from your agent code using a built-in sandbox tool, coming soon.

114. SSH: Gain secure shell access (SSH) directly into a running Cloud Run container, now in preview with select customers.

115. Cloud Run service bindings: Service-to-service communication for scalable microservices architectures, coming soon.

116. Ephemeral disk: Now in preview, ephemeral disk storage is created when an instance starts and deleted when it stops.

And for mobile and web developers, we added new capabilities to Firebase, our premier platform for client developers:

117. Firebase SQL Connect: Firebase Data Connect is now Firebase SQL Connect, letting you connect your mobile and web client apps to the power of Cloud SQL for PostgreSQL, now with support for realtime syncing, offline cache, and native SQL— all available via a no-cost trial.

118. Firebase integration with Application Design Center: The new Firebase integration with Application Design Center (ADC) offers a unified model for deploying and managing the resources used by your mobile and web client apps, such as Firestore, App Check, Authentication, and Firebase AI Logic, in the same way you manage the rest of your Google Cloud infrastructure.

119. Firestore Enterprise edition: Introducing native full-text search, geospatial queries, relational-style JOIN using subqueries, and a robust data manipulation language (DML) as part of the new query engine in Firestore Enterprise.

120. Firebase AI Logic: Build generative AI features directly into your mobile and web client apps with no server-side setup. New features, like auto function calling and JSON schema mapping, help reduce technical overhead and give you more control to optimize your resources and costs. Read more here.

121. Firebase Phone Number Verification: Helps onboard new users to your app by retrieving the phone number assigned to the SIM in a device over cellular or WiFi. Now supported by 10+ carriers across six regions.

Then, there’s Google Kubernetes Engine, or GKE, which has emerged as the platform of choice for complex, demanding AI and agentic applications. Here’s what’s new in GKE:

122. GKE Agent Sandbox: A GKE add-on based on the open-source Agent Sandbox controller project that manages isolated, stateful, and single-replica workloads on GKE, optimized for AI agent runtimes. Learn more here.

123. GKE hypercluster: A single, conformant GKE control plane to manage millions of accelerators across Google Cloud regions, in private GA.

124. Predictive latency boost in GKE Inference Gateway: Reduce time-to-first-token latency by up 70% by replacing heuristic guesswork with real-time capacity-aware routing — no manual tuning required. In preview. Learn more here.

125. Automatic KV Cache storage tiering across RAM, Local SSD, and Google Cloud Storage/Lustre, solving long-context memory bottlenecks. Learn more here.

126. RL Scheduler: Solve for the "straggler effect" and inter-batch tail latency, maximizing throughput via intelligent routing. More here.

127. RL Sandbox: Kernel-level isolation for tool-calling and reward evaluation with millisecond-scale provisioning.

128. RL Observability and Reliability dashboards: The deep visibility you need to troubleshoot and optimize the entire RL loop instantly, out of the box. See more here.

129. Scaling on custom metrics: Support for intent-based autoscaling on triggers besides CPU and memory. Learn more here.

130. Accelerated node and pod startup: GKE nodes now start up to 4x faster, while pod startup times have been slashed by up to 80%. Learn more here.

131. Rapid model loading: Leveraging the run:AI Model Streamer and Rapid Cache in Google Cloud Storage, models now load 5x faster, removing a traditional storage bottleneck. Learn more here.

Developers and operators, get ready for the next generation of Gemini Cloud Assist, our AI-assisted cloud operations platform. Be on the lookout for:

132. Redesigned Application Design Center: Using natural language and the power of Gemini, reduce the time from architecture design to deployment of new or existing multi-resource environments.

133. Support for gcloud, kubectl, and Terraform: Automate infrastructure operations while using proactive multi-turn agents to troubleshoot and resolve incidents.

134. Proactive cost anomalies detection: A new FinOps agent analyzes spending spikes and generates granular cost reports on demand.

135. MCP servers: Get help wherever you work, including in your integrated development environment (IDE), CLI, or third-party tools with MCP servers for Gemini Cloud Assist’s design, operations, troubleshooting and optimization capabilities. Read more here.

Whatever your preferred Google Cloud management interface, you’ll find a variety of new features:

136. App-centric visibility for Unified Maintenance: Gives a consolidated view of maintenance events across your entire application landscape, moving from resource-level to business-level visibility. Learn more here.

137. App Lifecycle Manager: Now in preview, App Lifecycle Manager lets you design, deploy and operate Apps at scale across tenants with automated Day 2 ops and visibility, while feature flags decouple your feature releases from code deployments. Learn more here.

138. FinOps Explainability agent: Autonomously investigates the drivers of your AI-related cloud costs. This is in addition to new FinOps tooling which provides commercial auditability. Get started here.

139. Spend Caps: In private preview, FinOps and DevOps managers can set budgets and enforce cost boundaries at the project level for Google AI Studio, Gemini Enterprise Agent Platform, Cloud Run, Cloud Run Functions, and Maps. These caps alert and ultimately pause API traffic once your set budget is reached. Sign up for the preview here.

Customers

The real impact of these new capabilities is in the value they bring to our customers. We updated our running list of AI customer stories with more than 300 new entries, bringing the total to 1,302!

We also talked about several new customer use cases:

140. Capcom worked with Google Cloud to build agents that improve the gameplay experience, freeing developers for creative work. These specialized AI agents — including visual inspection, predictive and institutional knowledge agents — support the highly complex process of playtesting.

141. Citi Wealth is launching Citi Sky, an always-on AI-powered agent for customers to access market insights, act on opportunities, and engage with their advisors. Citi Sky is conversational and multilingual for a more intuitive, responsive and personalized wealth experience.

142. Citadel Securities built a scalable, cloud-based research environment that uses TPUs to run AI workloads up to 4x faster with 30% lower costs. Work that used to take days is now executed in minutes.

143. Deutsche Telekom announced MINDR, a multi-agent system built on Gemini models that enables autonomous, self-healing network operations across multiple domains to resolve network issues before they impact customers. MINDR reduces event management times by more than 95%.

144. GE Appliances is deploying more than 800 AI agents with Gemini Enterprise across its manufacturing, logistics, and supply chain operations.

145. Highmark Health scaled employee impact with Sidekick, an AI assistant that has delivered $27.9 million in value in 2025 alone by automating research protocols and providing "AI search proxies" for internal teams.

146. Home Depot created Magic Apron, a sophisticated digital agent built on Gemini Enterprise for Customer Experience that brings the brand’s "Orange Apron" expertise to every customer, and a new AI-powered phone agent that allows customers to skip complicated phone menus — identifying needs in under 10 seconds.

147. Macy’s created “Ask Macy’s,” a new digital shopping concierge and a leap forward in multimodal, agentic AI shopping powered by Gemini Enterprise for Customer Experience. Far beyond a chat bot, this multimodal agent is capable of handling text and images, and it also has a unique virtual try-on feature.

148. Mars selected Gemini Enterprise as the primary AI operating system for its global workforce. The company is providing its associates with a suite of agentic capabilities, including sophisticated AI assistants designed to handle complex, multistep tasks that can help them work more efficiently and creatively.

149. Merck will deploy a new agentic platform across their research and development, manufacturing, commercial and corporate operations. Core to the work will be digitizing data and boosting productivity for Merck’s 75,000 employees worldwide.

150. Papa John’s is the first user of the Food Ordering agent in Gemini Enterprise for Customer Experience. The agent features a unified voice-and-text AI ordering system, , while integration with the Omnichannel Gateway helps the agent remember customer preferences and past orders and get food to the door faster.

151. Tata Steel is rapidly scaling autonomous capabilities across its vast global organization, deploying a fleet of more than 300 specialized AI agents in just nine months to drive efficiency and precision across its global operations. These include a low-code platform for easy agent deployment and a digital assistant for informed decision-making.

152. Unilever is designing and deploying agents at scale — safely, securely, and globally — to transform how they serve billions of customers. Using Agent Platform, Unilever has developed a multi-agentic solution to help procurement teams make quicker and smarter buying decisions.

153. Virgin Voyages’ new Rovey agent is a personal concierge for the high seas, powered by Gemini Enterprise and Google Distributed Cloud. Rovey makes sure the crew has the latest, most on-brand information 24/7, and even on ships with limited connectivity at sea, Google Distributed Cloud keeps everything connected.

154. Vodafone Business is launching a managed detection and response service enabled by Google Security Operations, and the Vodafone Business AI Concierge with Google Gemini. The detection and response service will further protect Vodafone’s customers from increasingly frequent and sophisticated cyber threats.

Powering many of these organization’s new agents is Gemini Enterprise for Customer Experience, which comes with new agents and capabilities:

155. Shopping agent and Food Ordering agent: Bring new conversational sales and ordering capabilities direct to businesses and third-party chat and digital interfaces.

156. Omnichannel Gateway: Helps agents maintain context across web, mobile, and voice, so a company’s agents can offer more personalized service.

157. Agent Assist: For the most complex customer service situations, coaches employees to deliver fast and more accurate answers to customer questions with organizational data readily available through generative AI grounding.

Networking

We also announced a variety of enhancements to the broader Cross-Cloud Network family:

158. Virgo Network: A scale-out fabric for modern AI workloads. Built on high-radix switches that allow more ports per switch, it delivers massive bisection bandwidth and deterministic low latency for distributed training and serving workloads. Check out the deep dive.

159. Agent Gateway partnerships: Allows customers to integrate their trusted vendors to help secure agentic communications. Includes partner solutions from Broadcom, Check Point, Cisco, CrowdStrike, Exabeam, F5, Netskope, Okta, Palo Alto Networks, Ping Identity, Saviynt, Silverfort, Thales, and Zscaler. In preview soon.

160. Ambient networking: This new integrated data plane for GKE and Cloud Run provides service discovery, zero-trust access, and traffic management without the need for complex and resource-heavy sidecar proxies. Now in preview.

161. Service bindings: Enabled by ambient networking, service bindings automatically establish service-to-service connectivity between agentic applications and services.

162. Network Services Monitoring: Also enabled by ambient networking, the preview of Network Services Monitoring bridges application and network observability gaps for faster root-cause analysis and simplified troubleshooting.

163. Multi-region support in GKE Inference Gateway: Scale inference services across regions, for cross-regional failover, optimized utilization, and reduced global latency. In preview.

164. GKE Inference Gateway disaggregated serving: Choose between vLLM and SGLang for model serving by leveraging llm-d’s SGLang support. Now GA.

165. Accelerator network profiles: Set up accelerator-equipped GKE node pools with DRANET, and improve the bandwidth for distributed AI/ML workloads by up to 60%. Now GA.

166. Cloud Interconnect 400 Gbps circuits: Use up to 3.2 Tbps in a single connection. Now GA.

167. Partner Cross-Cloud Interconnect: GA for AWS; in preview soon for CoreWeave and Lumen.

168. Ultra Low Latency Solution for financial exchanges: In partnership with CME Group, this fully managed cloud environment will allow CME Group and its clients to migrate core trading systems to Google Cloud. Now in preview.

169. Cloud Network Insights: Network performance monitoring (NPM) and digital experience monitoring (DEM) enabled by Broadcom AppNeta to reduce the mean time to detect and mitigate network-related agent, application, and API issues. Now in preview.

170. Private Service Connect published services: Enable secure private global access to your managed services through 40+ Google and third-party published services. Read more here.

171. Private Service Connect endpoint-based security: Use granular authorization policies for producer-to-consumer service communications, now in preview.

172. Cloud Number Registry: Now in preview, this agentic IPAM solution finds free IP ranges, tracks utilization, and allocates resources. It also integrates with Infoblox Universal DDI for Cross-Cloud Network IPAM discovery and enforcement.

173. Hybrid Subnets: Migrate legacy workloads from on-premises to a VPC without needing to change hard-coded IP addresses. GA.

174. Cloud NAT: Connect IPv6-only workloads to private IPv4 destinations using the combined power of DNS64 and private NAT64. In preview soon.

175. Granular networking charge visibility: Cost Explorer and the new App Optimize API now provide attribution of associated data transfer costs to the originating resources for Google Cloud products, in preview soon.

176. Gemini Cloud Assist network security agent: Streamlines network security operations with policy generation, recommendations, and impact analysis. In preview soon.

177. Gemini Cloud Assist network agent: Optimizes workload placement for performance and reliability, Private Service Connect troubleshooting, and also provides advanced cost estimation for observability services. In preview soon.

178. Network observability MCP tools and agent skills: Partners can build agents to leverage connectivity tests, query VPC Flow Logs in natural language. In preview.

As part of Cross-Cloud Network, new capabilities in Global Front End include:

179. Global Front End Enterprise for simplified consumption: Combining capabilities from global Cloud Load Balancing, Google Cloud Armor, Cloud CDN, and Service Extensions with up to 15% lower TCO. In preview soon. Learn more here.

180. Post quantum cryptography (PQC): Helps secure your workloads with industry-standard algorithms that provide a layered defense against both classical and quantum adversaries. Learn more here.

181. Google tag gateway: This allows advertisers to serve tags from their own domain, which can significantly improve the accuracy and resilience of measurement signals. GA soon.

182. Image optimization in Cloud CDN: Deliver content that best fits your end users’ screens and save on bandwidth costs (in preview soon).

183. Cloud CDN GKE Gateway support: Enable and manage caching services using GKE APIs, now GA.

Cloud WAN is a fully managed, reliable global backbone to connect your enterprise, and includes the following new capabilities:

184. Expanded geographic reach: Network Connectivity Center’s site-to-site data transfer is now available in over 25 countries. Read more here.

185. NCC Gateway: Use third-party secure service edge (SSE) integrations from Palo Alto Networks (GA soon) and Symantec (preview).

186. Expanded Verified Peering Provider program: Get highly reliable internet connectivity to Google through 175+ providers worldwide. Read more here.

187. Last mile connectivity: Provision site-to-cloud private connectivity in minutes with preferred partners from the Google Cloud console. In preview soon.

We added new capabilities in our AI-powered network security offerings to protect against evolving threats.

188. Cloud Armor managed rules: Available in preview, and powered by Thales Imperva, detects Layer 7 application attacks and zero-day vulnerabilities.

189. Cloud Armor adaptive protection for Network Load Balancers and VMs: Advanced machine learning to L3/L4 traffic, to detect and mitigate volumetric DDoS attacks. In preview soon.

190. Cloud Armor visual rule builder: Makes custom rule creation easier with a simplified user experience, in preview soon. Read more here.

191. Cloud NGFW Internal Application and proxy Network Load Balancer support: Helps enforce consistent, service-centric security for abstracted services like GKE, Cloud Run, and Private Service Connect traffic. In preview.

192. Cloud NGFW project-level policies: Creates and manages Cloud NGFW endpoints, security profiles, and security profile groups at the project level. In preview soon.

193. Cloud NGFW Advanced Malware Sandbox: In preview later this year, helps defend against highly evasive zero-day threats. This capability is powered by Palo Alto Networks Advanced Wildfire, trained on data from more than 70,000 Palo Alto Networks customers to stop 99% of known and unknown malware.

We added new functionality to Network Connectivity Center (NCC), our hub for managing global connectivity:

194. Support for static routes: By using an internal load balancer as the next hop, you can integrate Secure Web Proxy and third-party network security virtual appliances. Now GA. Learn more here.

195. Support for privately used public IP (PUPI): Allows the exchange of PUPI IPv4 addresses with VPC spokes and producer VPC spokes, now GA. Learn more here.

Partners

A huge part of Google Cloud’s success is built on our partner ecosystem, which helps conceive, build, and implement solutions for our joint customers. To strengthen this partner partnership, we made a number of important announcements, most of which are detailed here.

196. $750 million innovation fund: Investing in partners with new models, tools, engineering support, and incentives to accelerate agent development and deployment globally for every business process, function, and industry.

197. Sell partner agents in Gemini Enterprise: The new Agent Marketplace with the Agent Gallery of Gemini Enterprise allows partners, startups, and more to sell their agents directly to other enterprise users.

198. 70-plus partner-built agents in Agent Marketplace: We’re launching the Agent Marketplace with more than 70 pre-built agents. Discover and manage specialized AI from partners like Accenture, Adobe, Atlassian, Deloitte, and more. Read more here.

199. Partnering on forward deployed engineers: Google Cloud will make its engineering talent available for the customers of select partners, like Accenture, Deloitte, and McKinsey, who are also establishing advanced engineering practices.

200. Deeper model integration: Palantir joins many of the leading SaaS and AI platforms already featuring Gemini access, while Box, DocuSign, Oracle, Salesforce, and SAP have added Gemini features in more of their agents and tools.

201. AI-powered features in Google Cloud Partner Network: A new partner agent and agentic features in the Earnings Hub and Partner Finder help partners amplify their impact for joint customers.

202. New Unified Data Foundation with SAP: Integrates SAP and non-SAP data into a bi-directional source of truth, helping power AI workflows and deliver data-driven value. Zero-copy data sharing opens up access to outside platforms securely.

203. More SAP news: We also unveiled new Gemini features in SAP Engagement Cloud, solution accelerators in Cortex Framework, and support for SAP Sovereign applications. Read all our SAP announcements here.

204. Oracle AI Database@Google Cloud expansion: Our recently launched Oracle data connectors are expanding to 20 regions, offer new AI capabilities, feature Oracle GoldenGate support for near real-time data replication to BigQuery and Knowledge Catalog integration, and more. Read more here.

205. Partners of the year: Celebrating hundreds of partners across specialties, industries, and geographies. See them all here.

Security

From security operations, to fraud detection, to cryptography and key management, there was a ton of security news at Next ‘26.

206. Threat hunting agent in Google Security Operations: Helps teams proactively hunt for novel attack patterns and stealthy adversary behaviors that bypass traditional defenses.

207. Detection Engineering agent Google Security Operations: Identifies coverage gaps and creates new detections for threat scenarios, reducing toil and transforming detection creation from a manual craft into an automated science.

208. Third-Party Context agent Google Security Operations: Enriches your workflows with contextual data from third-party content.

209. Google Security Operations: A robust cohort of new partner integrations for Google Security Operations has been designed to deliver high-fidelity security workflows right out of the box. Our latest participating Google Cloud Security integration ecosystem partners include Darktrace, Gigamon, and SAP. Read more here.

210. Model Armor: Now integrated with Agent Gateway, Agent Runtime, and Langchain available in preview, and Firebase, generally available, to help developers add inline enforcement and sanitization of agent traffic and interactions without the need to change code. These integrations expand Model Armor's protection against runtime risks such as prompt injections, tool poisoning, and sensitive data leakage across Google Cloud services and our AI portfolio. Learn more here.

211. Google Cloud Fraud Defense: Now generally available, we’re evolving reCAPTCHA into a comprehensive platform designed to secure the agentic web by discerning the legitimacy and authorization of bots, humans, and agents. Using the same scale and signals that protect Google’s own ecosystem, Fraud Defense will soon offer in preview agent-specific capabilities for human users and AI agents that can help secure the digital commerce journey, from account creation and login to payment and checkout. Learn more here.

212. Chrome Enterprise: Now in preview, AI-aware extension threat detections can surface advanced extension telemetry that helps security teams detect and respond to anomalous AI agent activity. Learn more here.

213. Chrome Enterprise: Generally available soon, new shadow AI reporting can help you gain visibility into the shadow AI landscape by flagging employee use of unsanctioned web-based AI and SaaS applications. Learn more here.

214. Identity and Access Management: To help achieve least privilege quickly and simply, we’ve streamlined our predefined roles catalog with easy-to-use administrator, editor, and viewer roles, such as the IAM role picker and the ability to re-authenticate sensitive actions.

215. Confidential Computing support for G4 VMs: In partnership with NVIDIA, Confidential Computing support for G4 VMs is available in preview globally, featuring NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs on Compute Engine Confidential G4 VMs, to help strengthen confidentiality and integrity for a wide spectrum of sensitive AI workloads. Learn more here.

216. C4 Confidential VMs: In partnership with Intel, we’re also introducing the preview of C4 Confidential VMs, bringing Intel TDX to 6th Gen Xeon processors to help protect diverse AI and analytics workloads while providing industry-leading compute density and performance.

217. Cloud Key Management Services (KMS): Available in preview, the new Confidential External Key Manager (cEKM) can give you the flexibility to host and protect external keys in any region and maintain verifiable control in a confidential environment.

218. Post-quantum cryptography (PQC): Available in preview, KMS Quantum Safe Key Imports can help you bring your own keys with quantum-safe algorithms.

219. Secret Manager: Generally available, the native integration of our Secret Manager with Agent Development Kit can help prevent password leaks and mitigate prompt injection risks.

Google has completed its acquisition of Wiz. By bringing two industry leaders together, we are building a comprehensive, AI-ready cybersecurity platform designed to protect your organization across all your cloud environments.

Wiz is constantly expanding and deepening your ability to adopt AI quickly and securely, with the following new capabilities:

220. Support for agent studios: Now supports Databricks as well as new agent studios like AWS Agentcore, Gemini Enterprise Agent Platform, Microsoft Azure Copilot Studio, and Salesforce Agentforce to help customers gain visibility however their teams choose to build. Read more here.

221. Security detections updates from Wiz Defend: We updated how we integrate security detections from Wiz Defend with Google Security Operations and Mandiant Threat Defense to help analysts more easily configure automatic threat information forwarding.

222. Vibe coding integration: Generally available in May, Wiz can help secure vibe-coded applications. This new integration runs Wiz security scanning directly inside the Lovable platform so vulnerabilities, secrets, and misconfigurations caught by Wiz are surfaced in Lovable's built-in security view.

223. Inline AI security hooks: Integrates Wiz directly into integrated development environments (IDEs) and agent workflows to evaluate prompts and scan AI-generated output instantly, injecting security guardrails before code is committed.

224. Wiz Skills: Help agent-based remediation by equipping coding agents and AI-native IDEs with full code-to-cloud context and validated attack surface findings from the Wiz Security Graph. These capabilities enable teams to trigger automated, agent-driven remediation workflows locally from the developer's individual IDE and globally at the repository and pull request level in your version control system.

225. AI-Bill of Materials: To help eliminate shadow AI, Wiz’s dynamic AI-Bill of Materials automatically inventories all AI frameworks, models, and IDE extensions across your environment. This helps provide visibility into what is writing code across your stack, allowing you to track sanctioned corporate tools like Gemini Code Assist and GitHub Copilot while simultaneously uncovering unapproved shadow AI plugins.

Storage

Enhancements to Google Cloud’s storage offerings span high-performance storage infrastructure, smart storage, new management capabilities, and expanded ecosystem offerings.

226. Rapid Bucket: As part of the new Cloud Storage Rapid family of high-performance object storage, Rapid Bucket delivers more than 15 TB/s of bandwidth, 20 million requests per second, and sub-millisecond latency in a single zonal bucket. Now GA; read more here.

227. Rapid Cache: Formerly Anywhere Cache, Rapid Cache accelerates bandwidth for bursty workloads, delivering an aggregate read throughput of 2.5 TB/s for existing buckets, with no code changes. Learn more here.

228. Rapid Cache ingest-on-write: Get up to 2.2x faster checkpoint restores, allowing training clusters to recover faster from interruptions. Learn more.

229. Increased throughput for Google Cloud Managed Lustre: With up to 10 TB/s of throughput, Managed Lustre writes and restores checkpoints 2.6x faster than other Google Cloud storage solutions. Learn more.

230. Managed Lustre Dynamic tier: Low-latency performance for intense AI workloads like training and checkpointing, offered at $0.06/GB-month.

231. Managed Lustre support for new TPUDirect and RDMA: Minimize latency by allowing data to bypass the host, moving directly to the accelerators. Learn more here.

232. Smart storage object context: Now GA, this metadata substrate adds structured, mutable, IAM-governed context to every object. Learn more.

233. Smart storage automated annotations: Unlock unstructured data with automated metadata annotation.

234. Cloud Storage MCP server: Read, write, and analyze Cloud Storage data using the standard MCP protocol. Learn more.

235. Storage Intelligence zero-configuration dashboards: Surface cost anomalies and integrate Security Command Center’s Data Security Posture Management (DSPM) data governance feature, to detect security vulnerabilities across Cloud Storage. Now in preview.

236. Storage Intelligence object events and bucket activity tables: Available in Insights Datasets, use these new tables to, for example, optimize bucket placement based on egress patterns, or quickly troubleshoot 429 errors.

237. Enhanced batch operations in Storage Intelligence: New change ACL and storage class operations make it simpler to act on billions of objects, and adds support for multi-bucket operations.

238. Google Cloud NetApp Volumes: With new Flex Unified, provision both block (iSCSI, NVMe/TCP) and file (NFS/SMB) on the same storage pool. New ONTAP-mode lets you bring your existing automation (Terraform, Ansible) and ONTAP APIs directly to NetApp Volumes.

239. Filestore for GKE: Developers building AI workloads on GKE can start small, with shares as small as 100 GiB, and scale capacity and IOPS independently. Tighter integration with Colossus provides more scale and enterprise capabilities.

240. Google Cloud Backup and DR agentic AI capabilities: Let these agents autonomously audit your backup estate and remediate coverage gaps, with new GA integrations for AlloyDB and Filestore.

Startups

Startups are building on Google Cloud for our models, purpose-built infrastructure, and unique security solutions. You can read more about some of the exciting new ones, and new startup programs, here.

241. Notable startups using Google Cloud: We announced dozens of new startups using our services or expanding with us, including Lovable, Notion, Proximal Health, Gamma, Mavvrik, Insilica, and Parrallel.

242. Access to NVIDIA hardware: A number of leading startups use Google Cloud for training and inference workloads on NVIDIA chips, including Thinking Machines, Interfact, Nucleus AI, Reflection, and Neuraldefend.

243. Global Gemini Startup Forum: This June, we’re hosting a two-day, in-depth gathering of founders from seed to Series A startups; attendees will engage with Google leaders and engineers, build and scale products, and share feedback.

244. Google for Startups AI Agents Challenge: Open to anyone, this global competition runs for six weeks and gives teams $500 in credits and tools — like the new Agent Platform — to build autonomous systems and compete for a $90,000 prize pool.

245. New gen media report: Our Future of AI: Perspectives on generative media for Startups report compiles exclusive interviews with eight startup founders, investors, and Googlers.

Workspace

Google Workspace with Gemini is where work happens for over 3 billion users and more than 13 million customers — from solopreneurs to small businesses to the Fortune 500. We introduced new innovations that transform how people work with AI, including:

246. Workspace Intelligence: Unified, real-time understanding to power agentic work, by understanding complex semantic relationships within Workspace apps (such as Docs, Slides, or Gmail), active projects, collaborators, and your organization's domain knowledge. See it here.

247. Ask Gemini in Chat: Acts as a unified command line for all of your work; simply state your goal and Gemini can complete tasks, generate documents or slides, or find files using just a description.

248. New Gemini capabilities in Sheets, Docs, and Slides: Build or edit entire spreadsheets using natural language, collaborate with Gemini in Docs to create infographics grounded in your data or edit based on comment feedback, and create full, editable slide decks in one shot.

249. AI Inbox and AI Overviews in Gmail: Get a clear, streamlined look at what’s most important in your inbox and use AI Overviews in Gmail search to get answers without digging.

250. New Gemini capabilities in Drive: Use AI Overviews and Ask Gemini to find information fast and get detailed insights about all of your files, and use Drive Projects to organize your files and emails, ensuring your colleagues and Gemini have the full context.

251. Third-party data in Sheets: Import third-party data from platforms like HubSpot and Salesforce to Gemini in Sheets, and create interactive visualizations — such as dashboards and kanban boards — directly within the new Sheets canvas.

252. Workspace skills: Organizations can now deploy agentic automation across workflows using skills. Collaboratively build and share skills in Workspace Studio, then invoke them wherever you use Gemini in Workspace.

253. Custom Avatars in Vids: Add branding elements to custom avatars — everything from a company logo on a shirt to a fully branded backdrop. And you can soon add avatars when you convert a presentation into a video.

254. Help from Gemini in every meeting: We’re expanding Take Notes for Me in Google Meet to capture automated meeting summaries and action items for any meeting, whether it’s in-person, or hosted on another provider like Zoom or Teams.

255. Workspace capabilities in the Gemini Enterprise app: Coming to private preview, customers will soon also be able to schedule Google Calendar meetings and seamlessly create, view, and edit Docs and Slides directly from canvas mode in the Gemini Enterprise app.

256. Auto browse with Gemini in Chrome: Workspace customers in the US can now turn on the new Gemini auto browse capabilities in Chrome Enterprise. Learn more here.

257. Workspace MCP Server: The new Workspace MCP Server, available in preview, lets developers bring advanced Workspace capabilities such as synthesizing Drive documents and drafting Gmail responses into their AI applications and agents. Learn more here.

258. Simplified agent governance: New agent governance controls, such as AI control center, agent management, and Workspace Studio controls, help you monitor, control, and audit agent access to your data in Workspace to reduce indirect prompt injection, oversharing, and data loss risks.

259. New sovereign controls and client-side encryption: To meet regulatory requirements, you can lock data processing and storage to the US and EU, with more countries supported in the future, such as Germany and India. And for your most sensitive data, client-side encryption lets you deny access to any agent and any entity, including Google.

260. Improved migration and interoperability features: Migrating your entire organization from Microsoft 365 to Workspace with a new cloud service that lets you easily move emails, files, and conversations. Improved interoperability features allow collaboration with customers and partners who use Office apps.

A big thank you to all of our customers, partners, and attendees who joined us on the ground or tuned in virtually. We’ll see you next year for Cloud Next ‘27 on April 13-15 in Las Vegas!

Day 2 at Google Cloud Next: A marathon developer keynote

Thu, 23 Apr 2026 13:00:00 +0000

At Google Cloud, every day is Developer Day, but none so much as day 2 of Google Cloud Next, when we hold the developer keynote. This year’s topic? An in-depth look at Gemini Enterprise Agent Platform. This year’s theme? Planning a marathon for 10,000 participants through the Las Vegas Strip.

OK, let’s run with it.

Gemini Enterprise Agent Platform: A warm up

As the evolution of Vertex AI, Agent Platform “allows you to build autonomous agents that proactively help users — and complete tasks independently,” said Brad Calder, President, GCP and SRE. The platform does so with a whole suite of tools and capabilities to build, scale, govern, and optimize your agents.

Brad then passed the baton to keynote emcees Richard Seroter, Chief Evangelist, and Emma Twersky, Developer Relations Engineer, for an in-depth run-through of the agentic marathon simulator.

The system uses three main agents: A planner to determine routes; an evaluator that assesses routes based on business and community requirements; and a simulator that takes the route, adding actors and randomized behaviors to test the impact on the city,

This, Emma said, turns out to be “a great example of how agents can help us plan, simulate, and think about solving a really big challenge.”

First off the blocks: Building the agent

Mofi Rahman, Developer Relations Engineer, came onstage to demo how the Agent Development Kit (ADK), Google Cloud remote Model Context Protocol (MCP) servers, and Agent Runtime provide the planner agent with the Instructions, Skills, and Tools it needs to improve the initial agent. By the end of the demo, the simulator had generated a route.

“The simulated route looks beautiful,” said Mofi. “Looks like the runners are going to get an amazing view of the entire Las Vegas Strip.”

An agent to evaluate the agent

Next, Ivan Nardini, Developer Relations Engineer, and Casey West, Architecture Advocate, showed us how to evaluate the agent, and build a UI for it. “We want to show you how to move from fragile, unpredictable agentic loops, to a rigorously evaluated network of experts that literally build their own UI,” Casey said.

They did so by deploying a separate model to judge the route, checking both deterministic (e.g., route length) and non-deterministic (e.g., community impact) criteria. For UI development, they showed off the Agent-to-User Interface, or A2UI, an open-source standard developed by Google that created an interface in a single shot. They also used the Agent-to-Agent Protocol, or A2A, and Agent Platform’s Agent Registry, to connect and see which agents are deployed.

“Think of Agent Registry as the DNS of your internet of agents,” Casey said.

Agents that never forget

Richard then mused about how to build agents that get better with time, that “take the learnings from the simulation and optimize for the next run.” Because the answer shouldn’t be to “cram raw text in every request we send back to our agents.”

To capture this learned knowledge, Agent Platform offers Agent Platform Sessions and Memory Bank, plus the ability to turn to tools like Spark or a database to retrieve more information, resulting in an even stronger simulator.

When agents go off course

Thus far, everything had gone swimmingly, but then Richard accidentally “broke” the simulator agent. That provided a perfect opportunity for Megan O’Keefe, Senior Staff Developer Advocate, to show off Agent Interoperability and Gemini Cloud Assist, and how to use them to debug agents at scale.

“With these autonomous agents, the production challenge isn’t just scaling the infrastructure, it’s managing the reasoning, the tool calls — all the places in the system where something can go wrong!” Megan said.

Megan used Agent Runtime trace view to see where the problem was, and using natural language, launched a Cloud Assist Investigation to explore logs and events, which pointed to a specific line of code as the offender. Megan then opened up her Antigravity IDE (powered by Gemini 3, and connected via MCP) to find the problem (an insufficiently run “event compaction” run) and to suggest a fix (add a token_threshold parameter to the event compaction config). She approved the fix and committed it to source, triggering a redeployment to Agent Platform. Problem solved!

Scaling the agents

To this point, all of the presenters had been showing off agent services running as Cloud Run services. Bobby Allen, Group Product Manager, then showed how to convert the apps to Google Kubernetes Engine (GKE), which provides greater control, as well as to use a customized Gemma 4 model, all by vibe coding in the Antigravity editor, which is connected to Cloud Assist. Along the way, Bobby also migrated the agents from GCSFuse to a high-performance Lustre file system.

Closely related to scaling is sharing — making agents available for the world to use and build on. Ines Envid, Senior Director, Product Management and Jason Davenport, Area Technical Lead, showed how to build no-code agents from the Gemini Enterprise app, and how to integrate them with other, “high-code” agents.

Shifting down

Last but not least, it was time to talk about security and governance. “Agents give users and other agents new ways to intentionally — or unintentionally — expose data and behavior in ways that we may not want,” mused Emma.

The standard response to that is to “shift left” — move testing, quality, and performance evaluation earlier in the development process — but for developers, that usually means more work, Richard said. “It’s not sustainable for developers to be responsible for all the layers of the stack,” he said. Instead, “we need to shift down.”

To help, there’s Agent Identity and Agent Gateway, demoed by Ankur Kotwal, head of Cloud Developer Relations. Ankur showed how Agent Gateway uses IAM policies to ensure agent actions are only accessible by approved sources, and how Agent Identity provides each agent with a unique and immutable credential. Then, Agent Policies can be configured to provide guardrails for the agents.

Yinon Costica, Co-Founder and VP of Product at Wiz, then went a step further and showed how Wiz can scan your agent code and infrastructure, and Wiz Green Agent can suggest root cause remediations.

“It’s a full architecture for security to easily understand what you built without you having to actually explain it,” Yinon said. Better yet, he also showed using this functionality from Anthropic’s Claude Code with Opus. “With Wiz, we want to enable your choice of tools and models to fix and prevent real risks,” he said.

The finish line

With this, the developer keynote came to an end. But for Google Cloud developers, it’s just the beginning, as the entire solution is available as source code in GitHub, and all the demos are available as Codelabs. Because when it comes to agentic development, these resources will really help you hit the ground running.

Day 1 at Google Cloud Next ‘26 recap

Wed, 22 Apr 2026 23:00:00 +0000

Last year at Google Cloud Next ‘25, we asked you to imagine a new future for AI. At Next ‘26, the question before you is how do you move AI into production across your entire enterprise?

According to Google Cloud CEO Thomas Kurian, the answer is straightforward: You need a unified stack, with “chips that are designed for models, models that are grounded in your data, agents and applications that are built with those models,” and the whole thing “secured by the infrastructure,” Thomas said in his keynote. (This is the same unified stack that Google uses for Search, YouTube, Chrome, and Android. As Alphabet CEO Sundar Pichai said in his opening remarks, “a big focus of ours is to always be customer zero for our own technologies.”)

As AI matures, we’ve laid out a blueprint on how to succeed. Read on for a whirlwind tour of what we announced from the keynote stage.

Gemini Enterprise: The end-to-end system for the agentic era

Throughout this unified stack is Gemini Enterprise — “the connective tissue between your data, your people, and your goals,” Thomas said, providing a combination of intelligence and automation across multiple layers. Here’s what’s new.

1. Gemini Enterprise Agent Platform

Gemini Enterprise Agent Platform is where you go to build, scale, govern, and optimize agents. As the evolution of Vertex AI, it’s built on top of our leading infrastructure, and deeply integrated with our data and security capabilities — the foundation of the Agentic Enterprise. Here’s a sampling of Agent Platform’s new features and capabilities:

Build: Choose the right environment for the job — from the low-code, visual interface of the new Agent Studio, to the code-first logic of the upgraded Agent Development Kit (ADK). We’ve simplified the entire lifecycle with AI-native coding capabilities to help you ship production-grade agents faster.
Scale: Clear the path to production with the re-engineered Agent Runtime. This supports long-running agents that maintain state for days at a time and are backed by Memory Bank for persistent, long-term context.
Govern: Establish centralized control with Agent Identity, Agent Registry, and Agent Gateway. These capabilities help ensure every agent — whether built on Agent Platform or sourced from our partner ecosystem — has a trackable identity and operates within enterprise-grade guardrails.
Optimize: Guarantee quality with Agent Simulation, Agent Evaluation, and Agent Observability. These tools provide full execution traces and a real-time lens into agent reasoning to help ensure your agents always hit their goals.

To dive deep into Agent Platform, read more in our announcement blog here.

2. Gemini Enterprise app

The Gemini Enterprise app is “the primary environment where your business actually operates,” Thomas explained. The app is where many workers, especially non-technical ones, can ask questions of enterprise agents, create generative media, engage with prebuilt agents, and even create their own with conversational interfaces — all with governance, compliance, and security built in. Here’s a sample of what’s new in this foundational interface:

Gemini Enterprise Projects give your agents permanent memory.
Deep Think solves your most complex business challenges without context pollution.
Microsoft 365 interoperability makes it easy to export docs you create with Canvas into Microsoft Office formats.

To illustrate the power of Gemini Enterprise, Shaun White, three-time Olympic gold medalist, entrepreneur, and snowboarding legend, joined us on stage.

“Back when I was training, our tools were camcorders and guesswork. You’d land a trick and watch it back. And you’d be thinking, ‘How can I make that trick better?’” he said.

Alongside Jason Davenport, Google Cloud Tech Lead, they showed a model that Google Cloud built in collaboration with Google DeepMind that tracked Shaun in space from a two-dimensional video, helping him understand what he was doing right and wrong.

“Learning the trick on the mountain is one thing, but actually understanding the physics of a trick is a whole other thing,” Shaun said.

Read more on the Gemini Enterprise app here.

3. AI Hypercomputer

The same technology foundation that athletes like Shaun White use to understand their performance is being used by enterprises to transform their businesses.

Amin Vadhat, SVP and chief technologist, AI and Infrastructure, took to the stage to announce enhancements to AI Hypercomputer, the integrated supercomputing underneath every AI workload on Google Cloud.

First, there’s the eighth-generation Tensor Processing Unit, or TPU — “a thing of beauty,” Amin said. And because “the demands of training and serving have completely diverged,” this TPU family actually consists of two chips:

TPU 8t, optimized for training, uses new Inter-Chip Interconnect (ICI) technology to scale up to 9,600 TPUs and 2 petabytes of shared, high-bandwidth memory in a single superpod. It achieves three times the processing power of Ironwood and delivers up to double the performance per watt.
TPU 8i, optimized for inference, uses the new Boardfly topology to directly connect 1,152 TPUs in a single pod. It features three times more on-chip SRAM compared to previous versions and a specialized Collectives Acceleration Engine offloads resource-heavy tasks. Taken together, TPU 8i delivers 80% better performance per dollar for inference than the prior generation, helping millions of concurrent agents to run cost-effectively.

AI Hypercomputer also supports Arm-based Google Cloud Axion processors, such as the N4A, now generally available, and will be one of the first platforms to offer NVIDIA’s Vera Rubin NVL72 platform when it is released.

Other parts of the network need to keep up with the demands of the agentic era. For example, the new Virgo Network doubles connectivity to scale training beyond AI Hypercomputer superpods, and Google Cloud Managed Lustre now supports an industry-leading 10 terabytes per second of throughput.

Learn more about all of our AI infrastructure innovations here.

4. Agentic Data Cloud

The AI era hinges on data. Lots of data. That data comes with a catch: It needs to be grounded in context. That’s because “reasoning without context is just a guess,” explained Karthik Narain, chief product and business officer.

To that end, we’re totally rethinking our data platform, and giving it a new name: the Agentic Data Cloud.

Here’s a sampling of what you’ll find in the Agentic Data Cloud:

Knowledge Catalog constructs a unified, dynamic context graph of your entire business enabling you to ground agents in all of your business data and semantics. With Smart Storage and the Object Context API, files in Google Cloud Storage are instantly tagged and enriched with metadata before an agent touches them. Knowledge Catalog is also integrated with Gemini’s Deep Research Agent.
Data Agent Kit delivers a Gemini-powered data science authoring experience across your IDEs, Notebooks, and agentic terminals.
Lightning Engine for Apache Spark is a real-time, serverless engine that is up to 4.5 times faster than open-source alternatives and offers up to double price-performance over the leading competitor for large datasets.
Finally, Cross-Cloud Lakehouse, based on Apache Iceberg, lets you query data in Amazon Web Services or Azure without having to copy it.

Learn all about all the innovations in the Agentic Data Cloud here.

5. Agentic Defense

AI makes — and demands — that everything go faster, and security operations are no exception. Increasingly, “human analysts can’t keep up with AI-driven attacks,” said Francis deSouza, COO, and president, Security Products.

“Security must become an autonomous force, responding faster than the threat itself,” he said.

To help, we introduced three new agents in Google Security Operations to help you defend at the speed of AI.

Threat Hunting agent: Helps teams proactively hunt for novel attack patterns and stealthy adversary behaviors that bypass traditional defenses. Now in preview.
Detection Engineering agent: Identifies coverage gaps and creates new detections for threat scenarios. Now in preview.
Third-Party Context agent: Enriches workflows with contextual data from third-party content. Coming soon to preview.

You can also build your own security agents with remote Google Cloud MCP server support for Google Security Operations, now generally available, and access it from the Google Security Operations chat interface, now in preview.

Then there’s Wiz, now a part of Google Cloud, whose AI-Application Protection Platform (AI-APP), Wiz Security Agents, and Wiz Workflow help you identify and respond to risks and threats at machine speed. New in the Wiz family today are:

Secure vibe-coded applications: A new integration runs Wiz security scanning directly inside the Lovable platform so vulnerabilities, secrets, and misconfigurations caught by Wiz surface in Lovable's built-in security view. Generally available in May.
Secure AI-generated code: Wiz can now remove risks from AI-generated code with inline AI security hooks integrated directly into IDEs and agent workflows, injecting security guardrails before code is committed.
Agent-based remediation: Wiz Skills can equip coding agents and AI-native IDEs with full code-to-cloud context and validated attack surface findings from the Wiz Security Graph.
AI-Bill of Materials (AI-BOM): Work towards eliminating shadow IT by automatically inventorying all AI frameworks, models, and IDE extensions across your environment.
Google Cloud Fraud Defense: The evolution of reCAPTCHA, this platform is designed to discern the legitimacy and authorization of bots, humans, and agents. Now generally available.

Read more about these security innovations here.

6. Workspace Intelligence

For a look at AI for end-users, we heard from Yulie Kwon Kim, VP, Product, Google Workspace, who shared new ways that AI is manifesting in our collaboration and productivity suite. Workspace Intelligence is a unifying semantic layer that breaks down information and context silos for you and your agents. It understands your work, your priorities and the people you work with to help you get more done.

“Think of it as a unified intelligence layer that lives inside every Workspace app. It connects the dots and lets AI do the heavy lifting," Yulie said.

Here’s what’s new:

Ask Gemini in Google Chat allows you to instantly synthesize information, surface insights, and query projects from across Workspace directly from your Google Chat window. It provides proactive daily briefings to help you prioritize, and also lets you take immediate action — such as scheduling a meeting on your calendar or creating a Google Doc to develop a pre-meeting brief — turning your conversations into momentum without having to switch tabs.

Reimagined content creation in Docs, Sheets, and Slides uses Workspace Intelligence to synthesize information from across Workspace and the web and creates professionally formatted drafts that match your voice, style, and brand.
AI Inbox and AI Overviews in Gmail creates a personal, proactive inbox assistant with Gemini.
Google Drive Projects instantly organizes your team's files and emails to manage workflows, generate content, and deliver specific answers based on rich project context. In addition to newly added AI Overviews and Ask Gemini, Projects is another way we’re transforming Drive from a storage tool into an active collaborator to provide insights about your data.
Workspace agent in Gemini Enterprise executes complex, multi-step tasks across Google Workspace apps without having to leave Gemini Enterprise.

For more, check out the full Workspace Intelligence blog.

7. Agentic Commerce

For enterprises, AI agents are reshaping how consumers engage with companies and their products.

“In the agentic era, an agent isn’t just a tool; it’s a strategic extension of your business, built to expand your reach, deepen engagement, and personalize service at scale,” said Carrie Tharp, Google Cloud Vice President of Go To Market Strategic Industries.

Gemini Enterprise for Customer Experience offers a suite of tools to enhance the entire customer journey, from the first moment of discovery through on-going interactions that remember the customer like the best shopkeeper would.

Shopping agent and Food Ordering agent bring new conversational sales and ordering capabilities direct to businesses and third-party chat and digital interfaces.
Omnichannel Gateway helps agents maintain context across web, mobile, and voice, so a company’s agents can offer more personalized service.
Agent Assist helps during complex customer service situations, coaching employees to deliver fast and more accurate answers to customer questions by having organizational data readily available through gen AI grounding.

With Omnichannel Gateway in particular, about bridging the physical, digital, and agentic shopping experience, so consumers always have a familiar, brand-aware experience.

“If a customer moves from text chat to a phone call, the agent seamlessly remembers exactly where they left off,” said Carrie. Now that’s progress!

For more customer stories, check out all 1,302 of the latest gen AI use cases from businesses around the globe.

Innovate all the things

From new products, to new solutions, to new ways of working, there are so many other ways that we’re helping organizations take their AI from pilot to production.

Over the following week, we’ll share even more news, helpful how-to guides, and go deeper on today’s announcements. Stay tuned!

Enabling the agentic enterprise: business and industry agents arrive in Gemini Enterprise

Wed, 22 Apr 2026 21:00:00 +0000

We are officially moving past the era of one-size-fits-all AI. Enterprises today require highly specialized, role-specific tools to drive real productivity — but they cannot afford to sacrifice security, oversight, or control to get there.

What's new: Today, we’re bringing partner-built agents from our Agent Marketplace directly into the Agent Gallery inside the Gemini Enterprise app. This creates a powerful, centrally governed hub for discovering and managing specialized AI from partners like Accenture, Adobe, Atlassian, Deloitte, Lovable, Oracle, Palo Alto Networks, Replit, S&P Global, Salesforce, ServiceNow, Workday, and more.

Why this matters for your business: With millions of paid seats across thousands of companies, the Gemini Enterprise app allows organizations to automate complex, end-to-end workflows in a single, unified environment. Five key benefits include:

Move beyond agentic overclaims: The market is flooded with passive chatbots and rigid workflows. Our featured Marketplace partners deliver true autonomous agents capable of executing complex, end-to-end tasks. They achieve this by combining deep context and memory, intelligent orchestration, and safe tool use.
High-caliber agents: Every featured agent earns the "Google Cloud Ready - Gemini Enterprise" designation only after passing a strict four-step evaluation for basic functionality, output accuracy, autonomous execution, and enterprise standards.
Built-in safeguards: Gemini Enterprise Agent Platform assigns every agent a cryptographically secure identity for a clear audit trail, while our Agent Gateway and Model Armor actively screen traffic to guarantee your data is never used for model training.
Strong IT governance: To balance workforce productivity with strict organizational oversight, we've built Agent Gallery around a secure, two-step governance model. Employees can browse the gallery and submit procurement or access requests for the agents that fit their needs, while IT remains firmly in the driver's seat. Administrators review all incoming requests to approve or deny deployments, retaining granular control over exactly who can interact with each agent across the organization.
A unified ecosystem breaking down silos: Typically, purchasing third-party AI means bolting on disconnected platforms. Our approach eliminates that friction. These partner agents are built to natively run within Gemini Enterprise Agent Platform. This means third-party agents and internal custom builds live side-by-side in a single, cohesive ecosystem, allowing your organization to scale its AI capabilities infinitely without ever fracturing your infrastructure.

Go-to-market opportunity for Partners: Today, we announced a $750 million partner fund for agentic development. In addition, partners can tap into a $460 billion plus backlog of committed enterprise spend and reach millions of Gemini Enterprise users directly in their daily workflows. We are turning everyday user discovery into a procurement flow for IT, standardizing contracts and accelerating purchasing cycles by up to 50%. With Google sales reps incentivized to work with you, and Marketplace vendors closing deals 112% larger, there has never been a better time to build with Google Cloud.

Where you can get started:

For customers: Discover these agents directly via the Agent Gallery in the Gemini Enterprise app. Start your 30-day trial today.

For partners: Take advantage of our rapid agent deployment framework, which provides a clear path to onboard, commercialize, and make agents discoverable to Gemini Enterprise users. Apply for our AI Agents Program now.

In addition, we’re launching the Google for Startups AI Agents Challenge today, including a track dedicated to startups with agents that can be primed for use in Gemini Enterprise.

Meet the agents your organization can start using today

A sampling of featured partner-built agents from Agent Marketplace accessible via the Agent Gallery in the Gemini Enterprise app

Acalvio: ShadowPlex Preemptive Cyberdefense Agent deploys honeytokens in Cloud IAM and in cloud workloads to disrupt agentic AI attacks at the reconnaissance phase. This helps secure cloud workloads from malicious or unsafe agent actions in real time.

Accenture: The Supply Chain Inventory Intelligent Advisor agent is an agentic AI-powered decision support system designed to optimize inventory planning and tracking efficiently. Using real-time Data and AI, it delivers actionable insights to balance service levels and minimize excess inventory.

Adobe: Adobe Marketing Agent for Gemini Enterprise connects natural language queries to Adobe's Customer Experience agentic capabilities. It enables instant insights on campaign performance, target audiences and journey monitoring powered by Adobe, helping teams work smarter without leaving their workflow.

Alteryx: Alteryx AI Insights Agent delivers analyst-curated insights inside Gemini Enterprise. Built on Alteryx One, analysts define trusted datasets while business users simply ask questions to receive governed, repeatable answers. Requires an Alteryx One subscription.

Ambiguous AI: Ambiguous Recruiting Coworker manages candidate communication, multi-round scheduling, and pipeline tracking from application to offer. It works natively across Gmail, Google Calendar, Google Sheets, Google Drive, and more, while looping you in for key decisions.

Amdocs: The Telco Customer Experience Agent autonomously resolves complex customer service issues across digital and contact center channels. Combining Amdocs Cognitive Core with Gemini Enterprise for CX, it reasons, orchestrates, and executes end‑to‑end telco processes, improving resolution speed, cost efficiency, and customer satisfaction.

AODocs: AODocs AI Agent provides instant, trustworthy answers grounded exclusively in the latest validated versions of controlled documents. It supports diverse use cases including standard operating procedures (SOPs), engineering specs, legal policies, and risk management procedures.

Articul8 AI: Articul8 Table Understanding Agent converts complex documents into analytics-ready table data. It reconstructs hierarchical layouts and relationships missed by traditional OCR for accurate analysis. Articul8 Diagram Understanding Agent turns complex diagrams into structured data, interpreting symbols and spatial relationships for automated analysis and seamless workflow integration.

Ascendo: Ascendo AI Agent helps critical infrastructure teams automate workflows, improve diagnostics, and reduce field escalations. By ingesting logs, telemetry, and service manuals, it recommends optimal actions to support faster, accurate decision-making in complex service environments.

Atlassian: Rovo is an AI teammate seamlessly integrated across Jira, Confluence, and your tool stack. Built on Atlassian’s Teamwork Graph, it surfaces knowledge, automates tasks, and takes intelligent actions to help teams pull context, move faster and stay aligned.

AutoCIO: Portfolio managers and investment teams use the AutoCIO Financial Forecasting Agent to design, optimize, and manage portfolios. It automates end-to-end workflows, delivers predictive signals across 50,000+ securities, and enables real-time, hyper-customized strategies via natural language prompts.

Avalara: Avi Agent is an AI-powered gateway for tax and compliance automation. Built on Avalara’s Agentic platform, it smartly coordinates between external systems and compliance agents to instantly and securely calculate taxes, file returns, and manage invoicing.

Backstory: Backstory Revenue Answers Agent tells revenue leaders what's happening on any account, identifies deal risks, and suggests actions. It achieves this by synthesizing CRM data, engagement activity, stakeholder relationships, and public company news into actionable insights.

Botcopy: TrueQ is an AI analyst for government contact centers, enabling supervisors to understand and improve conversational AI performance in the Gemini Enterprise app. It identifies conversation patterns, escalation triggers, and knowledge gaps to improve automation, strengthen governance, and reduce support costs.

Carto: Carto Site Selection AI Agent helps enterprises analyze and visualize prospective commercial locations using spatial analytics. Users ask conversational questions about foot traffic, demographics, or competitors, and results automatically fuel a comprehensive location analysis.

Corvic AI: Corvic AI Agent connects to your enterprise data and delivers production-ready decision intelligence — no pipelines and no data engineering required. Use cases include turning competitive signals and market trend data into actionable intelligence, root cause analysis and customer support automation.

Cotality: The Cotality Payoff Analysis Agent helps mortgage lenders improve customer retention and reduce loan runoff. By analyzing loan originations and payoffs, it delivers actionable intelligence to understand recapture performance, borrower behavior, and competitive insights.

Crescendo Lab Ltd.: DAAC AI Agent transforms complex data into actionable growth strategies for marketing and sales. Using natural language, it delivers instant conversational insights and automated dashboards, effectively eliminating manual reporting and guesswork.

Deloitte: Tariff Management Agentic Suite automates customs regulatory workflows via sub-agents that ingest documentation, extract attributes, and reconcile tariffs against filings. It flags discrepancies to ensure full audit coverage, eliminate manual effort, and recover unnoticed costs.

Devoteam: Demand Sensing Agent helps leaders improve forecast accuracy and optimize inventory. It analyzes sales and real-time market signals to provide immediate demand insights, enabling proactive adjustments. Request for Proposal Agent helps sales teams draft proposals, leveraging centralized knowledge to rapidly generate tailored and compliant proposal drafts.

Dun & Bradstreet: The Dun & Bradstreet Business Verification Agent enables trusted business verification in the Gemini Enterprise app. Powered by Dun & Bradstreet’s trusted data, it resolves businesses to a D-U-N-S® Number and validates identity attributes, enabling front-line teams to move faster and focus on exceptions.

Dynatrace: Dynatrace for Gemini Enterprise integrates AI agents with Dynatrace observability data via Gemini Enterprise Agent Platform and A2A protocol. This enables real-time insights into performance, infrastructure, and user experience, seamlessly enhancing AI workflows while ensuring enterprise-grade security and compliance.

Enigma: Enigma KYB Agent gives compliance teams instant, AI-driven business verification backed by ground-truth business identity from 100M+ registrations and authoritative government records, plus sanctions watchlists and real-time transaction signals.

Eon: Eon AI Agent empowers cloud infrastructure teams to streamline data protection operations. It connects Google Cloud accounts, triggers resource discovery, monitors protected resources, tracks backup jobs, and inspects restore status — all through natural conversation, replacing manual console workflows.

EPAM: Creative Conductor Marketing Agent turns creative briefs into polished promotional videos. It coordinates specialized sub-agents for visuals, voiceover, and music, assembling assets into a final MP4 to deliver fast, scalable, high-quality marketing content.

Exa AI: Exa Agent gives workers a live web view, turning it into a searchable database of code, people, and companies. It searches billions of pages to accelerate tasks like developer tooling, recruiting, and market mapping.

Foxtrot Communications: Forge - AI Agent helps teams eliminate manual ETL pipeline development. Connect any API, and Forge autonomously delivers clean, normalized, analytics-ready datasets, reducing time-to-insight to minutes and freeing engineers to focus on strategic initiatives.

Genpact (UK) Ltd: Genpact Finance One – Revenue Lens and PnL Agents are a suite of agents built on Google ADK and the A2A protocol, enabling finance users to gain actionable insights from revenue and profit & loss data through natural language conversations in Gemini Enterprise.

Genspark: Genspark AI Agent turns multi-step work into finished outputs with one prompt by combining search, web interaction, content generation, and media creation. It helps users execute research, create documents, and automate workflows through a single agent built for real work.

HCLTech: HCLTech ITOps Agent is built on Google Cloud and the HCLTech AI Force platform and enables autonomous IT operations within ServiceNow. Using natural language, it triages incidents, surfaces relevant SOPs, and drives guided remediation to improve overall efficiency, consistency, and operational performance.

Invideo: Invideo empowers marketing teams to scale video production while maintaining consistency. By leveraging brand guidelines and assets, it delivers production-ready ads and short films with built-in quality controls and output in 50+ languages.

Iron Mountain, Inc.: The Iron Mountain InSight® DXP AI Agent bridges Gemini Enterprise and your managed content for unified search and AI-powered Q&A. Its intended purpose is to accelerate results and reduce hours of manual document review to seconds of conversational inquiry.

Lilt: LILT Assist empowers global content managers to transition from manual tasks to autonomous orchestration, managing end-to-end multilingual production and brand governance. Assist scales global content volumes with AI while ensuring enterprise-grade consistency and compliance.

Lovable: Lovable empowers anyone to build real apps and websites by chatting with AI. Teams across product, design, marketing, sales and operations can turn ideas into real, working software.

LTM: LTM Video Intelligence Agent is based on Google Cloud AI technology and converts recorded videos into BDD test cases in a standard format for faster time-to-test and improved productivity.

LumApps: The LumApps AI Agent helps organizations retrieve company knowledge and announcements, trigger workflows, and complete tasks across enterprise systems.

Manhattan: Manhattan Active Agents are AI digital teammates for warehouse management, transportation management, order management, and platform operations. These agents help monitor work, resolve exceptions, guide users, automate tasks, and recommend actions to reduce manual effort; improving speed and accuracy, increasing utilization, and delivering better operational resilience.

Menlo Security: The Menlo Security Orchestrator for Gemini Enterprise reduces mean time to repair (MTTR) from hours to milliseconds. Command global policy and perform deep threat reporting through natural conversation in the Gemini Enterprise app.

monday.com: monday.com agent in Gemini Enterprise turns planning into execution. Teams can quickly build boards, summarize updates, and trigger automations without leaving their workspace—keeping everyone aligned, tracking progress, and driving measurable results.

Nativeorange: LexAI Agent streamlines property and casualty (P&C) underwriting and operations. It ingests submissions, extracts data, and automates risk evaluation, delivering actionable insights on classification and offering next steps to improve speed, accuracy, and decision consistency.

Neo4j: Neo4j Agent translates natural language into Cypher, enabling applications to read from and write to the Neo4j database. Integrated with Gemini Enterprise, it can be used to build agents grounded with GraphRAG and connected enterprise knowledge.

Obin AI: Obin Financial Agent helps private markets and commercial lending teams run financial analysis from Gemini Enterprise. Users submit model spreadsheets and deal-related documents with a scenario description in plain-language. The agent then extracts relevant data and computes financial metrics across modeling assumptions.

Onix: The Onix Risk & Compliance Intelligence Agent automates compliance reviews and risk assessments across regulated industries. It reduces review times from weeks to minutes, proactively identifies regulatory risks, and creates detailed audit trails with citations.

OpenText: OpenText™ Content Aviator™ is a generative AI assistant that enhances content search and productivity. Built on Gemini Enterprise Agent Platform, it allows users to perform natural language searches and efficiently summarize enterprise documents.

Oracle: Oracle AI Database Agent lets you query Oracle data directly from Gemini Enterprise. Available via Google Cloud Marketplace, it delivers fast, enterprise-grade answers without requiring custom chatbots or NL-to-SQL maintenance.

Orion by Gravity: Orion by Gravity is an AI analyst that connects to Looker, BigQuery, and more to proactively hunt for insights. It empowers data teams to identify anomalies and recommend actions that uncover 'unknown-unknowns' in analytics-rich industries.

Palo Alto Networks: Prisma AIRS Model Security agent scans over 35 model formats to eliminate malicious code hidden deep within model layers. This proactive defense goes beyond standard scanning to ensure neural integrity, detecting poisoned weights and backdoors to prevent unauthorized data exfiltration and model hijacking.

Pendo: The Pendo Agent connects your product intelligence to Gemini Enterprise. Query usage trends across SaaS and AI tools, surface adoption gaps, and understand the impact of your AI investments without leaving the Gemini Enterprise app.

Persistent Systems: Ethical Content Auditor Agent helps editors at publishing houses audit political and social content for potential biases and risks. Editorial Assistant Agent enables efficient manuscript evaluation by reading and understanding long manuscripts, analyzing market trends and providing data-backed summaries and recommendations.

Pluto7: Pluto7’s Planning in a Box - Pi Agent orchestrates 50+ sub-agents for supply chain and manufacturing teams. This enables self-healing, autonomous supply chains that optimize demand and production in real-time.

Product Genius: Product Genius brings AI-powered ecommerce intelligence directly into Gemini Enterprise. Share your website URL to get instant insights, actionable policies, and a conversational analyst that helps you optimize your store—no code install required.

Quantum Metric: Felix Agentic for Gemini Enterprise delivers instant insight, contextualized in business objectives, to digital leaders and analytics teams. Leveraging Gemini models, Felix continuously analyzes your digital experience to explain what changed, why it matters, and where it’s costing you.

Red Hat, LLC: Red Hat Lightspeed Agent for Google Cloud helps SREs and admins manage Red Hat Enterprise Linux infrastructure on Google Cloud via natural language. Built on A2A and Gemini, it securely connects with Red Hat Lightspeed services to streamline enterprise IT operations on Google Cloud.

Replit: Replit Agent enables business users to build enterprise-grade applications using natural language. With no coding knowledge required, users turn complex requirements—like a BigQuery-backed customer health dashboard—into functional custom tools in minutes.

S&P Global: The S&P Global Data Retrieval Agent by Kensho enables financial professionals to access cited data from S&P Global's proprietary datasets. It allows analysts and portfolio managers to analyze earnings calls, conduct market research, and instantly retrieve specific financial metrics.

Salesforce: Agentforce Sales for Gemini Enterprise can engage leads, create meeting briefs, surface deal risks and guidance, and manage pipeline and CRM updates — securely and in real-time — without leaving Gemini Enterprise. With the agent handling the grind of manual tasks, sellers can focus on relationship building and strategic selling to win.

Sana from Workday: Sana Self-Service Agent instantly finds and summarizes information from Workday and other knowledge sources, providing personalized answers and assistance with tasks. With over 300 skills across areas like pay, time, and absence, the agent handles everyday HR and finance tasks globally, reducing support tickets and allowing teams to focus on high-value work.

Saviynt: Identity Security for AI enables continuous discovery and inventory of agents, identifying their functions, access, and risks. It treats agents as first-class identities with defined purposes and accountable owners. This blend of posture visibility and enforced accountability converts opaque automation into governable, enterprise-ready identities.

ServiceNow: Now Assist for IT Operations Management operates inside the ServiceNow AI Platform, harnessing the power of GenAI to elevate operational intelligence and automation across IT environments. It introduces a suite of AI-driven capabilities designed to optimize every stage of alert and incident management.

Sheetgo: Sheetgo is an AI-ready data execution platform that transforms spreadsheet-driven operations into governed, reliable pipelines for BigQuery and Gemini Enterprise.

Skyflow: The Skyflow Runtime Data Security Agent helps teams securely deploy agents on Gemini Enterprise Agent Platform. It provides expert guidance and on-demand data protection with fine-grained access controls, enabling enterprises to meet compliance without risking privacy.

Supermetrics: The Supermetrics Marketing Intelligence Agent delivers live, cross-channel performance insights directly in Gemini Enterprise. Connecting to 175+ sources like Google Ads and GA4, it provides fast, hallucination-free analysis using simple natural language prompts.

Synthpop Inc.: The Synthpop Patient Journey Orchestration Agent turns unstructured referrals into validated, payer-ready orders in real time by executing intake and patient-facing workflows. These flows help payers save time as part of Synthpop’s coordinated, production-grade administrative backbone.

Tech Mahindra: Frontline agents use Order Assist to resolve complex orders and service cases faster. By delivering real-time, contextual intelligence through conversation, it reduces handling time, improves first-contact resolution, and enables consistent, data-driven responses.

Teradata: Data Analyst AI Agent enables natural-language analytics on enterprise data in Google Cloud. It translates business questions into governed SQL and Python workflows on Teradata, delivering faster, secure insights without requiring data movement.

Typeface: Typeface Agent acts as a seamless bridge to creative campaigns. It enables instant access to Typeface teams, brand assets, and campaigns via natural language from Gemini Enterprise. Users can trigger high-quality, brand-aligned content creation workflows, streamlining the journey from ideation to production.

UKG: Agentic People Assist — HR Service Delivery elevates the agentic capabilities of the UKG Workforce Operating Platform by empowering employees and managers with knowledge and connected agents to turn intent into action. It orchestrates workflows across time, pay, and HR to deliver real-time, personalized, and measurable outcomes.

WRITER: WRITER Agent is autonomous AI for Fortune 500 enterprises. Built for high-stakes enterprise workflows, it safely plans and executes autonomous work across data and tools based on company context, from brand standards to repeatable playbooks.

XM Cyber: XM Cyber’s PostureAI agent serves as an autonomous posture specialist, providing continuous assessment of Google Workspace applications against security best practices. It is part of a broader Continuous Exposure Management offering that allows organizations to safely adopt AI and secure their critical business processes.

Together, we are creating the world's largest ecosystem of agents — one that will redefine every industry, supercharge every function, and accelerate every business process. I cannot wait to see what we all achieve together.

What’s new in Cloud Run at Next ‘26

Wed, 22 Apr 2026 12:00:00 +0000

From vibe-coded and large-scale apps to AI models and agents, Cloud Run delivers on-demand compute with zero overhead and pay-per-use pricing for all of your workloads. Last year, the number of external active developers and applications on Cloud Run doubled, with more new customers and apps coming to Cloud Run in 2025 than in its first 6 years combined!

Today, we’re announcing new features and improvements to Cloud Run to help you run your workloads:

Build and deploy full-stack apps in Google AI Studio with Cloud Run, Firestore, and user authentication.
Build, scale, govern, and optimize reliable AI agents with the all new Gemini Enterprise Agent Platform and Cloud Run.
Enable easy deployments from developers and agents with Cloud Run's fully managed remote MCP server.
Combine high-performance inference and serverless compute with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs on Cloud Run.

Empowering the new era of developers

For decades, software development has had an inaccessible learning curve, but thanks to AI, anyone can be a digital builder. With Cloud Run, you can go from prototype to deployed app in seconds.

Build full-stack apps in Google AI Studio
AI Studio now supports full-stack applications that can run server-side code, a Firestore database, and user authentication. Deploy your vibe-coded apps with a single click to Cloud Run, now generally available.

Cloud Run's fully managed remote MCP server
To make it even easier for developers or agents to deploy code, we are launching an official remote Cloud Run MCP (Model Context Protocol) server, giving you the tools to manage and deploy apps. Now GA.

Billing caps
Soon, you’ll be able to define your maximum spend per month. If your bill reaches this amount, your Cloud Run resources will be de-activated.

"Cloud Run has been one of the best technical choices we made in our deployments platform. It is our primary target, powering us to over 1 million live projects being hosted on Replit." - Scott Kennedy, VP of Engineering, Replit

Embracing the agentic era

AI agents are just like people in that they need access to a compute environment to perform their tasks. For a cloud-based AI agent to take complex actions, it can use Cloud Run’s on-demand compute service.

Cloud Run integration with Gemini Enterprise Agent Platform
Through its integration with Cloud Run, Agent Platform helps agents transition from experimental environments into fully managed, production-grade systems without having to rebuild them. Now in preview with select customers.

Cloud Run instances
Traditionally, Cloud Run services, jobs, or worker pools have been opinionated ways to manage Cloud Run infrastructure. Now, we are giving you access to the underlying primitive: you can create individual Cloud Run instances. Coupled with Cloud Storage volume mounts, these instances are ideal for hosting long-running background agents like OpenClaw in one simple command:

code_block: <ListValue: [StructValue([('code', 'gcloud run instances create \\\r\n --image alpine/openclaw:latest \\\r\n --port 18789 \\\r\n --memory 4Gi \\\r\n --default-url \\\r\n --add-volume mount-path=/home/node/.openclaw,type=cloud-storage,bucket=$BUCKET_NAME'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7fabd8442dd0>)])]>

This functionality is available in preview to select customers.

Cloud Run sandboxes
Agents often need a safe place to execute code or other commands as quickly as possible. Coming soon, while processing a single request, you will be able to very quickly spin up an ephemeral sandbox that’s strictly isolated from your agent code using a built-in sandbox tool:

code_block: <ListValue: [StructValue([('code', 'app.post(\'/execute\', (req, res) => {\r\n const escapedCode = req.body.code.replace(/"/g, \'\\\\"\');\r\n\r\n exec(`sandbox do -- /usr/bin/python3 -c "${escapedCode}"`, (e, stdout, stderr) => {\r\n res.send({ stdout, stderr });\r\n });\r\n\r\n});'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7fabd968c4d0>)])]>

"Cloud Run’s concurrency model has been instrumental in simplifying our AI workloads for our customer service AI tool, Lumi. Using Cloud Run alongside Gemini and AlloyDB, we’ve created a unified action layer that enables real-time call summarization and flow guidance, leading to improved first-call resolution rates and faster onboarding for our contact center team." - Edward Wright, Head of Engineering, VirginMedia O2 UK

Automatic scaling for high-demand applications

Cloud Run automatically scales to meet your demand, making it a great fit for large customers that need to serve and respond instantly to heavy traffic spikes.

SSH support for Cloud Run
Developers can now gain secure shell access (SSH) directly into a running Cloud Run container, enabling advanced troubleshooting and inspecting the container's file system on the fly, now in preview with select customers. Open a secure interactive shell session with simple command:

code_block: <ListValue: [StructValue([('code', 'gcloud run services ssh SERVICE'), ('language', ''), ('caption', <wagtail.rich_text.RichText object at 0x7fabd968c7d0>)])]>

Cloud Run service bindings
Coming soon, unlock seamless service-to-service communication for your scalable microservices architectures with Cloud Run service bindings.

“Cloud Run’s serverless architecture empowers us to meet exponentially growing demand through near-instant scaling, while its streamlined developer experience simplifies building and running our applications.” - Mimi Chen, Member of Technical Staff, Anthropic

Running AI models

From serving custom models with GPUs to training and fine-tuning models with jobs, you can use Cloud Run for your most AI-intensive workloads.

Support for NVIDIA RTX PRO 6000 Blackwell GPU on Cloud Run
We’re bringing the serverless experience to high-end inference with support for NVIDIA RTX PRO™ 6000 Blackwell GPUs on Cloud Run, now GA. This means you can serve up to 70B+ parameter models without having to manage any underlying infrastructure, including scaling to zero when the resource is not in use.

Ephemeral disk
With per-instance temporary disk storage, workloads can process large files or use scratch space without eating up your container memory. Now in preview, ephemeral disk storage is created when an instance starts and deleted when it stops.

"Cloud Run has fundamentally changed how we manage our model deployments. By moving to a usage-based, scale-to-zero model, we’ve eliminated idle GPU costs for low-traffic models. We are now running over 17 model variants in production across multiple regions, each independently deployable and isolated, without the burden of capacity planning or fleet management." - Ajay Nair, Global VP, Elastic

On-demand compute for every workload

Whether you’re a seasoned software developer or a vibe coder looking to deploy the next viral app, Cloud Run delivers on-demand compute for everyone and every workload. Get started today.

Introducing Gemini Enterprise Agent Platform, powering the next wave of agents

Wed, 22 Apr 2026 12:00:00 +0000

In the early days of generative AI, building safe and reliable business tools took massive engineering effort and a high tolerance for trial and error. We helped solve that with Vertex AI, our trusted AI development platform. But today, we’re managing a different level of complexity, with agents interacting across multiple systems — and often without security and governance guardrails.

To move toward a truly autonomous enterprise, one where agents can act with the same independence and reliability as a member of your team, you need a foundation that can sustain that level of trust.

What’s new: Today, we’re launching Gemini Enterprise Agent Platform — our new, comprehensive platform to build, scale, govern, and optimize agents. It’s the evolution of Vertex AI, bringing the model selection, model building, and agent building capabilities that customers love, together with new features for agent integration, DevOps, orchestration, and security.

Agent Platform provides a single destination for your technical teams to build agents that can transform your products, services, and operations. These agents can be seamlessly delivered to your employees through the Gemini Enterprise app, all while remaining tightly integrated with your IT operations to help ensure control, governance, and security as you scale.

The platform also provides first-class access to more than 200 of the world’s leading models through Model Garden. This includes our latest first-party breakthroughs like Gemini 3.1 Pro, Gemini 3.1 Flash Image, and Lyria 3, alongside our open models like Gemma 4. And, of course, customers have full flexibility to use the best model for the job with support for third-party models like Anthropic’s Claude Opus, Sonnet and Haiku.

Moving forward, all Vertex AI services and roadmap evolutions will be delivered exclusively through the Agent Platform, rather than as a standalone service, to power the next generation of agent development.

Why Agent Platform matters for your business: Agent Platform helps you move from managing individual AI tasks to delegating business outcomes with total confidence. You can:

Build: Choose the right environment for the job — from the low-code, visual interface of the new Agent Studio, to the code-first logic of the upgraded Agent Development Kit (ADK). We’ve simplified the entire lifecycle with AI-native coding capabilities to help you ship production-grade agents faster.
Scale: Clear the path to production with the re-engineered Agent Runtime. This supports long-running agents that maintain state for days at a time and are backed by Memory Bank for persistent, long-term context.
Govern: Establish centralized control with Agent Identity, Agent Registry, and Agent Gateway. These capabilities help ensure every agent — whether built on Agent Platform or sourced from our partner ecosystem — has a trackable identity and operates within enterprise-grade guardrails.
Optimize: Guarantee quality with Agent Simulation, Agent Evaluation, and Agent Observability. These tools provide full execution traces and a real-time lens into agent reasoning to help ensure your agents always hit their goals.

Get started with Agent Platform: Visit Agent Platform in the Google Cloud console to explore new features and start building today.

Keep reading for a deeper look at our latest releases and how Agent Platform helps you deliver the production-ready agents you can trust at every stage of the journey.

How customers are achieving more with Gemini Enterprise Agent Platform

"Burns & McDonnell uses Agent Platform to transform how organizational knowledge is applied across the enterprise. Using ADK, we are building an AI agent that turns decades of project data into real-time, actionable intelligence. Agent Platform enables this innovation to scale responsibly by combining deterministic business rules with probabilistic reasoning — making AI a trusted operational capability, not just a productivity tool. With Agent Platform, we aren’t just managing knowledge; we are activating experience to drive faster, more confident decisions." – Matt Olson, Chief Innovation Officer, Burns & McDonnell

“Color Health uses Agent Platform to power our Virtual Cancer Clinic, delivering end-to-end care. By building our Color Assistant with the Agent Development Kit (ADK) and scaling it via Agent Runtime, we are helping more women get screened for breast cancer. The Color Assistant engages users to check screening eligibility, connects them to clinicians, and helps schedule appointments. The power of the agent lies in the scale it enables — helping us reach more people and respond to individual risk and eligibility in real time.” – Jayodita Sanghvi, PhD., Head of AI Platform, Color

“By rebuilding Comcast’s Xfinity Assistant with Agent Development Kit (ADK), we’ve moved beyond simple scripted automation to conversational generative intelligence that delivers personalized troubleshooting and self-service support to our customers. Agent Runtime has been a massive accelerator, allowing us to deploy a sophisticated multi-agent architecture that increases digital containment while ensuring secure, grounded interactions via Gemini. We aren't just reducing repeat interactions by solving customers’ issues the first time; we're redefining the customer experience at scale.” – Rick Rioboli, Chief Technical Officer, Connectivity & Platforms, Comcast

“Geotab uses Agent Platform to rapidly accelerate our AI Agent Center of Excellence. Google's Agent Development Kit (ADK) provides the flexibility to orchestrate various frameworks under a single, governable path to production, while offering an exceptional developer experience that dramatically speeds up our build-test-deploy cycle. For Geotab, ADK is the foundation that allows us to rapidly and safely scale our agentic AI solutions across the enterprise” – Mike Branch, Vice President, Data & Analytics, GeoTab

"Gurunavi uses Agent Platform to power 'UMAME!', an AI restaurant discovery app that leverages Memory Bank to achieve a deep understanding of user context. Unlike conventional prompt-based systems, our agent remembers a user's past actions and preferences to proactively present the best options. This eliminates the need for manual searches and creates a seamless experience that will improve user satisfaction by 30% or more. We view this memory function as a non-negotiable feature for the future of new culinary experiences.” – Toshiaki Iwamoto, CTO, Gurunavi

"At L'Oréal, Beauty Tech is not just a support function — it is a powerful catalyst to create the beauty that moves the world. To live up to that ambition, we decided to build our own proprietary Beauty Tech Agentic Platform, powered by Google Cloud. Leveraging Agent Development Kit (ADK), we are leading a fundamental shift: moving from deterministic workflow automation to autonomous, outcome-oriented agent orchestration. Our agents are not locked in a vacuum — through Model Context Protocol (MCP), they are securely connected to our single sources of truth, including our Beauty Tech Data Platform and core operational applications. Google Cloud gives us the resilience, the multi-LLM flexibility, and the enterprise-grade trust framework we need to scale this platform globally, while keeping human oversight at the center." – Etienne BERTIN, Group CIO, L'Oréal

“Payhawk uses Agent Platform to transform our AI agents from simple task executors into genuine financial assistants. By leveraging Memory Bank, we have moved from stateless interactions to long-term context retention. Our agents now act like dedicated team members, autonomously recalling user-specific constraints and history. For example, our Financial Controller Agent now remembers a user’s habits to auto-submit expenses, reducing submission time by over 50%. This shift allows our agents to anticipate needs based on past behavior rather than just reacting to prompts.” – Diyan Bogdanov, Principal Applied AI Engineer, Payhawk

"PayPal uses Agent Platform to rapidly build and deploy agents in production. Specifically, we use Agent Development Kit (ADK) and visual tools to inspect agent interactions, and manage multi-agent workflows. This provides the step-by-step visibility we need to visualize the flow of intent and payment mandates. Finally, Agent Payment Protocol (AP2) on Agent Platform provides the critical foundation for trusted agent payments. helping our ecosystem accelerate the shipping of secure agent-based commerce experiences." – Nitin Sharma, Principal Engineer, AI, PayPal

Build AI agents

Build agents quickly and easily by empowering your developers, business users and everyone in between to build and deploy agents at scale.

Build smarter agents, faster

A major upgrade to ADK: More than six trillion tokens are processed monthly on Gemini models through ADK. Unlock more powerful reasoning by organizing agents into a network of sub-agents. This new, graph-based framework allows you to define clear, reliable logic for how agents work together to solve complex problems.
Workspaces are secure-by-design: Give agents a hardened, sandboxed environment to run bash commands and manage files safely, isolated from your core systems.
Multimodal streaming: Bring human-like stability to real-time interactions with multimodal support for live audio and video cues.

Connect your agents to the enterprise

Securely access any system: Use plug-and-play architecture with Native Ecosystem Integrations to connect agents to your internal data and tools without custom coding.
Automate background operations: Activate your data in BigQuery and Pub/Sub with Batch & Event-driven agents. This way, you can run massive, asynchronous tasks like content evaluation or data analysis in the background.

Go from idea to production in hours

Enable AI-driven development: A programmatic interface for coding agents to access Google’s complete suite of agentic capabilities, allowing them to build, evaluate, and deploy production-ready agents on your behalf.
Bringing agent building directly to Agent Studio: Now, you can move seamlessly from building simple prompts to deploying complex agents in Agent Studio. Once you're ready for deep customization, export your logic directly into ADK to continue development in a full-code environment.
Get a head start with pre-built agents: Access a curated set of agent templates in Agent Garden — including code modernization, financial analysis, economic research, invoice processing, and more — that serve as immediate building blocks for your multi-agent systems.

Scale AI agents

To move from a proof-of-concept to a live environment, you need a platform that can handle the performance, state, and security requirements of real-world work.

Powering high-performance agent execution

The latest Agent Runtime: Our revamped Agent Runtime delivers sub-second cold starts and allows you to provision new agents in seconds.
Support for multi-day workflows: You can now deploy long-running agents that run autonomously for days at a time. This allows your agents to manage complex, multi-step workflows and deep reasoning tasks that require extended persistence, like managing a sales prospecting sequence.
Autonomous action with security-by-design environments: Agent Sandbox provides a hardened environment to safely execute model-generated code and perform computer use tasks like browser-based automation without risk to your host systems.
Agent-to-agent orchestration: Enables agents to seamlessly delegate tasks to one another, including support for complex, generative, and deterministic orchestration patterns. This ensures that for critical flows such as compliance, your agents follow well-specified paths every time.

Move beyond temporary session data to high-accuracy context

Personalize interactions: Agent Memory Bank dynamically generates and curates long-term memories from conversations. Using new Memory Profiles, agents can recall high-accuracy details with low latency, ensuring context is never lost.
Link AI interactions to your existing records: Store and manage history using Agent Sessions. With Custom Session IDs, you can use your own unique identifiers to track sessions and map them directly to your internal database and CRM records.
Enable real-time, human-like interactions: Using the WebSocket protocol for Bidirectional Streaming, you can help ensure your agents are highly responsive during live customer or employee interactions, processing audio and video without lag.

Govern AI agents

Govern with a secure-by-design architecture that applies enterprise rigor to every agent in your fleet – from the ones you build on Agent Platform to the ones you source from our partner ecosystem.

Manage all of your agents through a single source of truth for identity and access.

Assign every agent a verifiable identity: Agent Identity improves the security posture of your agents by ensuring every agent receives a unique cryptographic ID. This creates a clear, auditable trail for every action an agent takes, mapped back to defined authorization policies.
Maintain a central library of approved tools: Our new Agent Registry provides a single source of truth for your enterprise. It indexes every internal agent, tool, and skill, simplifying discovery and ensuring only governed, approved assets are available to your users.
Manage your agent fleet from one control point: Agent Gateway acts as the air traffic control for your agent ecosystem. It provides secure, unified connectivity between agents and tools across any environment, while enforcing consistent security policies and Model Armor protections to safeguard against prompt injection and data leakage.

Use AI-powered insights to detect hidden risks and suspicious behavior before they impact your business.

Detect suspicious behavior in real-time: Agent Anomaly Detection uses statistical models and an LLM-as-a-judge framework to flag unusual reasoning. This works alongside Agent Threat Detection to provide visibility into malicious activity, such as reverse shells or connections to known bad IP addresses.
Uncover vulnerabilities automatically: A new Agent Security dashboard, powered by Security Command Center, unifies threat detection and risk analysis. It allows your teams to map relationships between agents and models, automate asset discovery, and scan for vulnerabilities in the underlying operating system and language packages.

Optimize AI agents

Agent Platform gives you the visibility needed to understand how your AI is performing, making it easy to refine their logic and get smarter over time.

Test your agents before they ship

Simulate realistic conversations: Use Agent Simulation to test agents against human-like synthetic user interactions and virtualized tools in a controlled environment. Agents are automatically scored based on task success and safety across multi-step conversations.

Monitor and improve in production

Track live performance: Use Agent Evaluation to continuously score agents against live traffic using multi-turn autoraters that can evaluate the logic of an entire conversation, not just a single response. With turnkey dashboards and Agent Observability, you can visually trace complex reasoning to debug issues as they happen.
Automate agent refinement: Instead of manually digging through logs, Agent Optimizer automatically clusters real-world failures and suggests refined system instructions to improve accuracy.

Detailed technical guides and a full list of updates are available in our updated documentation and release notes. Agent Platform is the new standard for enterprise agent development, built to help you move from experimentation to production-scale impact, starting today.

Next ‘26: Redefining security for the AI era with Google Cloud and Wiz

Wed, 22 Apr 2026 12:00:00 +0000

aside_block: <ListValue: [StructValue([('title', 'Our news today from Next ‘26'), ('body', <wagtail.rich_text.RichText object at 0x7fabd964d1d0>), ('btn_text', ''), ('href', ''), ('image', None)])]>

The AI era demands a new security era. Organizations are facing the dual challenge of harnessing the potential of AI while defending against its malicious use, and Google Cloud can help you adapt and thrive.

The latest research from Google Cloud shows that adversaries are using AI to accelerate the speed, scale, and sophistication of attacks. Meanwhile, M-Trends 2026 also showed that increased threat actor coordination has driven down the time to hand-off from an initial access to a secondary threat actor from eight hours to 22 seconds in the last three years.

Today at Google Cloud Next, we are showcasing how Google Cloud can help you defend against increasingly sophisticated threats at machine speed, protect AI and multicloud environments, and secure cloud workloads at scale.

Delivering agentic defense

Our full-stack AI approach, from the chips to the models, gives you a competitive advantage with better integration and velocity to help protect customers. Not only can Google action insights from the world’s largest threat observatory and Mandiant frontline experts, but we also bring cutting-edge insights and breakthroughs from Google DeepMind, to help make your platforms more secure.

Today we are introducing three new agents in Google Security Operations to help you defend at the speed of AI.

Threat Hunting agent, now in preview, can help teams proactively hunt for novel attack patterns and stealthy adversary behaviors that bypass traditional defenses.
Detection Engineering agent, now in preview, can identify coverage gaps and create new detections for threat scenarios, reducing toil and transforming detection creation from a manual craft into an automated science.
Third-Party Context agent, coming soon to preview, can enrich your workflows with contextual data from third-party content.

Initiating a threat hunt with the Threat Hunting agent

Our Triage and Investigation agent processed over 5 million alerts in the last year, reducing a typical 30-minute manual analysis to 60 seconds with Gemini.

“Operational resilience and cybersecurity are the bedrock of customer trust at BBVA. By integrating advanced artificial intelligence, such as the Triage and Investigation agent, we are able to scale in new ways," said Diego Martinez Blanco, head of Security Technology, BBVA.

“It handles the initial heavy lifting and filters out false positives so we can prioritize issues that require human attention. The agent's transparent explanations allow our team to understand recommendations and ultimately dedicate our resources to more complex investigations,” he said.

You can build your own security agents with remote Google Cloud model context protocol (MCP) server support for Google Security Operations, now generally available. To make it even easier, you can also access the MCP server client directly from the Google Security Operations chat interface, available in preview.

Organizations leveraging an intelligence-led, AI-augmented approach to modern security operations with Google Cloud's agentic defense can realize a strong ROI. Christopher Kissel
Research Vice President, IDC

Findings report created by the Threat Hunting agent

Security teams can also automate response actions with agentic automation in Google Security Operations. To further move teams from manual triage to agentic defense, we introduced dark web intelligence in Google Threat Intelligence, now in preview. Internal tests show it can analyze millions of daily external events with 98% accuracy to elevate threats that truly matter.

"IDC found that organizations experienced measurable operational gains, including substantial reductions in mean time to detect and mean time to respond, fewer false positives, and higher analyst productivity with AI-powered context and automation. These operational improvements translate into significant business outcomes, such as shorter disruption periods, lower incident-related costs, and improved executive confidence in security posture and decision-making," said Christopher Kissel, research vice president, IDC. "Organizations leveraging an intelligence-led, AI-augmented approach to modern security operations with Google Cloud's agentic defense can realize a strong ROI."

New partner-supported workflows for Google Security Operations

Today, we are also announcing a robust cohort of new partner integrations for Google Security Operations. Designed to deliver high-fidelity security workflows right out of the box, our latest participating Google Cloud Security integration ecosystem partners include Darktrace, Gigamon, and SAP.

Protecting AI and cloud applications across any infrastructure

AI and cloud applications are built across multiple platforms and models. To protect them end-to-end, we want to make it easier and faster to mitigate risk, regardless of where and how you build. This support includes major cloud environments like Amazon Web Services, Google Cloud, Microsoft Azure, and Oracle Cloud; software-as-a-service (SaaS) environments like OpenAI; and even custom hosted environments.

Wiz, now a part of Google Cloud, expands and deepens our ability to protect the apps you build and run. Wiz empowers you to quickly and securely adopt AI, while also helping protect the AI development lifecycle.

Wiz announced its AI-Application Protection Platform (AI-APP) at the RSA Conference, providing deep visibility, risk posture, and runtime analysis for your AI applications. Wiz also announced Wiz Security Agents and Wiz Workflows, helping you identify and respond to risks and threats at machine speed.

Today, we’re taking our commitment to secure customers in any cloud, platform, and AI environment further. Wiz now supports Databricks as well as new agent studios like AWS Agentcore, Gemini Enterprise Agent Platform, Microsoft Azure Copilot Studio, and Salesforce Agentforce, so customers gain visibility however their teams choose to build.

In addition, Wiz continues to support security ecosystems with integrations to the outer layer of the cloud, including Google Cloud Apigee, Cloudflare AI Security for Apps, and the Vercel platform, further extending the power of the Wiz Security Graph. We’ve also updated how we integrate security detections from Wiz Defend with Google Security Operations and Mandiant Threat Defense to help analysts more easily configure automatic threat information forwarding.

Wiz is also announcing new capabilities designed to secure the AI-native development lifecycle, helping teams to innovate faster and more securely:

Secure vibe-coded applications: Wiz is announcing a new integration, generally available in May, that runs Wiz security scanning directly inside the Lovable platform so vulnerabilities, secrets, and misconfigurations caught by Wiz surface in Lovable's built-in security view, right where teams are already building.
Secure AI-generated code: Wiz removes risks from AI-generated code the moment it is created. Inline AI security hooks integrate directly into IDEs and agent workflows to evaluate prompts and scan AI-generated output instantly, injecting security guardrails before the code is ever committed.
Agent-based remediation: Wiz Skills equip coding agents and AI-native IDEs with full code-to-cloud context and validated attack surface findings from the Wiz Security Graph. These capabilities enable teams to trigger automated, agent-driven remediation workflows either locally from the developer's individual IDE or globally at the repository and pull request level within your version control system.
Eliminate shadow AI: Wiz’s dynamic AI-Bill of Materials (AI-BOM) automatically inventories all AI frameworks, models, and IDE extensions across your environment. This provides complete visibility into what is writing code across your stack, allowing you to track sanctioned corporate tools like Gemini Code Assist and GitHub Copilot while simultaneously uncovering unapproved shadow AI plugins.

You can learn more about the Wiz announcements here.

Securing your agents and the agentic web

In addition to securing your cloud and AI workloads, Google Cloud’s secure-by-design foundation can help you innovate at the speed of AI — from agents to fraud defense to the web.

Securing and governing agents with the Gemini Enterprise Agent Platform
To build, orchestrate, govern, and optimize agents, today we are announcing Gemini Enterprise Agent Platform including:

Agent Identity to enable access management and AI governance at scale. Our new capability provides agents unique identities to operate autonomously with specific authentication flows, and with scoped human delegation.
Agent Gateway, which enables policy enforcement for all agent-to-agent and agent-to-tool connections. It governs your enterprise agent traffic and understands agent protocols like MCP and Agent2Agent (A2A) to inspect and secure every agent interaction.
Model Armor, our runtime protection for model and agent interactions, now integrates with Agent Gateway, Agent Runtime, and Langchain available in preview, and Firebase, generally available, to help developers add inline enforcement and sanitization of agent traffic and interactions without the need to change code. These integrations expand Model Armor's protection against runtime risks such as prompt injections, tool poisoning, and sensitive data leakage across Google Cloud services and our AI portfolio.

Securing the agentic web with Google Cloud Fraud Defense and Chrome Enterprise
Today, we are evolving reCAPTCHA with the launch of Google Cloud Fraud Defense, generally available. This comprehensive platform is designed to discern the legitimacy and authorization of bots, humans, and agents. Using the same scale and signals that protect Google’s own ecosystem, Fraud Defense will soon offer in preview agent-specific capabilities for human users and AI agents that can help secure the digital commerce journey, from account creation and login to payment and checkout.

Our commitment to securing AI extends to the browser, a vital endpoint for interacting with AI. Chrome Enterprise provides comprehensive data protection for the AI era with the visibility and controls needed to embrace AI safely without compromising corporate data:

AI-aware extension threat detections, now in preview, can surface advanced extension telemetry that helps security teams detect and respond to anomalous AI agent activity.
New shadow AI reporting, generally available soon, can help you gain visibility into the shadow AI landscape by flagging employee use of unsanctioned web-based AI and SaaS applications.

What’s new in Trusted Cloud

We continue to offer new security controls and enhance capabilities across identity, data, and networking on our cloud platform to help you secure your environments. Today we’re announcing the following updates:

Simplifying permissions with modern IAM
To help achieve least privilege quickly and simply, we’ve streamlined our predefined roles catalog with easy-to-use administrator, editor, and viewer roles, such as the IAM role picker and the ability to re-authenticate sensitive actions.

Data security
We are announcing several new capabilities for our cloud platform data security portfolio to help protect your most sensitive data and accelerate AI transformation.

Confidential Computing: In partnership with NVIDIA, today we’re announcing Confidential Computing support for G4 VMs, featuring NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs on Google Compute Engine (GCE) Confidential G4 VMs, available in preview globally, to help strengthen confidentiality and integrity for a wide spectrum of sensitive AI workloads. In partnership with Intel, we’re also introducing the preview of C4 Confidential VMs, bringing Intel TDX to 6th Gen Xeon processors to help protect diverse AI and analytics workloads while providing industry-leading compute density and performance.
Cloud Key Management Services (KMS): We are announcing the new Confidential External Key Manager (cEKM) in preview, giving you the flexibility to host and protect external keys in any region and maintain verifiable control within a confidential environment.
Post-quantum cryptography (PQC): We are introducing KMS Quantum Safe Key Imports, available in preview, to help you bring your own keys with quantum-safe algorithms.
Secret Manager: To help prevent password leaks and mitigate prompt injection risks, we are announcing the general availability of the native integration of our Secret Manager with Agent Development Kit.

Network security
Google Cloud’s Cross-Cloud Network security products offer several new capabilities:

Cloud NGFW: We’re announcing the Cloud NGFW advanced malware sandbox, in preview later this year, to help defend against highly evasive zero-day threats. This capability is powered by Palo Alto Networks Advanced Wildfire, trained on data from more than 70,000 Palo Alto Networks customers to stop 99% of known and unknown malware.
Cloud Armor: We have released new Cloud Armor managed rules, powered by Thales Imperva and available in preview, to detect Layer 7 application attacks and zero-day CVEs (like React2Shell).

Advancing Google Cloud security with SCC
As our Google Cloud-native security solution, Security Command Center (SCC) establishes a cloud security baseline to protect both your traditional and AI applications on Google Cloud:

AI agents, models, and MCP servers are secured by providing continuous discovery and comprehensive risk analysis to identify threats, vulnerabilities, and misconfigurations.
SCC will add deep runtime visibility to uncover shadow AI for your Google Cloud workloads. Coming soon in preview, SCC will automatically discover unmanaged agentic workloads — including agents, MCP servers hosted on Cloud Run, GKE, and inference endpoints running on GKE, and surface those as posture findings in SCC.
Our enhanced Security Command Center Standard tier provides data security posture management, compliance, vulnerability management, and risk analysis to help any Google Cloud customer establish strong security, compliance and risk coverage from the start at no additional costs.

Take the next step

When you make Google part of your security team, you gain the power of an intelligence-driven, AI-native defense; the freedom of an open cloud that’s secure-by-design; and the industry's most-battle tested experts as an extension of your organization.

For more on these new innovations and how you can secure what’s next, tune in to watch our security spotlight. And be sure to check out the many great security breakout sessions — live and on-demand — to learn more about all of our Next ‘26 announcements.

The new Gemini Enterprise: one platform for agent development, orchestration, and governance

Wed, 22 Apr 2026 12:00:00 +0000

The first wave of AI changed how we find information; the next wave is changing how we get work done. Today, we’re enhancing our most powerful AI tools and bringing them together under one roof.

Gemini Enterprise is now an end-to-end system for the agentic era, built for agents that can execute complex, multi-step work processes. It combines access to frontier AI models, an intuitive UI, a secure development framework, and the ability to successfully deploy agents at scale.

At Google Cloud Next, we’re launching an expanded Gemini Enterprise portfolio that includes several key components:

Gemini Enterprise Agent Platform is our new developer platform and evolution of Vertex AI. It brings together our full suite of models, development, and tuning services, with new features for businesses to build, scale, govern, and optimize agents that can work autonomously to execute complex business workflows. Learn more here.
Gemini Enterprise app lets teams discover, create, share, and run AI agents in a single, secure environment. The Gemini Enterprise app is built on the Agent Platform, meaning all governance, security, and identity capabilities come standard, and it integrates with your enterprise data – including data in third-party systems–through connectors.
An open partner ecosystem to discover and deploy a wide range of third-party agents from leaders like Oracle, Salesforce, and ServiceNow – all within a secure and governed environment.

As organizations start building and managing thousands of agents, they need new ways to provide oversight and minimize sprawl. Gemini Enterprise enables IT departments to manage agent permissions and activities securely. It provides the same level of oversight and auditability found in essential business applications like payroll or quarterly financial reporting.

Gemini Enterprise’s unified approach provides multiple advantages across an organization:

For developers: Agent Platform is the foundation for the agentic enterprise built on our leading infrastructure and deeply integrated with our data and security capabilities. It’s where technical teams can build innovative agents that transform your products and services, alongside agents that scale your internal operations. With tools like low-code Agent Studio, support for Agent Development Kit (ADK), and protocols like A2A and MCP, Agent Platform is the key to building, scaling and optimizing agents across the entire organization.
For IT teams: We provide a single control plane for governance in Agent Platform, so every employee can use and share agents with full IT visibility. Both no-code and pro-code agents are managed through a consistent model for identity, security, and auditing – preventing AI sprawl without slowing down innovation.
For knowledge workers: Gemini Enterprise app is the front door for AI for every employee in your organization. Knowledge workers can build their own agents with our no-code Agent Designer or leverage a wide range of partner-built agents from Google Cloud Marketplace. And these agents can be easily shared across the organization with full governance and compliance controls built-in. Inbox in Gemini Enterprise provides a central location to manage and monitor agents, including long-running agents.

Gemini Enterprise Agent Platform: securely orchestrating the agentic enterprise

At the heart of the Gemini Enterprise experience is our new Agent Platform. It’s the central place where developers can build, scale, govern, and optimize fleets of agents. Today, we're introducing new capabilities across four key areas:

Build: An enhanced Agent Development Kit (ADK) helps you unlock more powerful reasoning by organizing agents in a network of sub-agents. With ADK’s new graph-based framework, you can orchestrate complex processes with new ecosystem integrations and support for MCP.
Scale: With our improved Agent Runtime, your agents can be both fast and persistent, and with agent-to-agent orchestration, they can seamlessly delegate tasks to one another. The launch of Memory Bank and Memory Profiles also ensure agents operate with long-term contextual memory, allowing them to recall user preferences and past history across different sessions.
Govern: Our secure-by-design architecture gives IT teams centralized control over their entire agent lifecycle. Agent Identity assigns every agent a unique cryptographic ID for complete traceability and auditing, while Agent Gateway provides air traffic control to ensure interactions between agents and data is secure. Model Armor then protects against activities like prompt injection, tool poisoning, and sensitive data leakage.
Optimize: To move beyond trial-and-error, the new Agent Simulation tool allows you to stress-test your agents against real-world scenarios before they ship. Agent Observability also helps ensure agents function exactly as intended, providing a real-time view into safety and performance.

Any agent built on Agent Platform can then be instantly surfaced in the Gemini Enterprise app for your employees to access. By building on a foundation where security and identity are included by default, you can stop worrying about managing risk and start focusing on managing agent-led business outcomes.

Learn more: What's new in Agent Platform

The Gemini Enterprise app: your hub for everyday work

In the Gemini Enterprise app, teams can discover, create, share, collaborate, and run agents in a single environment. With data governance and compliance built in, it gives your teams a safe place to leverage agents grounded with your business data. It also provides a simple interface for users to engage, and democratizes the power of agent creation, orchestration, and scaling to every individual in the organization.

New capabilities available in the Gemini Enterprise app include the ability to:

Innovate with no-code agents: Using the newly-integrated Agent Designer, anyone can create agents to automate complex, multi-system processes with simple, natural language. This includes schedule or trigger-based agents, including those triggered by another business application, even via a partner connector.
Centralize agent oversight: Inbox in Gemini Enterprise provides a unified hub for managing agents at scale, including long-running agents executing complex workflows. Users can monitor active agents, adjust configurations, and receive real-time status alerts directly through email and chat, ensuring automated processes stay on track.
Establish persistent team memory: Moving beyond individual chats, Projects creates a shared workspace for teams and agents. It centralizes the context and history of a given topic, ensuring that critical project intelligence remains a permanent company asset even as team members or roles change.
Collaborate in a dynamic workspace: Canvas provides teams with an interactive editor to co-create and edit in Google Docs and Slides. We’ve also added Microsoft 365 interoperability, allowing you to export the Docs and Slides you create within Canvas into common Microsoft Office formats.

Learn more: What's new in Gemini Enterprise

The agent economy: amplified by our partner ecosystem

No organization can build every agent they need from scratch. The agentic enterprise thrives on an open, interoperable ecosystem. That's why we are helping to make it easier than ever to discover and deploy specialized agents from our partners.

Benefits include:

Activating third-party agents from within the Gemini Enterprise app: We are now bringing the full catalog of partner-built agents from Google Cloud Marketplace directly into the Agent Gallery. You no longer need to leave your workflow to find and deploy powerful third-party agents. The app provides a secure request-and-approval gateway that connects employee needs to IT-validated solutions.
A growing library of trusted solutions: Agent Gallery features an expansive collection of validated agents from SaaS leaders and innovative startups. For example, you can use specialized agents from partners like Adobe, Salesforce, ServiceNow, and Workday–directly within the Gemini Enterprise app.
Deploy with confidence: Every partner agent in the gallery has been validated by Google Cloud to meet our requirements for security and interoperability. This allows you to benefit from a world of innovation, while maintaining the centralized governance and control your business demands.

Learn more: Partner-built agents available in Gemini Enterprise

Gemini Enterprise brings the best of Google AI to every employee, every developer, and every customer — empowering them with cutting-edge AI agents. Get started with Gemini Enterprise today.

What’s new with compute: Scaling core and agentic workloads

Wed, 22 Apr 2026 12:00:00 +0000

At Google Cloud Next, we’re announcing a range of compute capabilities to enable your core general purpose and AI workloads for the agentic world with higher performance and lower costs.

Why it matters: IT leaders and builders are faced with balancing compute investments and resources between agentic AI and the general purpose use cases, including the web servers, databases, and enterprise applications that drive everyday customer experiences.

On one side, agents can place unpredictable demand on compute infrastructure, often scaling exponentially. A single user interaction can instantaneously kick off hundreds of concurrent, high-throughput, and low-latency tasks. On the other side, general-purpose workloads generate and hold the data required to fuel the agentic world. Relying on static and siloed infrastructure to run them can risk performance bottlenecks and spiraling costs, leaving your organization unable to respond to surges in demand.

Consider a global travel application where a simple vacation search instantly triggers a massive orchestration of agentic inventory checks, dynamic pricing models, and AI-driven personalized itineraries. Without a modern architecture, this sudden surge in demand can overwhelm the core booking database and bring business to a halt.

We address this with fluid compute, Google Cloud infrastructure that adapts to your general-purpose and agentic workflows, enabling both to win by flexing in performance, capacity, and scale, all in real time. This dynamic flexibility relies directly on the automated orchestration of Google Kubernetes Engine (GKE) and our new Agent Sandboxes to instantly provision secure, isolated execution environments at machine speed.

Let’s take a deeper look at the new compute capabilities announced at Next ‘26.

Run AI and general purpose workloads together

Agentic planning and reinforcement learning depend on highly fluid compute to process unpredictable bursts of autonomous tasks. Relying on static infrastructure to isolate agent-generated code can create severe provisioning delays and heavily inflate your cloud budget. You can remove these bottlenecks by adopting an adaptive cloud foundation. Leveraging GKE Agent Sandboxes empowers your teams to securely launch thousands of execution environments. Pairing these scalable sandboxes with efficient Google Axion processors helps your organization optimize total cost of ownership while fueling artificial intelligence innovation.

Here’s what’s new in Google Cloud compute launches and announcements:

Google Axion N4A is GA: Harness the agility of Google’s custom Arm-based Axion CPUs and achieve up to 2x better price-performance than comparable current-generation x86-based VMs for cost-sensitive workloads such as Java applications, scale-out web servers, and SaaS built by startups, enterprises and partners. Learn more here.
GKE Agent Sandbox, with Axion N4A for price performance, is GA. As the industry’s only native sandbox service among hyperscalers, GKE Agent Sandbox offers scalable and low-latency infrastructure designed for agents to safely execute untrusted code and tool calls without sacrificing performance. With Google Axion, you can build agents on leading infrastructure without compromising on cost or choice. GKE Agent Sandbox running on Google Axion N4A instances provides up to 30% better price-performance than the next leading hyperscale cloud provider. Try GKE Agent Sandbox here.
Google Axion C4A.metal, our first Axion bare metal instance, is in preview: C4A.metal instances power Android development, automotive simulation, CI/CD pipelines, security workloads, and custom hypervisors, without the performance overhead and complexity of nested virtualization. C4A.metal will be GA this summer; learn more here.
C4 instances offer expanded support for Intel Xeon 6 (Granite Rapids) across all shapes: Achieve high-performance for AI workloads like LLM inference and vector search by using Intel AMX with native FP16 support to increase throughput and reduce latency, offering 13% better price-performance versus comparable Intel Xeon 6-based VMs from another leading hyperscaler. C4 VMs are available with Intel Xeon 6 processors across all shapes. Learn more here.
Flexible CUDs expanded support is GA: Shift spend across regions and VM families while optimizing for TCO, with flexible committed use discounts, now with support for a wider range of VM families and services, including memory-optimized (M1-M4) and HPC-optimized (H3, H4D) VM families, as well as Cloud Run. Learn more here.

Here’s what customers are saying:

Unity: Unity is redefining the economics of real-time AI with Unity Vector. By migrating its on-demand feature processor workloads to Google Axion N4A instances, Unity achieved a 20% improvement in cost efficiency without sacrificing latency. As Unity Vector scales to meet increasing demand, the move to N4A instances ensures that Unity continues to deliver industry-leading performance at a sustainable cost.

Deutsche Börse: A leading German market infrastructure provider, Deutsche Börse migrated and modernized dozens of core financial applications onto Google Compute Engine VMs, including latest generation C4 and C4D instances, supporting latency-sensitive Oracle databases and post-trade processing at scale, and boosting release speed, operational agility, and resilience. This delivered the consistent performance they needed to process millions of financial transactions every day and they achieved 58% faster time to market and 33% lower TCO.

WP Engine: WP Engine powers millions of digital experiences where every millisecond matters. By running GKE clusters on C4D and N4D instances, WP Engine has seen up to a 60% reduction in latency for mobile-optimized REST APIs and up to 51% faster processing for data-rich application requests.

eDreams ODIGEO: Operating a high-volume, AI-powered travel platform where every millisecond dictates the customer experience, eDreams ODIGEO migrated its foundational Java-based ecommerce modules on GKE to Axion virtual machines. This immediately eliminated weeks of manual code optimization, delivered a massive 75% improvement in P95 latency with zero code changes, and unlocked price-performance to scale their global services far more cost-effectively than their legacy x86 infrastructure could.

Chainguard: Prioritizing absolute isolation for their foundational software build system, Chainguard deployed the new Axion C4A bare metal instances. This allowed them to establish a strong hypervisor security boundary for package builds, secure their development pipeline with architectural parity, and ensure robust protection, all without compromising build performance.

Run I/O and latency-sensitive workloads together

Both AI and core workloads depend on the ability to store, read, and move data as a single, high-performance operation. Traditionally, these stages are slowed by network and storage limits tethered to vCPU counts, which can starve AI models of the data they need to function. You can remove these constraints by leveraging accelerated Hyperdisk performance for rapid data access and high-performance networking for consistent transit. By allowing your data pipelines to scale independently of compute, your AI training and I/O-sensitive workloads have the dedicated bandwidth they need to remain stable under peak demand.

C4N is in preview: Running high-volume network applications such as concurrent mobile app requests and real-time inventory updates can risk bottlenecks during peak traffic. Maximize your throughput with C4N, featuring Titanium adapters that offload complex packet processing to deliver a market-leading 95 million packets per second — a 40% performance advantage for high-traffic network applications compared to other leading hyperscalers. Designed to rapidly transfer large datasets, C4N provides nearly 400 Gbps of VM-to-VM bandwidth, a 4x improvement in bandwidth-per-vCPU, and achieves an 8x increase in egress network bandwidth through internet gateways compared to C4 VMs. C4N with Hyperdisk Extreme also provides the low-latency, high-speed data access that modern databases and enterprise AI applications need, with 25 GiB/s of block storage throughput and nearly 1M IOPS. Sign-up here for C4N preview access.
M4N is in preview: Running memory-intensive databases can push organizations to overprovision compute cores (vCPU) to meet memory speeds, driving up software licensing fees. We introduced the new M4N series to solve this exact problem. Running Oracle workloads on M4N with Hyperdisk Extreme can reduce TCO by over 20%, enabling you to run Oracle more efficiently, with 26.57 GiB of RAM per vCPU for scale and on far fewer cores. Paired together, M4N with Hyperdisk Extreme delivers the highest per-core IOPS and throughput for high-memory instances across leading hyperscalers. Sign-up for the preview here.
Announcing Z4D: Optimize I/O-intensive workloads and remove network-based storage bottlenecks with new Z4D instances. By securing up to 84 TiB of high-performance local SSD directly on the node, organizations can process massive datasets for SQL, NoSQL, and vector databases. Z4D provides up to 400 Gbps of VM-to-VM bandwidth, matching both C4N and M4N. Z4D virtual machines and bare metal instances will be in preview soon.

Here is what customers are saying:

Ericsson: 5G Core workloads are inherently network-heavy, demanding high-throughput packet processing and deterministic latency that standard public cloud instances often struggle to maintain at scale. By leveraging the Google Cloud C4N, they’ve found the ideal choice for network performance to power Ericsson On-Demand. C4N’s architectural focus on network-optimized compute allows its 5G Core-as-a-Service to reach unprecedented throughput levels, like its recent 1 Tbps milestone, while maintaining the carrier-grade reliability its customers expect.

Teradata: Teradata’s Autonomous Knowledge Cloud enables the world’s largest enterprises to activate enterprise intelligence and turn trusted data into measurable business outcomes. Customers rely on Teradata to run mission‑critical, highly I/O‑intensive analytics at scale where performance and efficiency directly determine value. C4N instances are well suited for these demanding workloads, delivering strong price‑performance and supporting more efficient, optimized deployments. With C4N, Teradata can help customers accelerate insights, scale with confidence, and drive greater impact from their data and AI investments.

Handle demanding storage requirements

Foundational workloads such as web servers, applications, and databases hold the data required to fuel the agentic world. Siloing this critical information on rigid hardware creates bottlenecks that can completely stall enterprise modernization. Imagine a global retail brand running a holiday promotion, but the inventory database times out and drops customer requests because the legacy hardware couldn’t process the sudden flood of agentic queries.

Organizations require the highest performing database hosts backed by high performance IOPS and throughput per vCPU to ensure non-blocking data delivery. Moving these applications to modern cloud infrastructure dramatically improves total cost of ownership and operational throughput. Through strategic cloud migrations, customers can eliminate the architectural walls that stall modernization and unlock their data for AI. Here is what is new in fluid compute for throughput- and capacity-sensitive workloads:

Hyperdisk Balanced improvements. Hyperdisk Balanced enables fast and efficient block storage for general purpose workloads, including applications and relational databases. With Hyperdisk Balanced you can drive up to 2.4 GiB/s and 160K IOPS per volume, higher than general-purpose block storage offerings from other hyperscalers, all while achieving lower mean latency than alternatives. With Hyperdisk Balanced High Availability you can now achieve a 4x performance improvement for high availability databases like SQL Server or PostgreSQL by dynamically routing full disk performance to the active VM, removing the need to overprovision storage. Leverage zero-downtime encryption key rotation and consistency groups for instant snapshots, making it easier to stay more secure. With these capabilities, you can deliver lower TCO, higher performance, and workload resilience for your general-purpose workloads. Learn more here.
Hyperdisk ML performance improvements and Hyperdisk Exapools are GA: With 2 TiB/s of aggregate throughput (up from 1.2 TiB/s), Hyperdisk ML helps eliminate AI storage bottlenecks, offering more than 200x higher throughput per disk than competitive offerings, so your valuable accelerator clusters never sit idle. This allows you to maximize AI compute ROI while powering the next generation of intelligent agents. In addition, for large-scale training needs, Hyperdisk Exapools offer the highest aggregate block storage performance and capacity, per AI cluster, of any hyperscaler. Learn more about Hyperdisk ML and Exapools here and here.
Announcing Z4M: Access up to 168 TiB of local SSD coupled with up to 400 Gbps of network bandwidth, support for RDMA, and bare-metal shapes to run distributed parallel file systems and large-scale AI/ML workloads. Z4M will be integrated with Cluster Director with the option to be colocated with accelerators to provide fast and low-latency access to data. Z4M VMs and bare metal instances are expected to be in preview in Q3 2026.

Here is what customers are saying:

Shopify: During Black Friday weekend sales, Shopify processed over $14.6B and tracked 136 million packages for 81M buyers using its Shop App built on Compute Engine’s Z-series backed storage — without compromising speed or reliability.

HubX: Operating a massive portfolio of AI-powered mobile applications where rapid model loading dictates the user experience, HubX deployed Hyperdisk ML on GKE to eliminate severe I/O bottlenecks. Leveraging this specialized storage layer allowed HubX to support hundreds of concurrent readers and accelerate pod initialization times by 30x during peak traffic surges, drastically reducing idle accelerator costs and helping ensure their complex inference workloads scaled as expected.

Fluid infrastructure for the agentic era

Now, your foundational workloads and agents no longer need to compete for capacity or performance. With Google Cloud’s fluid compute, you get adaptive cloud infrastructure that prevents bottlenecks and enables both your foundational and AI workloads to collaborate and thrive.

Ready to get started? Head straight to the Google Cloud console to spin up a VM for your next big project. Or start planning your migration by checking out Migration Center's AI-powered toolsets to perform cost estimates, create a business case, and evaluate your modernization options.

Inside the eighth-generation TPU: An architecture deep dive

Wed, 22 Apr 2026 12:00:00 +0000

At Google, our TPU design philosophy has always been centered on three pillars: scalability, reliability, and efficiency. As AI models evolve from dense large language models (LLMs) to massive Mixture-of-Experts (MoEs) and reasoning-heavy architectures, the hardware must do more than just add floating point operations per second (FLOPS); it must evolve to meet the specific operational intensities of the latest workloads.

The rise of agentic AI requires infrastructure that can handle long context windows and complex sequential logic. At the same time, world models have emerged as a necessary evolution from current next-sequence-of-data architectures, which means newer agents are simulating future scenarios, anticipating consequences, and learning through "imagination" rather than risky trial-and-error. The eighth-generation TPUs (TPU 8t and TPU 8i) are our answer to these challenges, ensuring that every workload, from the first token of training to the final step of a multi-turn reasoning chain, is running on the most efficient path possible. They are built to efficiently train and serve world models like Google DeepMind’s Genie 3, enabling millions of agents to practice and refine their reasoning in diverse simulated environments.

TPU 8: Specialized by design

Recognizing that the infrastructure requirements for pre-training, post-training, and real-time serving have diverged, our eighth-generation TPUs introduce two distinct systems: TPU 8t and TPU 8i. These new systems are key components of Google Cloud's AI Hypercomputer, an integrated supercomputing architecture that combines hardware, software, and networking to power the full AI lifecycle. While both systems share the core DNA of Google’s AI stack and support the full AI lifecycle, each is built to address distinct bottlenecks and optimize efficiency for critical stages of development. Additionally, by integrating Arm-based Axion CPU headers across our eighth-generation TPU system, we’ve removed the host bottleneck caused by data preparation latency. Axion provides the compute headroom to handle complex data preprocessing and orchestration, so that TPUs stay fed and don’t stall.

TPU 8t: The pre-training powerhouse

Optimized for massive-scale pre-training and embedding-heavy workloads, TPU 8t utilizes our proven 3D torus network topology at an even larger scale of 9,600 chips in a single superpod. TPU 8t is designed for maximum throughput across hundreds of superpods, ensuring that training runs stay on schedule.

Here are some key advancements of TPU 8t over prior-generation TPUs:

The SparseCore advantage: Central to TPU 8t is the SparseCore, a specialized accelerator designed to handle the irregular memory access patterns of embedding lookups. While the Matrix Multiply Unit (MXU) handles matrix math, the SparseCore offloads data-dependent all-gather operations, amongst other collectives, preventing the zero-op bottlenecks that often plague general-purpose chips.
VPU/MXU overlap and balanced scaling: TPU 8t is designed to maximize provisioned FLOPs utilization. By implementing more balanced Vector Processing Unit (VPU) scaling, the architecture minimizes exposed vector operation time. This allows for better overlapping of quantization, softmax, and layernorms with the matrix multiplications in the MXU, helping the chip stay busy rather than waiting on sequential vector tasks.
Native FP4: TPU 8t introduces native 4-bit floating point (FP4) to overcome memory bandwidth bottlenecks, doubling MXU throughput while maintaining accuracy for large models even at lower-precision quantization. By reducing the bits per parameter, the platform minimizes energy-intensive data movement and allows larger model layers to fit within local hardware buffers for peak compute utilization.

Figure 1: TPU 8t ASIC block diagram

Virgo Network topology and up to 4x data center network increase: To support the massive data requirements of TPU 8t, we introduced Virgo Network. This new networking architecture enables up to 4x increased data center network (DCN) bandwidth on TPU 8t training over DCN. Virgo Network is a scale-out fabric designed for the extreme requirements of modern AI workloads. Built on high-radix switches that reduce network layers by allowing more ports per switch, it employs a flat, two-layer non-blocking topology. Compared with traditional datacenter networks, this significantly reduces latency by minimizing network tiers. It features a multi-planar design with independent control domains to connect TPU 8t chips. The TPU 8t racks also connect with the Jupiter north-south fabric to access compute and storage services. Together, this streamlined architecture delivers the massive bisection bandwidth and deterministic low latency necessary for enabling the world's largest training clusters with high availability.

With 2x scale-up bandwidth on the inter-chip interconnect (ICI) and up to 4x raw scale-out DCN bandwidth compared to the previous generation, TPU 8t drastically reduces data bottlenecks. Then, to further accelerate the development of frontier models, we scale distributed training beyond a single cluster. With JAX and Pathways, we can now scale to more than 1 million TPU chips in a single training cluster. Virgo Network can link over 134,000 TPU 8t chips with up to 47 petabits/sec of non-blocking bi-sectional bandwidth in a single fabric. This fabric delivers over 1.7K ExaFlops with near-linear scaling performance.

Figure 2: TPU 8t rack level connectivity to Virgo fabric

Faster storage access: We are introducing TPUDirect RDMA and TPU Direct Storage in TPU 8t. TPU Direct RDMA enables direct data transfers between the TPU's memory (HBM) and the Network Interface Cards (NICs), bypassing the host CPU and DRAM. This reduces latency and host system bottlenecks, increasing the effective bandwidth for TPU-to-TPU communication. Similarly, TPUDirect Storage bypasses CPU host bottlenecks by enabling direct memory access between the TPU and high-speed managed storage like 10T Lustre, effectively doubling the bandwidth for massive data transfers. This architecture allows the silicon to ingest training data at line rate, ensuring that the MXUs stay fully saturated even when processing large multimodal datasets.

By combining Managed Lustre 10T and TPUDirect Storage to route hundred-petabyte datasets directly to the silicon, TPU 8t prevents training delays caused by data ingestion bottlenecks. This delivers 10x faster storage access compared to training on seventh-generation Ironwood TPUs.

Figure 3: The top diagram shows the data transfer path without TPUDirect Storage. The bottom diagram shows TPU 8t data transfer with TPUDirect Storage between 2 TPU 8t chips and TPUDirect Storage with Managed 10T Lustre storage.

TPU 8i: The sampling and serving specialist

Optimized for post-training and high-concurrency reasoning, we designed TPU 8i with our highest on-chip SRAM, a new Collectives Acceleration Engine (CAE), and a new serving-optimized network topology called Boardfly.

Large on-chip SRAM: With 3x more on-chip SRAM over the previous generation, TPU 8i can host a larger KV Cache entirely on silicon, significantly reducing the idle time of the cores during long-context decoding.

Figure 4: TPU 8i ASIC block diagram

The Collectives Acceleration Engine (CAE): To solve the sampling bottleneck, TPU 8i uses the CAE, which aggregates results across cores with near-zero latency, specifically accelerating the reduction and synchronization steps required during auto-regressive decoding and "chain-of-thought" processing. For each TPU 8i chip, there are two Tensor Cores (TC) on-core dies and one CAE on the chiplet die, replacing four SparseCores (SCs) on core dies in previous-generation Ironwood TPU. By integrating a specialized CAE, TPU 8i further reduces the on-chip latency of collectives by 5x. Lower latency per collective operation means less time spent waiting, directly contributing to higher throughput required to run millions of agents concurrently.
Boardfly ICI topology: While the 3D torus allows connecting thousands of chips to be used in cohesion, a large mesh does have more hops between chips and higher all-to-all latencies. For 8i, we changed how the chips connect together in fully connected boards that are then aggregated into groups. Utilizing a high-radix design, we connect up-to 1,152 of these chips together, reducing the network diameter and the number of hops a data packet must take to cross the system. By slashing the hops required for all-to-all communication (the heart of MoE and reasoning models), Boardfly achieves up to a 50% improvement in latency for communication-intensive workloads.

Figure 5: TPU 8i hierarchical Boardfly topology building up from a building block of four fully connected chips into a fully connected group of eight boards, with 36 of such groups fully connected into a TPU 8i pod

Boardfly consists of the following elements, and its topology is hierarchical by nature:

Building Block (BB): Each tray forms a four-chip ring using internal ICI links, providing 16 external connections for broader networking.
Group (G): Eight boards are fully connected via copper cabling to create a localized group, utilizing 11 of the available external links for intra-group communication.
Pod structure: The final architecture scales to 36 groups (up to 1,024 active chips) linked through Optical Circuit Switches (OCS), ensuring a maximum latency of seven hops for any chip-to-chip communication.

Deep dive: The Boardfly vs. torus math

Why move away from the torus for TPU 8i? It comes down to network diameter.

In a 3D torus, nodes are arranged in a grid where each dimension wraps around like a ring. To reach the furthest possible chip in a 8 x 8 x 16 (1024-chip) configuration, a packet must traverse half the distance of each ring:

3D torus = 8/2(X) + 8/2(Y) + 16/2(Z) = 16 hops

While the torus is highly efficient for the neighbor-to-neighbor communication typical of dense training, it creates a latency tax for all-to-all communication patterns. In the era of reasoning models and MoE, where any chip may need to talk to any other chip to route a token, this hop count matters.

Boardfly’s high-radix topology is inspired by Dragonfly topology principles. By increasing the number of direct optical long-haul links between groups of boards, we flatten the network. For that same 1024-chip pod, Boardfly reduces the network diameter from 16 hops down to just seven.

This 56% reduction in network diameter translates directly to lower tail latency, so that the TPU 8i CAE isn't left waiting for data to arrive from across the pod.

Figure 6: A visual representation of the maximum seven-hop ICI network diameter via optical circuit switch on TPU 8i pod

TPU 8t and TPU 8i at a glance

Feature	TPU 8t	TPU 8i
Primary Workload	Large-scale pre-training	Sampling, serving, and reasoning
Network Topology	3D torus	Boardfly
Specialized Chip Features	SparseCore (Embeddings) & LLM Decoder Engine	CAE (Collectives Acceleration Engine)
HBM Capacity	216 GB	288 GB
On-Chip SRAM (Vmem)	128 MB	384 MB
Peak FP4 PFLOPs	12.6	10.1
HBM Bandwidth	6,528 GB/s	8,601 GB/s (~1.3x of TPU 8t)
CPU Header	Arm Axion	Arm Axion

Software enablement: A performance-first AI stack

Hardware is only as powerful as the software that drives it. The eighth generation of TPUs are built on the same performance-first stack we pioneered with the seventh-generation Ironwood TPUs, designed to make custom kernel development accessible without sacrificing the abstraction of high-level frameworks. This stack includes:

Pallas and Mosaic: We provide first-class support for Pallas, our custom kernel language that lets you write hardware-aware kernels in Python. This enables you to squeeze every drop of performance out of the TPU 8i CAE and the TPU 8t SparseCore.
Native PyTorch experience: We're thrilled to share that native PyTorch support for TPUs is now in preview. If you're currently building and serving models on PyTorch, we've made it easier than ever to start using TPUs. You can bring your existing models to our TPUs just as they are, complete with full support for the native features you rely on, such as Eager Mode.
Portability: The same JAX, PyTorch, or Keras code that runs on Ironwood scales to this generation. Accelerated Linear Algebra (XLA) handles the complex translation of the Boardfly topology and CAE synchronization behind the scenes, allowing you to focus on your model, not the interconnect.

Generation over generation: The performance leap

Our commitment to co-designing hardware and software continues to pay dividends. When compared to seventh-generation Ironwood TPU, the eighthgeneration TPUs deliver massive gains:

Training price-performance: TPU 8t delivers up to 2.7x performance-per-dollar improvement over Ironwood TPU for large-scale training.
Inference price-performance: TPU 8i delivers up to 80% performance-per-dollar improvement over Ironwood TPU, particularly at low-latency targets for large MoE models.
Energy efficiency: Both chips deliver up to 2x better performance-per-watt, critical for scaling the next generation of AI sustainably.

Looking ahead

To empower Google Cloud customers pioneering the next wave of innovation, we designed TPU 8t and TPU 8i as two distinct, specialized systems tailored to the multifaceted future demands of the AI lifecycle. TPU 8t and 8i are both purpose-built for the most demanding serving and training workloads, fully integrating with the AI Hypercomputer software stack: JAX, PyTorch, vLLM, XLA, and Pathways. This specialization and ground-up redesign, all in deep collaboration with Google Deepmind, delivers exceptional price-performance and power efficiency.

The modularity of our eighth-generation architecture provides a clear and unique roadmap for the future. Just as every major shift in the computing landscape has required infrastructure breakthroughs, so does the agentic era. Reasoning agents that plan, execute, and learn within continuous feedback loops cannot operate at peak efficiency on hardware that was originally optimized for traditional training or transactional inference; their operational intensity are fundamentally distinct. Our eighth-generation TPU infrastructure has evolved to meet these specific requirements head-on.

To learn more about the eighth-generation TPU family:

Gemini Enterprise for the agentic task force: introducing long-running agents, agentic collaboration spaces, advanced governance, and more

Wed, 22 Apr 2026 12:00:00 +0000

As we have deployed Gemini Enterprise across various industries over the last few months — including the world's most complex and heavily regulated organizations — a clear shift has emerged. Enterprises don't just need AI to answer questions or complete isolated tasks. They also need autonomous agents that can execute complex, multi-step workflows, capable of running for hours or days.

Companies are ready to build their agentic task force, but this demands doing so within a secure and governed environment. This includes creating and deploying agents with their own identity, registry, and gateway so they can always be traced, monitored, and managed.

Today, we are expanding Gemini Enterprise to meet this demand, transitioning AI from an isolated productivity tool into a secure, collaborative, autonomous engine for your business. Gemini Enterprise is the end-to-end system for the agentic era, including both the Gemini Enterprise app and our newly announced Gemini Enterprise Agent Platform.

At Google Cloud Next, we are announcing several new capabilities in the Gemini Enterprise app, including the ability to:

Combine generative intelligence and deterministic business logic to manage enterprise reliability: Build agents ranging from simple assistants to highly complex autonomous orchestrators—without writing a single line of code. And, using our enhanced Agent Designer and reusable Skills, you can construct these workflows entirely with natural language or through a visual interface.
Scale the agentic task force with autonomy: Build and deploy long-running agents that independently solve complex problems up to days at a time, managed entirely through a new, unified Inbox command center.
Collaborate with your colleagues and agentic task force in real-time: Transition from individual AI to collaborative, team-level productivity with Projects — a shared workspace for teams and agents—and the integrated Canvas editor.
Govern with confidence with complete agent observability and traceability: The Gemini Enterprise app is built on our new Agent Platform, meaning enterprise-grade governance, observability, and granular control are built-in from day one.
Connect your ecosystem: Orchestrate actions across Google Workspace, Microsoft 365, and a growing ecosystem of partner agents and data connectors.

Automate complex work with a digital taskforce

With our suite of agent building and management tools, anyone can manage a full spectrum of work — from immediate, ad-hoc tasks, to asynchronous, multi-step workflows that don’t require constant manual prompting.

Enhanced Agent Designer: Create an agent using natural language or a visual interface. Balance generative intelligence with deterministic nodes and flows that carry your business logic reliably. You can inspect, test, and approve every step of a workflow before it ever runs, ensuring enterprise transparency and trust. You can also seamlessly integrate human-in-the-loop checkpoints — ensuring your team can review actions, grant approvals, or provide expert guidance exactly when it matters most. These agents don’t just execute; they learn and improve with every human interaction.

Skills for reusable actions: Codify your unique expertise into reusable Skills. Skills let you formalize a specific workflow — like applying brand guidelines or formatting a report — and save it as a reusable action for your whole team. Instead of re-explaining the context every time, the agent uses the appropriate skill automatically. Because our models dynamically load these Skills, your agents remain incredibly cost-efficient and laser-focused. They pull in deep expertise only when the task requires it, without cluttering the model’s reasoning.

Long-running agents: Build and direct agents that can handle massive, multi-step workflows like end-to-end financial reconciliation or deep sales prospect sequencing, from ad hoc to recurring tasks. Operating autonomously from hours to days inside Google's secure cloud sandboxes, these agents orchestrate complex business logic in the background — powering through mission-critical work without requiring constant human supervision.

Inbox for agent management: Inbox is a central command location to monitor, guide, and securely manage all of your agent activity. Notifications are intuitively categorized into actionable groups like "Needs your input," "Errors," and "Completed," giving you a clear, consolidated view of ongoing agent progress.

Dynamic and interactive Agent UI experiences: Move beyond simple text chat. With new support for the Agent-to-UI protocol (A2UI), custom agents can dynamically generate rich, native UI components — like interactive data visualizations and structured forms — directly within the Gemini Enterprise app.

High-code agents: Allow your engineering teams and IT leaders to build sophisticated agents with full code and publish them directly into the Gemini Enterprise app for the business to use.
Data Insights agent for unified analysis: Data Insights is a native, out-of-the-box agent built by Google that uncovers deep patterns across your entire data landscape. By bridging structured data in warehouses and databases with the rich context of unstructured sources like docs, email, and chat, it provides a 360-degree view of your operations. The agent dynamically generates SQL and search queries to produce rich data visualizations — answering not just the “what,” but the critical “why” behind your business outcomes.

Enhanced Deep Research agent: Execute complex, multi-step research plans that autonomously synthesize information from the open web and internal company data. By combining iterative reasoning with long-running inference, the Deep Research agent can work in the background for hours to deliver comprehensive, citation-backed reports and executive summaries.

Team productivity: Projects and Canvas

Real work is a team sport. We are moving AI from a solo conversation to a persistent, shared team asset.

Projects: A dynamic, highly collaborative workspace where human experts and agents actively co-create side-by-side. By unifying your shared context across various sources — Google Workspace, Microsoft OneDrive, team chats, and more — the agent acts as a dedicated expert on your specific initiative. Whether your team is collaborating in real time or catching up asynchronously, you can work together to brainstorm new ideas, co-create key artifacts and make informed decisions that ultimately drive impactful outcomes, without ever having to sift through fragmented context.

Canvas: Stop tab-switching with an interactive editor built directly into Gemini Enterprise. Canvas allows teams to create and edit Docs or Slides in a single pane, with rich formatting and live co-editing. We’ve also added Microsoft 365 interoperability, allowing you to export the docs and slides you've created within Canvas into common Microsoft Office formats.

An open and connected enterprise ecosystem

The enhanced Agent Gallery: Your Agent Gallery is the unified destination for every agent your teams need — whether built by Google, created internally by your company, or discovered through our partners. Today, we are launching Agent Marketplace integrated into the Agent Gallery, giving your users the ability to easily browse and request specialized third-party agents from partners like Accenture, Oracle, and ServiceNow. Once approved, these agents seamlessly appear in the Agent Gallery, giving your team access to execute complex workflows with natural language right from within Gemini Enterprise.

Integrations for Bring Your Own Model Context Protocol (BYO-MCP): In addition to a rich set of rigorously validated, out-of-the-box connectors, we are introducing support for BYO-MCP. This allows admins to securely connect the Gemini Enterprise app directly to your own custom or third-party business tools, internal workflows, and custom-built applications. This capability expands your automation potential by allowing no-code agents to discover and execute tools hosted on your own servers — tailoring the AI experience to your specific enterprise workflows.
Cross-platform unified intelligence: Gemini Enterprise delivers unified intelligence across your entire fragmented data landscape. Whether your data lives in Google Workspace, Microsoft 365, or other business applications, Gemini Enterprise strictly respects native permissions — ensuring your teams get powerful insights without ever compromising your company's governance or user access controls.

Enterprise-scale governance comes standard without additional cost

Confidence in enterprise AI is built on layers of trust. The Gemini Enterprise portfolio provides it at every level – without added cost. The entire system is built on Google’s secure-by-design infrastructure, with Agent Platform at its core providing access to universal governance controls. Any agent built on both Agent Platform and the Gemini Enterprise app is governed by the same universal rules, giving businesses a deeply integrated way to help secure agentic work.

Key controls now include:

Enforce granular access controls: Agent Identity lets you assign unique digital IDs and help enforce the principle of least privilege for every agent no matter where it runs.
Organize and share approved agents: Agent Registry gives IT teams a central catalog to curate and distribute approved, reusable agents for their employees.
Turn “Shadow AI” into “Managed AI”: Agent Gateway gives IT teams a centralized management point for network policies, data access, and security guardrails — protecting against risks like prompt injection. Powered by our Context Aware Access capabilities, the Gateway enforces your company-specific rules and ensures agents never send data to unapproved endpoints.

Learn more about Gemini Enterprise Agent Platform here.

Availability

New features will roll out to customers over the coming months. Ready to explore these new features and empower your teams? Get started with Gemini Enterprise today.

Announcing new partner-supported workflows for Google Security Operations

Wed, 22 Apr 2026 12:00:00 +0000

Security teams are frequently burdened with manually stitching together telemetry, alerts, and response playbooks. This fragmentation can limit visibility, increase alert fatigue, and slow down investigations.

Defending the modern enterprise requires tools that work together. Today at Google Cloud Next, we are thrilled to announce a robust cohort of new partner integrations for Google Security Operations as part of the Google Cloud Security integration ecosystem.

Designed to deliver high-fidelity security workflows right out of the box, our newest partners to join our ecosystem with more than 300 vendors include: Beacon Security, Contrast Security, Darktrace, Gigamon, GreyNoise, Intezer, Prophet Security, SAP, Synqly, Thinkst, Tidal Cyber, Torq, and Vali Cyber.

Here’s how our partners are building in the Google Security Operations ecosystem, the integration types supported, and how security operations centers (SOC) can use them.

Specificity and depth: Supported integration types

The Google Security Operations platform supports several distinct integration patterns. Here is how our current cohort is using these architectures to deliver specific technical capabilities:

1. Data feed integrations for deep visibility across your stack
These integrations pipe crucial telemetry directly into the Google Security Operations data lake, pre-mapped to our unified data model (UDM) schema so your team doesn't have to write custom parsers:

Beacon Security: Architects ingestion for both normalized and raw data. Beacon expands your coverage by collecting data from sources including APIs, syslog, webhooks, and cloud storage. Using a real-time streaming pipeline, it normalizes these raw events directly into out-of-the-box UDM mappings in minutes. Before data even reaches Google Security Operations, Beacon applies security-driven data reduction to filter and aggregate events preserving detection fidelity. Finally, it uses AI-powered data orchestration and continuous security data posture management to track collection health and help reduce the risk of blind spots becoming breaches.
Contrast Security ADR: Detects, investigates, and responds to application-layer attacks with the Contrast ADR and Google Security Operations integration. Verified runtime attack telemetry streams into Google's UDM, powering purpose-built detection rules that automatically surface confirmed exploits as cases and correlate application-layer findings with signals from WAFs, EDR tools and database security sensors.
Gigamon GigaVUE Cloud Suite: Introduces a new integration to help organizations close visibility gaps across hybrid cloud environments. This integration amplifies the power of Google Security Operations with actionable application and network-derived telemetry — including packets, flows, and metadata — from Gigamon, giving teams the context they need to detect threats earlier and investigate with greater precision.
SAP Logserv: Closes the visibility gap between SAP Logserv and security operations, empowering analysts to detect, investigate, and respond to SAP-specific threats alongside their existing IT landscape. The integration features out-of-the-box ingestion and uses SAP-specific standard parsers to normalize raw, complex infrastructure and application logs into the UDM format. This gives teams unified, enterprise-wide visibility to defend business-critical data while reducing the need for deep SAP technical expertise or custom log pipelines. This integration has been developed by Google, in partnership with SAP.
Synqly Mesh: Offers a unified API that performs bi-directional data normalization between Google Security Operations' UDM and the Open Cybersecurity Schema Framework (OCSF). It supports event ingestion configurations (Sink) as well as full bi-directional SIEM connectivity.
Vali Cyber Zero Lock: Streams hypervisor-level security events directly into your existing Google Security Operations workflows. This integration provides visibility into emerging ESXi threats and is designed to help keep virtual infrastructure protected and operational.

2. Response integrations for streamlined alert and case management
These integrations hook directly into your workflows, allowing external platforms to trigger alert delivery, create cases, and execute automated actions.

Darktrace: Currently in development, this response integration enables Google Security Operations to ingest Darktrace Incidents and Model Alerts. By pulling in pre-parsed raw logs via API or webhook, this integration provides your team with network context needed to streamline alert delivery, manage cases, and trigger automated response actions.
GreyNoise: New integrations that enhance detection and response capabilities in Google Security Operations. Spanning both SIEM and SOAR, the integration delivers standardized indicator ingestion, pre-built dashboards, YARA-L detection rules, saved searches, webhook support, response actions, and ready-to-deploy playbooks.
Thinkst Canary: Integrates directly with Google Security Operations SOAR, allowing security teams to ingest high-confidence Canary incidents as actionable cases. It preserves full alert context, surfaces extracted entities like IP addresses and hostnames, and allows analysts to acknowledge incidents without ever leaving their Google Security Operations workflow.
Torq: Brings its AI SOC Platform to Google Security Operations to help automate the threat lifecycle. Torq pulls detections directly via API, applies agentic AI auto-triage to filter out noise, and executes autonomous response actions — like isolating endpoints or revoking access — across the security stack while keeping Google Security Operations updated with case status.

3. Pulling Google Security Operations data (bi-directional API workflows)
Security doesn't just happen in one console. These integrations use secure APIs to pull Google Security Operations detections and intelligence natively into partner platforms, bridging the gap between tools.

Intezer: Allows you to natively query, investigate, and triage Google Security Operations detections without leaving your established environment. It automatically ingests Google Security Operations alerts directly into Intezer, which then queries your underlying Google Security Operations data during active investigations to drive autonomous triage. This bi-directional workflow ensures your team has the full picture — eliminating the need to pivot between consoles, reducing manual data gathering, and freeing your analysts to focus on high-level decision-making and rapid response.
Prophet Security: Integrates with Google Security Operations to provide AI-powered alert investigation and natural language threat hunting. It is designed to automatically ingest alerts, queries the Chronicle API for real-time UDM event context, and bidirectionally syncs investigation findings and comments back to Google Security Operations, with the goal of reducing analyst workload.
Tidal Cyber: Pulls configuration and policy data from your cyber defense intelligence (CDI) environment. It can retrieve ATT&CK-mapped curated detection rules and user-created rules from Google Security Operations. It also synchronizes the detection rules states with Tidal to reflect enabled and disabled capabilities. By knowing both what a product is capable of and what's currently enabled in your environment, Tidal helps identify configuration gaps and assists in keeping your defensive stack and coverage map accurate as policies change.

Details on all partner integrations can be found in our technical documentation or in your Google Security Operations Content Hub console.

Unify your defense today

For technology vendors and developers looking to join the Google Cloud Security integration ecosystem, you can get started by downloading the Google Security Operations Build Partner Guide to understand our UDM schema and API requirements, and reach out to our Google Cloud Security Tech Partners team to request a development environment to accelerate your build in time for our next release cycle.

You can follow all of our security announcements at Next ‘26 here.

Next-gen FinOps for the AI era

Wed, 22 Apr 2026 12:00:00 +0000

Today we’re excited to announce the next generation of our FinOps product suite to help our customers increase operational efficiency, better understand their costs, and control them with Spend Caps.

What’s new: We’re introducing a new FinOps Explainability agent, which is designed to operate autonomously, and investigate the drivers of your AI-related Cloud costs. This is in addition to new FinOps tooling which provides commercial auditability.

We’re also announcing a private preview of Spend Caps in Google Cloud, enabling FinOps and DevOps managers to set budgets and enforce cost boundaries at the project level for Google AI Studio (AIS), Gemini Enterprise Agent Platform (the evolution of Vertex AI) , Cloud Run, Cloud Run Functions, and Maps. These caps alert and ultimately pause API traffic once your set budget is reached.

Why it matters for your business: These new FinOps tools give you clear visibility into AI costs, increase control with Spend Caps to prevent overspending, and offer the commercial flexibility needed to scale your AI innovations efficiently. Customers who are using our existing FinOps tools are seeing huge improvements. Since launching Gemini Cloud Assist (GCA) for FinOps last year, cost reporting adoption has surged 75% while simultaneously slashing customer time spent doing FinOps cost analysis by 18%.

Where to get started: You can access the FinOps Explainability Agent in the console here, along with the FinOps tooling. Customers can sign-up for the private preview of Spend Caps here.

Goodbye static reporting

Given the number of variables and the number of services commonly running for a large enterprise, cloud cost reporting can be noisy. Even when, in theory, AI costs are just the result of quantity (q) times price (p). Quantity can be driven by a large mix of variables such as API request traffic, error logs, fluctuating token counts, or even cloud storage. And price often fluctuates with different AI model types, and frequent provider price shifts. This challenges FinOps and DevOps managers to synthesize this data to identify efficiency opportunities, or take timely action.

In Google Cloud Billing we used Gemini to develop our new FinOps Explainability agent to autonomously help users understand the drivers of AI costs. Attributing ROI to AI projects requires a clear understanding of its costs, but because AI often piggybacks on existing infrastructure, its expenses frequently blur into the general cost of doing business.

Now you can use the FinOps Explainability agent to identify your AI cost drivers automatically, and use it to answer questions like: “How much did I spend on Gemini 1.5 Pro versus Gemini 1.5 Flash?” Or, “Break down my total spend by API Key so I can see which integration is expensive.” Or, “Show me the split between Input Token costs and Output Token costs for Gemini 3.0 Pro.” Users can quickly discover what services and which projects are driving your AI costs.

FinOps Explainability agent helps you analyze AI costs, drivers & trends

Hello automated Spend Caps

The speed of AI adoption and usage is driving cloud spend that behaves differently than traditional cloud spend. AI uses specialized hardware (TPU/GPUs) and a single runaway training job or unoptimized model running on that hardware can drain a budget in a very short amount of time. Users are also constantly experimenting. Traditional cost control tools typically alert managers, but don’t enforce budget caps. The result: many enterprises have been forced to build their own complex custom spend guardrails that are enforced through destructive actions that may be time consuming to adjust, such as disassociating forms of payment.

We’re excited to announce that Spend Caps are coming soon to Google Cloud. Designed to work with Google Cloud Budgets, FinOps and DevOps can set budgets that enforce automated cost boundaries (caps) at the project level for AIS, Agent Platform, Cloud Run, Cloud Run Functions, and Maps. These caps alert and ultimately pause API traffic once your set budget is reached, but leave your resources intact. If you need the traffic to resume, simply suspend the Spend Cap.

We expect customers that want to contain the costs of AI R&D to benefit immensely from this new feature. You can sign up for the private preview today.

Spend Caps help prevent cost overruns

Real commercial incentive auditability.

Google Cloud meets you at every stage of growth—offering the commercial flexibility, startup programs, and enterprise incentives needed to help your costs scale efficiently. To help users more clearly understand the connection between commercial agreements and the services being billed, we’ve designed our FinOps tooling to provide end-to-end auditability of our commercial obligations. With the private preview rollout of enhanced billing account hierarchies, customers can view their aggregated spend across multiple billing accounts, including Other Eligible Services (OES) spend. Additionally, we are announcing a private preview for Google Cloud contract commitment reporting, providing visibility into Google Cloud commit contract burndown within your Enterprise Agreement.

The future of FinOps is here. Built with AI for AI.

With the FinOps Explainability agent for deep visibility, Spend Caps for increased control, and enhanced billing account hierarchies with contract commitment reporting for ultimate commercial flexibility, Google Cloud is empowering you to scale your AI innovations with confidence and precision.

Storage innovations to accelerate your AI workloads at Next ‘26

Wed, 22 Apr 2026 12:00:00 +0000

At Google Cloud Next, we are announcing innovations across every layer of our storage stacks — performance, intelligence, and management — to ensure your data is as fast and as useful as the AI models, apps and agents you are building.

Why it matters: Storage is no longer just a place to keep data. When training AI models, storage is the engine that feeds data-hungry accelerators. During AI inference, it’s the access layer that makes it responsive, acting as the source for the context that AI agents need to be effective. When storage performance falls short, accelerators sit idle, agents respond slowly, and data remains invisible to AI models.

But storage performance is only half the battle; you also need storage that’s smart. With the help of Google’s AI models integrated directly into the storage layer, you’re no longer just storing bits, but data that has full context about its content. In this new era of smart storage, raw data becomes a valuable asset that’s ready to use by a variety of downstream AI and enterprise applications.

What’s new:

High-performance storage infrastructure: New Rapid family of features in Cloud Storage for high-performance object storage; delivering 10x performance enhancements plus a new cost-effective Dynamic tier for Google Cloud Managed Lustre.
Smart Storage: Unlocking unstructured data with automated metadata annotation, and AI agent connectivity via MCP.
Storage Intelligence: Streamlined data management through zero-configuration dashboards, aggregated activity views, and enhanced batch operations.
Enhanced ecosystem: Expanded capabilities across Google Cloud NetApp Volumes, Filestore for GKE, and our backup and data protection portfolio.

Let’s take a deeper look at the storage enhancements we are unveiling this week.

Storage infrastructure that keeps up with AI

As AI models scale, getting data from the storage to the compute layer fast enough can be a bottleneck. New storage capabilities move performance directly into the storage layer, reducing total cost of ownership (TCO) and keeping accelerators fully utilized.

Cloud Storage Rapid
Cloud-based object storage like our Cloud Storage is scalable and cost-effective, but bottlenecks can stall AI jobs and waste expensive compute cycles. Every time a training cluster waits on a read or a checkpoint write stalls, you're paying for accelerators that aren't doing useful work.

Cloud Storage Rapid marks a fundamental shift in designing AI infrastructure: you no longer have to choose between the reliability of object storage and the high performance of a specialized AI storage system. Cloud Storage Rapid lets you leverage the industry-leading durability, massive distributed scale, and cost-effective auto-tiering of object storage, while simultaneously achieving extreme throughput, frequent I/Os, and ultra-low latency. With native integrations into PyTorch and JAX, Cloud Storage Rapid is optimized out-of-box for the most popular AI/ML ecosystem frameworks, so that your data preparation, training, and inference workloads run on a high-performance and reliable foundation.

The Cloud Storage Rapid family consists of two offerings: Rapid Bucket and Rapid Cache.

Rapid Bucket, now generally available, leverages Colossus, the Google distributed storage system that powers Gemini and YouTube, to deliver more than 15 TB/s of bandwidth, 20 million requests per second, and sub-millisecond latency in a single zonal bucket. With access via high-performance gRPC and S3-compatible APIs, Rapid Bucket increases accelerator utilization for multi-modal training with 50% reduced GPU blocked time and 2.5x faster data loading. Checkpoint restores are 5x faster and checkpoint writes are 3.2x faster compared to traditional object storage, reducing workload interruptions and wasted GPU time.

Checkpoint writes are 3.2x faster and restores are 5x faster with Rapid Bucket

Rapid Cache, formerly Anywhere Cache, accelerates bandwidth for bursty workloads like model loading for inference, delivering an aggregate read throughput of 2.5 TB/s for existing buckets, with no code changes. The new ingest-on-write feature provides up to 2.2x faster checkpoint restores, allowing training clusters to recover faster from interruptions.

Rapid Cache’s combination of simplicity and performance has resulted in strong adoption, including from cutting-edge AI/ML customers like Thinking Machines Lab.

“Rapid Cache has become a core foundation of our AI/ML data infrastructure, supporting our critical workflows, from data prep and pretraining to training and model loading. By acting as a crucial bandwidth shield and booster, it enables us to scale our data-intensive workloads across our entire fleet without compromise, providing us with the on-demand high bandwidth and consistent stability that we need to innovate at speed.” - James Sun, Member of Technical Staff, Thinking Machines Lab

Google Cloud Managed Lustre
The Lustre parallel file system is the industry standard for organizations whose AI training and inference workloads require high throughput and sub-millisecond latency, and is trusted by AI labs and HPC centers worldwide to feed thousands of accelerators simultaneously and keep them saturated under pressure. Google Cloud Managed Lustre brings that capability as a fully managed service, and with today's announcements, it is the most performant managed Lustre offering available in any cloud.

Managed Lustre now delivers up to 10 TB/s of throughput — a 10x increase since last year and 4–20x higher than managed Lustre offerings from other hyperscalers for a single instance. Powered by C4NX VMs and Hyperdisk Exapools, Managed Lustre writes and restores checkpoints 2.6x faster when compared to other Google Cloud storage solutions.

The new Dynamic tier ($0.06/GB-month) delivers the low-latency performance required for intense AI workloads like training and checkpointing. By serving data from persistent disk rather than relying on object-based caching, we eliminate a performance cliff — helping ensure your data remains responsive and your accelerators stay productive. A single SKU provides simple, predictable billing without the hidden complexity of traditional data tiering.

“By integrating Managed Lustre we eliminated the typical onboarding bottlenecks, allowing us to hit the ground running with the inferencing workload. This high-throughput, low-latency storage keeps our B200 GPUs fully saturated, driving a substantial performance gain in LLM inference over the H200. For our customers, this translates directly into faster, more responsive AI agents that can handle complex reasoning at a fraction of the previous latency.” - Lavnaya Karanam, Software Engineering PMTS, Salesforce

Smart Storage: Context for the AI era

The beauty of an object storage system like Cloud Storage has long been its simplicity: the system knows the object’s name, its size, and when it was created. But if you want to understand the object’s content — what entities it references, whether it contains sensitive PII, or whether it’s relevant to a pending query — you need to use custom pipelines, separate databases, and bespoke enrichment systems.

AI has changed the equation. To fine-tune a model, you need to select the right objects from the get-go, from a corpus of millions. Building an agent requires retrieving the right context for each decision. To meet a compliance obligation, you need to know what every file contains up front, before it becomes a liability. In each case, the bottleneck isn’t compute or model quality — it’s the inability to describe, find, and act on objects at scale.

To bridge that gap between stored and usable data, last year we introduced Smart Storage, a set of capabilities built directly into Cloud Storage that makes every object self-describing. New Smart Storage capabilities include:

Automated annotations, which eliminates the need to build and maintain custom annotation pipelines. With Smart Storage enabled, Cloud Storage can now automatically generate context — including image annotations — so your data is discoverable and usable from the moment it lands. You pay to annotate the data once at write time, and every downstream system can use those annotations immediately for the life of the object.
Cloud Storage MCP server lets you read, write, and analyze Cloud Storage data using the standard MCP protocol.

Smart Storage enables these capabilities, and others, thanks to its object context, now generally available. This metadata substrate adds structured, mutable, IAM-governed context to every object. Customers write their own tags and classifications; Google's annotation pipelines automatically attach labels, extracted entities, and compliance signals.

With Smart Storage, ML teams can select training datasets from semantic criteria without building retrieval pipelines. AI agents can ground their reasoning in enterprise data without a separate retrieval layer.

Storage Intelligence: Data management at AI scale

As data estates grow to hundreds of petabytes, storage costs can spike without warning, and security blind spots can multiply across billions of objects. To manage this, teams have to stitch together multiple tools just to answer basic questions about their own data.

Last year we launched Storage Intelligence to give enterprises a unified management experience built directly into Cloud Storage. Today, 70% of our largest customers use Storage Intelligence, each of whom manage over 50 billion objects.

Storage Intelligence provides a single view across your entire project or organization, with unique capabilities like bucket relocations across regions. Today, we're making it significantly more powerful with:

New zero-configuration dashboards instantly surface cost anomalies and integrate Security Command Center’s Data Security Posture Management (DSPM) data governance feature, to detect critical security vulnerabilities across Cloud Storage — no setup required.
New object events and bucket activity tables in Insights Datasets now drive deeper cost analysis and accelerate operational tasks. You can use these insights to perform a wide range of analyses, from optimizing bucket placement based on egress patterns to quickly troubleshooting 429 errors by finding the impacted objects.
Enhanced batch operations make it even simpler to act on billions of objects with new change ACL and storage class operations, and support for multi-bucket operations.

Enhancing the storage ecosystem

Beyond our core storage offerings, we are streamlining how enterprises migrate to and protect data in the cloud.

Google Cloud NetApp Volumes: With the launch of Flex Unified, NetApp Volumes now provides a unified enterprise storage platform that bridges the data center and the cloud, provisioning both block (iSCSI, NVMe/TCP) and file (NFS/SMB) on the same storage pool. New ONTAP-mode lets you bring your existing automation (Terraform, Ansible) and ONTAP APIs directly to NetApp Volumes.
Filestore for GKE: Developers building AI workloads on Google Kubernetes Engine (GKE) can start small, with shares as small as 100 GiB, and scale capacity and IOPS independently. At the same time, tighter integration to the Colossus distributed file system provides more scale and enterprise capabilities.
Data protection: Google Cloud Backup and DR now features agentic AI capabilities that can autonomously audit your backup estate and remediate coverage gaps, with new GA integrations for AlloyDB and Filestore.

Where to start

As you navigate today’s generational AI shift, you need a storage foundation to support ever-larger, more intelligent, and autonomous models. With new high-performance and intelligent storage layers, plus enhanced storage management tools and a deeper data protection bench, Google Cloud’s storage platforms lets you understand and use your enterprise data in ways that weren’t previously possible, allowing you to:

Reduce the AI data bottleneck: Saturate compute and accelerate ROI. Keep your expensive GPUs and TPUs fully productive with high-throughput storage that delivers the extreme performance required for large-scale training and inference.
Build agent-ready data foundations: Shift from building custom pipelines to an active knowledge base where self-describing objects let AI agents instantly reason over data without manual prep.
Minimize blind spots across exabytes: Replace fragmented management tools with zero-configuration dashboards and datasets to instantly surface cost anomalies and security risks across billions of objects.
Embrace the storage ecosystem: Streamline migration and protection. Bridge your data center to the cloud, scale containerized apps, and automate data resilience with agentic AI.

Visit the Google Cloud Storage console to explore these new features, read more about Cloud Storage Rapid, or explore our Next '26 storage sessions.

Welcome to Google Cloud Next ‘26

Wed, 22 Apr 2026 12:00:00 +0000

Editor's note: This is an adapted transcript of Thomas’ keynote remarks for Next ‘26 where we are announcing a number of new innovations including:

Gemini Enterprise Agent Platform, a comprehensive platform to build, scale, govern, and optimize agents.
Gemini Enterprise app includes a new Agent Designer, Inbox for managing agent activity, long-running agents, Skills, Projects and more.
AI Infrastructure innovations include our 8th Generation TPUs, along with new storage and networking capabilities.
Agentic Data Cloud including a cross-cloud Lakehouse and Knowledge Catalog that allow organizations to use their data at the speed and scale required by agentic AI.
Agentic Defense that combines Google’s Threat Intelligence and Security Operations with Wiz’s Cloud and AI Security Platform to prevent, detect, and respond to threats.
Agentic Taskforce includes new AI capabilities across Gemini Enterprise for Customer Experience and Google Workspace.
Customer Success across our entire portfolio.

Welcome to Google Cloud Next ‘26

When we gathered at Next a year ago, we talked about how generative AI was transforming organizations around the world. Today, that future is in-production–the Agentic Enterprise is real–-and deployed at a scale the world has never before seen. At Next, we show you everything you need to make your organization an Agentic Enterprise.

This past year, we have been relentlessly focused on delivering innovation that drives your growth, introducing thousands of new capabilities across our portfolio. Our seventh-generation Ironwood TPUs and our new Axion custom processors are now generally available, providing customers with purpose-built, cost-effective infrastructure designed for the Agentic Enterprise. We launched groundbreaking new foundational models, including Gemini 3.1 Pro and our new image generation model, Nano Banana 2. And we delivered Gemini Enterprise, bringing the best of Google AI to every employee. Instead of piecing together a collection of fragmented systems, we offer a vertically optimized stack where everything is co-developed.

All of this innovation is propelling customer adoption. In fact, nearly 75% of Google Cloud customers are using our AI products to power their businesses. In addition, over the past 12 months, 330 Google Cloud customers each processed more than one trillion tokens, while 35 reached the 10-trillion-token milestone with our models. Our first-party models now process more than 16 billion tokens per minute via direct API use by our customers, up from 10 billion last quarter.

Today, organizations are choosing Google Cloud for three primary reasons:

Fully integrated AI stack: Organizations partner with us for our deep-rooted expertise and our ability to deliver first-party solutions across the entire AI stack. This isn't about offering individual services that can be cobbled together; it is about providing a comprehensive backbone for innovation. Our full stack approach across AI infrastructure; leading models like Gemini; data management capabilities; enterprise-grade foundation; developer tools and platform; and agents and applications make us unique in the market.
Secure-by-design, enterprise-ready foundation: Our customers trust Google Cloud because we provide a secure, unified platform built on Zero Trust security and planet-scale reliability. This is the same foundation that supports Google’s most demanding global services. Not only is our platform secure by default, but we also offer leading intel-driven and AI-powered cybersecurity solutions. These integrate development and security operations across hybrid and multicloud environments to effectively prevent, detect, and respond to threats up and down the entire AI stack.
Gemini Enterprise, the front door to AI for every employee and customer: Gemini Enterprise enables every customer and employee to build, use, deploy, and manage cutting-edge AI agents. Whether it's bringing together sales and service interactions to improve customer touchpoints, automating complex workflows to improve your supply chain efficiency, or enabling marketers to prototype and create campaigns in record time, Gemini Enterprise is transforming the way companies get work done.

Today at Next ‘26, we are announcing everything you need to make your organization an Agentic Enterprise – all anchored by a unified stack designed to turn intelligence into a growth engine for your business. Gemini Enterprise is now the end-to-end system for the Agentic Era—the connective tissue between your data, your people, and all of your apps and agents that transforms all of your processes into a single, intelligent flow. New innovations announced today include:

Gemini Enterprise Agent Platform is our new, comprehensive platform to build, scale, govern, and optimize agents. Built on top of our leading infrastructure, and deeply integrated with our data and security capabilities, it is the foundation for the Agentic Enterprise. It brings together the best of Vertex AI with transformational new features, including Agent Studio, Agent-to-Agent Orchestration, Agent Registry, Agent Identity, Agent Gateway, Agent Observability, and much, much more. This empowers your technical teams with a comprehensive, unified platform to manage every aspect of your agents, and deliver them through the Gemini Enterprise app, the front door for AI for every employee.
Gemini Enterprise app includes new features that let every employee build with AI, including a new Agent Designer that helps build sophisticated schedule- or trigger-based agents, long-running agents to execute complex business processes, an Inbox for managing agent activity, Skills to create shortcuts for repetitive tasks, and Canvas to create and edit files without switching applications.
AI Infrastructure developments include our eighth-generation TPUs, with TPU 8t for accelerated training and TPU 8i for cost-effective, near-zero latency inference, along with storage innovations like Managed Lustre with 10 TB per second of throughput, and networking advancements like Virgo Networking for high-performance at massive scale.
Agentic Data Cloud introduces a new, AI-native architecture that allows your data to be utilized at the speed and scale required by agentic AI. We are announcing Knowledge Catalog to ground agents in trusted business context across your entire data estate, a Data Agent Kit for Gemini-powered data science authoring, and a cross-cloud, AI-native Lakehouse that gives you seamless access to all of your data, regardless of where it’s located, unleashing a borderless foundation for autonomous action.
Agentic Defense is delivered via a cybersecurity platform that combines Google’s Threat Intelligence and Security Operations with Wiz’s Cloud and AI Security Platform to detect, prevent, and respond to threats. A number of new agents for threat detection, detection engineering, remediation and more, along with Wiz’s AI Application Protection Platform (AI-APP), provide autonomous protection from code, to cloud, to runtime, across multicloud, hybrid, and AI environments.
Agentic Taskforce includes new innovations across our Gemini Enterprise for Customer Experience solution, as well as new capabilities in Google Workspace including Workspace Intelligence, which streamlines productivity by providing a unified context layer across Workspace apps.

These are just a few of the highlights, and I’ve included much more detail below.

Gemini Enterprise: Bringing the Agentic Enterprise to life

Gemini Enterprise is seeing tremendous momentum. In Q1, we saw 40% growth in paid monthly active users quarter-over-quarter. And today we are announcing Gemini Enterprise Agent Platform, which has everything your technical teams need to build and orchestrate agents. It brings together the model selection, model building, and tuning services of Vertex AI that customers love, along with new features for agent integration, security, DevOps, orchestration, and more. It also includes the Gemini Enterprise app, which brings the best of Google AI to every employee, for every workflow.

Agent Platform provides first-class access to the world’s leading models, including Gemini 3.1 Pro—our most advanced model, optimized for complex workflow orchestration; Gemini 3.1 Flash Image, also known as Nano Banana 2, to generate high-fidelity user interfaces and visual assets; Lyria 3—our state-of-the-art model for generating professional-grade audio and music; and leading models from Anthropic, including Claude Opus, Sonnet and Haiku. And today we are also adding support for Anthropic’s Claude Opus 4.7.

The platform is designed around four core pillars—build, scale, govern, and optimize—that allow teams to collaborate seamlessly, and we are announcing a number of new features that allow you to:

Build agents quickly and easily by empowering your developers, business users, and everyone in-between to build and deploy agents at scale.

Agent Developer Kit (ADK) has a new graph-based framework that allows you to define clear, reliable logic for how agents work together to solve complex problems.
Agent Studio, our new low-code interface, enables every developer to build, test, and publish agents using natural language.
Agent Registry provides a single point of control, indexing every internal agent and tool across your organization to ensure they are discoverable and governed.
Ecosystem Integrations via Agent Marketplace help instantly deploy specialized agents from partners like Atlassian, Box, Lovable, Oracle, ServiceNow, Workday, and many more—directly from Gemini Enterprise. Also, you can leverage new integrations that enable agents from partners that use a unified Agent Registry with reusable skills and managed environments to treat all protocols as first-class citizens.
Model Context Protocol allows you to connect our Agent Platform to any MCP Server. We are also exposing Google Cloud Services as MCP to allow you to interact seamlessly with Google Cloud from any agent.

Scale and orchestrate agents through a modular set of services that can be used together or independently to streamline the path from agent proof-of-concept to production via our highly scalable and secure container for running agents in production.

Agent-to-Agent Orchestration enables agents to seamlessly delegate tasks to one another, including support for complex, generative, and deterministic orchestration patterns. This ensures that for critical flows such as compliance, your agents can follow well-specified paths every time.
Runtime improvements deliver the immediate responsiveness your users expect with a revamped foundation that delivers sub-second "cold starts" and the ability to provision new agents in seconds. It also helps reduce the overhead of idle infrastructure.
Support for long-running agents enables your agents to sustain autonomous reasoning and manage multi-step workflows.
Secure sandboxes provide isolated environments for agents to execute model-generated code and perform computer use tasks like browser-based automation—without risk to your host systems.
Memory Bank and Memory Profiles give your agents scalable, long-term memory that replaces temporary sessions with persistent context. Agents can recall specific user details and project histories over months to deliver deeply personalized interactions.
Agent Sessions enhance developer flexibility by allowing the use of unique identifiers for precise session tracking across disparate enterprise systems.

Govern agents with our secure-by-design architecture that allows you to manage a fleet of agents with the same rigor you apply to other mission-critical systems.

Agent Identity improves the security posture of your agents by ensuring every agent receives a unique cryptographic ID and well defined authorization policies that are traceable and auditable. With Agent Identify, agents are governed through your enterprise policies, stay within assigned tasks, and ask a human for help when a situation gets too complex.
Agent Gateway provides the air traffic control for your agent ecosystem. It understands agentic protocols (MCP, A2A, etc.) and provides centralized, real-time policy enforcement—helping to ensure every interaction between your agents and your data is secure, compliant, and scalable by design. Integrated inline protection with Model Armor safeguards your agent interactions against prompt injection, tool poisoning, and sensitive data leakage.
Agent Anomaly Detection proactively flags suspicious behavior before it impacts your business. By analyzing the intent behind agent actions, it automatically detects risks, such as tool misuse, unauthorized data access, and reasoning drift that can lead to unintended outcomes.
Agent Security dashboard gives you a full view of AI risks and compliance using the Security Command Center. It allows your teams to map relationships between agents and models, automate asset discovery, and scan for vulnerabilities in the underlying operating system and language packages.

Optimize agents using data-driven feedback loops to improve accuracy, reduce latency, and control costs, helping to ensure every AI agent performs exactly as intended.

Agent Observability delivers turnkey dashboards with automated logging and agent auditing for total oversight. Using standardized, OTel-compliant telemetry, you can now verify every agent, tool, and API handoff to ensure accountability, visualize full execution paths, and quickly diagnose reasoning loops or monitor performance metrics.
Agent Simulation lets you stress-test agents at scale against human-like synthetic interactions in a controlled environment. Generating thousands of synthetic, multi-step customer interactions, Agent Simulation rigorously evaluates logic and conversation flow.
Agent Evaluation continuously scores agent performance against live production traffic using multi-turn autoraters to detect quality degradation and anomalies in real-time.

Gemini Enterprise app has all the key components to let teams discover, create, share, and run AI agents in a single environment. Today we are introducing several new capabilities in the Gemini Enterprise app:

Agent Designer uses the same no-code agent designer experience of Agent Platform and lets employees build sophisticated schedule- or trigger-based agents using any enterprise connector. It gives you a virtual flowchart of your agent, allowing you to inspect, test, and approve workflows, ensuring total transparency for executing critical business processes.
Long-running agents are designed to execute complex business processes. They can work autonomously in secure cloud sandboxes, giving agents the ability to orchestrate business logic, write code to build custom tools, and complete multi-step work like reconciliation activities or sales prospect sequencing—without needing constant prompting.
Inbox in Gemini Enterprise provides a central location to monitor, guide, and help manage all of your agent activity, including your long-running agents. Notifications are intuitively categorized into actionable groups like "Needs your input," "Errors," and "Completed.”
Projects create a dedicated space where the agent’s memory is confined to the files and conversations your team adds. By connecting it to data sources like Google Drive, NotebookLM, and Google Group Chats, the agent becomes an expert on a specific topic and can provide team members daily briefings or status updates without digging through months of documents.
Skills create simple shortcuts using an “@” mention for repetitive tasks like applying brand guidelines, formatting a report, or accessing specific data.
Canvas gives our customers an interactive editor directly within Gemini Enterprise. It allows teams to easily create and edit Docs and Slides, and even export to Microsoft 365 files, within the same experience.
Agent Gallery provides access to third-party agents from partners like Adobe, Atlassian, Lovable, and ServiceNow, and is adding more third-party connectors for Asana, Mailchimp, Workday, and more. These integrations enable your agents to retrieve data and execute tasks with your systems-of-record.

We have seen tremendous adoption of Gemini Enterprise to date from companies including:

Bosch is scaling innovation from power tools to automotive technology. By integrating Gemini Enterprise into AskBosch—the company’s internal platform for accelerating AI—employees across the organization are now deploying their own agents.
Capcom is using Gemini Enterprise Agent Platform and its proprietary technology to build playtesting and predictive agents that improve the gameplay experience, freeing developers for creative work.
Citi Wealth today unveiled “Citi Sky,” an always-on, AI-powered member of the Citi Wealth team that will reshape how clients in the United States access market insights, act on opportunities, and engage with their financial advisors.
EcoVadis is deploying Gemini Enterprise to increase its productivity and operational efficiency, enhance its client value, and develop new growth opportunities.
FairPrice Group is advancing its ‘Store of Tomorrow’ program with Google Cloud's Gemini Enterprise Agent Platform. It has integrated Gemini-powered AI agents into its Smart Carts to elevate the grocery shopping experience across in-store and digital touchpoints. Its employees are also using solutions built with Google Cloud's multimodal AI and generative media models to streamline store operations and accelerate thematic ad creation.
GE Appliances is democratizing AI by putting it in the hands of the people closest to the work, with more than 800 AI agents deployed across manufacturing, logistics, and supply chain.
KPMG achieved 90% Gemini Enterprise adoption among its employees, with more than a hundred agents in the first month to better serve its global clients.
Macquarie Bank is supercharging productivity with Gemini Enterprise, reclaiming more than 100,000 hours of team members’ time.
Mars is adopting Gemini Enterprise as its primary AI operating system for its global Associates to build sophisticated AI agents that streamline marketing, R&D, and enterprise search. Gemini Enterprise empowers Mars teams to automate complex tasks across its Petcare, Snacking, and Food & Nutrition segments.
Merck is deploying a Gemini Enterprise-powered “agentic engine” to enhance decision-making across its value chain—from early discovery, through clinical development and commercialization—to improve both the probability of success and time-to-market.
NASA partnered with Google Cloud to use Gemini Enterprise agents to power flight readiness and ensure astronaut safety for Artemis II, which set the human spaceflight record for farthest distance from Earth.
PepsiCo is using Gemini Enterprise Agent Platform to help their employees move from insight to action more quickly and build new digital capabilities across its global operations.
Signal Iduna is using Gemini Enterprise to deliver insurance answers to its customers 37% faster.
SQUARE ENIX uses Agent Platform to power Gemini-based companions into Dragon Quest X Online that provide real-time gameplay support and interactive storytelling.
The American Society of Clinical Oncology is using Gemini Enterprise to deliver trusted cancer expertise to 50,000 oncology professionals worldwide.
The Home Depot is building on its 10+ year partnership with Google Cloud and using Gemini Enterprise to power a suite of AI tools – from assistants like Magic Apron to new AI voice agents for customer service – delivering 'Orange Apron' expertise to shoppers online, on the phone and in the store for a truly seamless home improvement journey.
Unilever is deploying agents built on Gemini Enterprise Agent Platform throughout its organization to supercharge demand gen, build stronger partnerships, and serve its 3.7 billion consumers.
Virgin Voyages is using Gemini Enterprise to create and manage more than 1,000 specialized AI agents, including more than 50 to reduce campaign creation times by 40%. And Gemini Enterprise powers the company’s personal concierge, helping its crew members delight customers at every step of their journeys.
Vodafone is deploying Gemini Enterprise agents to proactively resolve outages and optimize infrastructure, scaling network reliability and saving millions every year.
WPP is using Gemini Enterprise to reshape creative and production, releasing an AI-led campaign every four days, at twice the speed, with 2.5x more value for clients due to its ability to quickly test assets in market and revise creative. WPP has built thousands of agents so far on Gemini.
Small and midsize businesses are seeing benefits as well—Thrive Restaurant Group is enabling its employees to explore data conversationally; the law firm Abogada Alexandra has created new ways to process contracts and prepare important case briefs more quickly; medical technology manufacturer PFM Medical has built a robust enterprise search tool that helps its teams quickly find data stored across systems, like SharePoint, Jira, and SAP, to improve decisionmaking.

The foundation for the Agentic Enterprise: AI Hypercomputer

Designed specifically for demanding AI workloads, our AI Hypercomputer is an advanced, purpose-built architecture that unites performance-optimized hardware for compute, storage, networking, open software and ML frameworks—as well as flexible consumption models—

into a single, integrated system. Our ability to co-evolve all of these components translates to leading power and performance efficiency for our customers.

At the heart of this system is the industry’s widest variety of compute options, spanning our custom Google TPUs and Axion CPUs, to the latest NVIDIA GPUs, giving customers the ultimate flexibility to pick the right architecture for the job without compromise.

Today we are announcing innovations at every layer of the AI Hypercomputer:

We offer the widest array of compute processors to customers, and our portfolio is expanding today to include:

TPU 8t, optimized for training, uses breakthrough Inter-Chip Interconnect (ICI) technology to scale up to 9,600 TPUs and 2 PB of shared, high-bandwidth memory in a single superpod. It achieves 3x the processing power of Ironwood and delivers up to 2x more performance/Watt.
TPU 8i, optimized for inference, uses our new Boardfly topology to directly connect 1,152 TPUs in a single pod. It features 3x more on-chip SRAM compared to previous versions to host larger KV caches entirely on-silicon and integrates a specialized Collectives Acceleration Engine. Taken together, TPU 8i delivers 80% better performance per dollar for inference than the prior generation, enabling millions of concurrent agents to run cost-effectively.
NVIDIA GPUs are a core part of our AI accelerator portfolio. We will be among the first to offer NVIDIA Vera Rubin NVL72, in addition to the Blackwell- and Hopper-based instances already available.
Google Cloud Axion includes our N4A Axion instances—launched in January—which deliver 100% better price-performance than comparable x86 instances, ensuring sustained, efficient operation for agents.
Network-optimized Compute is expanding with new C4N and M4N machine series, perfectly optimized for high-volume agent communication, network-heavy Telco 5G core, and enterprise databases. C4N instances can deliver an almost 4x increase in network bandwidth-per-vCPU over standard C4 instances.

New innovations in Storage include improvements in Managed Lustre, which now delivers 10 TB per second of throughput to A5X or TPU 8t over RDMA, supercharging training. In Rapid Storage, we’ve made significant performance improvements—going from 6 TB/sec to 15 TB/sec—to speed up AI training and inference workloads. And we are introducing Smart Storage, which applies semantic meaning to unstructured data, providing the foundation for our Enterprise Knowledge Graph.
New capabilities in Networking include the Virgo Network, our new, purpose-built, AI-optimized network that connects either NVIDIA Vera Rubin NVL72 systems or TPU 8t superpods into massive supercomputers with hundreds of thousands of accelerators, supercharging distributed training of the world's most capable frontier models.
We are also announcing new features in our Compute Services, including the expansion of our OSS Frameworks and Inference Engines to include native support for Pytorch for TPUs, as well as optimized vLLM support across GPUs and TPUs. And Google Kubernetes Engine (GKE) has new capabilities that deliver dramatically faster cold starts and scale-out for AI inference workloads, and new agent sandboxes that are capable of deploying 300 sandboxes per second, per cluster, with sub-second time to first instruction.

In this agentic era, you can’t have humans manually filing tickets to manage capacity or troubleshoot configurations; you need a cloud that drives itself. We’ve used the Model Context Protocol (MCP) to turn every Google Cloud service into a tool that agents can orchestrate directly, enabling them to troubleshoot our own infrastructure using decades of our own telemetry. And by integrating Gemini reasoning into our telemetry libraries, the system now performs autonomous root-cause analysis—identifying misconfigurations and remediations to fix before a human even realizes there’s a problem.

And of course, Google uses this infrastructure to power our own models, including Gemini and our portfolio of media generation models—driving new levels of efficiency, creativity, and customer engagement. By designing the model and the machine in tandem, we eliminate the constraints of fragmented stacks to deliver extreme scale and efficiency, as well as leading price-performance. In addition, this ensures that breakthroughs achieved through our research are immediately ready for the rigors of production for our customers.

Our industry leading AI Infrastructure serves a number of customers:

Leading AI labs like Thinking Machine Labs use our NVIDIA-based GPU infrastructure and are seeing >2x faster training times on its frontier models.
Capital markets firm Citadel Securities built a scalable, cloud-based quantitative research environment that uses Cloud TPUs to run workloads faster and at a lower cost.
Enterprises like Boston Dynamics are training vision-language models to safely deploy robots across a wide variety of industrial applications.
Government agencies, including all seventeen U.S. Department of Energy National Labs use the “AI co-scientist” built on TPUs as a research partner, generating and validating new hypotheses to accelerate scientific discovery.
Utilities like Axia Energy, Brazil’s largest power company, uses TPUs to forecast severe storms so it can proactively plan and execute preventative measures before storms hit—significantly reducing service interruptions year-over-year.

The Agentic Data Cloud: A new data architecture built for the speed and scale of agentic AI

The Agentic Enterprise demands a fundamental shift in data architecture. For a decade, "systems of intelligence" functioned as reactive data archives built for human scale. Today, we are moving to “systems of action” built for proactive, trusted agents that run at agent scale. The Agentic Data Cloud delivers an AI-native architecture—a unified foundation that frees your data from the silos of the last generation—allowing agents to perceive, reason, and act on your behalf in real-time.

New capabilities in the Agentic Data Cloud we are announcing today include:

Cross-Cloud Lakehouse, standardized on Apache Iceberg, is our Lakehouse that enables you to leave your data in AWS or Azure (coming later this year) while querying it instantly—without the friction of vendor lock-in or the cost of data movement. We provide zero-copy access to applications, operating systems, and AI platforms, such as Databricks, Palantir, Salesforce Data360, SAP, ServiceNow, Snowflake, Workday, and more.
Lightning Engine for Apache Spark delivers a real-time, serverless engine that is up to 4.5x faster than open-source alternatives and offers up to 2x better price-performance over the leading competitor for large datasets.
Data Agent Kit delivers a Gemini-powered data science authoring experience across your IDEs, Notebooks, and Agentic Terminals. It enables data practitioners with intent-driven development in Python, Spark, and SQL—simply state your business goal, and the Data Agent Kit handles the rest.
Knowledge Catalog constructs a unified, dynamic context graph of your entire business enabling you to ground agents in all of your business data and semantics. With Smart Storage and the Object Context API, files in Google Cloud Storage are instantly tagged and enriched with metadata before an agent touches them. Then our Knowledge Engine uses Gemini to autonomously tag, define logic and instantly map complex relationships across your entire enterprise, providing the semantic definition your agents have been missing.
Deep Research Agent for enterprise insight is an autonomous agent that combines research and analytical skills. While Deep Research in Gemini Enterprise already works with your documents and SaaS systems to synthesize your business data, it now connects to data platforms like BigQuery, ensuring a full picture of all your structured and unstructured data. Now it can answer complex business questions like “what is the root cause of my supply chain disruption?” or "can you perform a financial audit on this region?” and include citations and precision that previously required weeks of manual effort.

Customers are already using our Agentic Data Cloud today:

Flipkart, Lowe’s, and Meesho are accelerating their Apache Spark workloads with Lightning Engine.
American Express is moving a core on-premise warehouse and hundreds of production applications to BigQuery to enable trusted agentic commerce at scale.
Bayer Crop Science researchers are eliminating manual data management and analytics with data science agents, freeing them to focus on pioneering agricultural innovations for a sustainable, productive future.
Macquarie Bank unified its data on BigQuery and Spanner to protect and personalize banking for customers.
Vodafone built a self-healing diagnostic system to proactively resolve outages expected to save €100 million by 2030.

Protecting the Agentic Enterprise: Security built for the AI era

Companies across industries face a compounding security challenge. As they expand into complex hybrid and multicloud environments, their attack surface is growing. At the same time, organizations are rapidly deploying AI agents—feeding them sensitive, proprietary data to streamline core business operations. Meanwhile, adversaries are weaponizing AI to launch faster, more sophisticated attacks directly targeting these new enterprise models and the critical data powering them.

To combat this, human analysts need AI on their side. We offer an AI-powered cybersecurity platform that combines Google’s Threat Intelligence and Security Operations with Wiz’s Cloud and AI Security Platform to prevent, detect, and respond to threats across all environments. Customers are already benefiting from our Triage and Investigation Agent, which has processed more than 5 million alerts to date, reducing a typical 30-minute manual analysis to 60 seconds.

Today at Next we are announcing a number of new solutions for agentic defense:

Agentic SecOps: Supercharging agentic defense with threat intelligence
- Dark Web Intelligence builds a nuanced profile of your organization using the newest Gemini models, based on Google Threat Intelligence Group’s dark web expertise. Internal tests show it can analyze millions of daily external events with 98% accuracy to help elevate only the threats that truly matter to your organization.
- Threat Hunting Agent uses Google’s massive threat intelligence knowledge to allow teams to proactively hunt for novel attack patterns and adversary behaviors that bypass traditional defenses.
- Detection Engineering Agent generates persistent detection rules for threat scenarios, transforming detection creation from a manual craft into an automated process.
Wiz: Protecting AI and cloud apps across any infrastructure
- Wiz AI Application Protection Platform (AI-APP) secures every layer of AI applications—infrastructure, data, access, models, agents, and applications—from code to runtime, across every environment where you build. AI-APP brings together visibility, risk analysis, and runtime protection into a single, graph-powered platform that helps teams understand and secure AI systems as they operate. Today, we’re taking our commitment further. Wiz now supports Databricks, as well as new agent studios like AWS Agentcore, Gemini Enterprise Agent Builder, Microsoft Azure Copilot Studio, and Salesforce Agentforce so customers gain protection however their teams choose to build. Wiz now also connects to the outer layer of the cloud, including Apigee, Cloudflare, Vercel, and others to add more context on the external attack surface.
- Wiz Red Agent acts like a sophisticated security researcher, but at AI speed and scale. It can reason about application behavior, adapt its approach in real time, and validate exploitable risks across your web applications and APIs, empowering you to stay one step ahead of attackers.
- Wiz Blue Agent gathers evidence across cloud telemetry, runtime signals, and identity context to investigate a threat and assess its potential severity, providing its full investigation logic so you can resolve threats quickly.
- Wiz Green Agent acts as a built-in investigation and remediation engine, continuously analyzing your highest-risk issues. Like a seasoned security engineer, it synthesizes context from across Wiz—including the Security Graph, code-to-cloud relationships, identity ownership, and historical remediation patterns—to identify the true root cause of a risk and the safest, most effective resolution. Teams get environment-specific, step-by-step remediation guidance so fixes are durable.
- Wiz Agentic Workflows introduces a new hub for orchestrating cloud and AI security in Wiz. It brings together the rich context of Wiz, analysis from Wiz AI agents, and a customizable drag-and-drop interface. Together, this enables security teams to define how work gets done, from when and how agents act to where human input is required.

In addition, we are introducing Google Cloud Fraud Defense, a comprehensive platform designed to discern the legitimacy and authorization of bots, humans, and agents. Fraud Defense is the evolution of reCAPTCHA, and provides the intelligence that businesses need to trust their digital interactions and commerce.

Customers across industries and geographies are using Google Cloud Security to strengthen their security posture:

CME Group, the world's leading and most diverse derivatives marketplace, depends on a unified security stack of Google SecOps, Security Command Center, and Mandiant to protect against novel threats.
Colgate-Palmolive decreased external exposure issues by 44% and now sustains long periods of zero critical risks with Wiz.
Deloitte increased analyst efficiency by more than 60% using Gemini in Security Operations. The firm accelerated threat hunts across billions of logs from hours to seconds—and generating detection rules now takes moments rather than weeks.
Lloyds Banking Group accomplished full SOC transformation in under eight months, moving to a continuous detection and response engineering model, shifting more than 40% of engineering capacity towards proactive threat hunting, enabling human expertise to move up the value chain.
The Los Angeles Department of Water and Power is securing critical infrastructure ahead of the LA28 Games using Google security solutions.
Priceline uses Wiz to ensure all acquisitions comply with its security framework, regardless of cloud environment.
Redis uses Wiz Defend to achieve consistent threat detection, while the Blue Agent helps analysts contextualize alerts without the need for manual investigation.
Shell improved its ability to detect urgent and novel security vulnerabilities with Wiz. What used to take anywhere from three days to two weeks is now managed in near-real-time.
Singapore’s CSIT enables proactive defense against advanced digital threats by translating insights into action with Google Cloud security technologies.
Sutherland contained potential intrusions in under 24 hours and transitioned from reactive defense to proactive threat hunting using Google Cloud security.
University of California Riverside cut incident response times by 90% with Google Cloud security–slashing detection-to-resolution from more than 20 minutes to under two minutes.

Most enterprises run on a combination of clouds or hybrid environments, and we are committed to protecting our customers across all of their environments. Bringing Wiz’s cloud context together with Google’s threat intelligence, we are able to uniquely give you a unified view of every AI asset from code to cloud—ensuring governance without slowing down innovation.

The Agentic Taskforce: Redefining engagement and productivity

At Google Cloud, we are solving the two challenges facing the modern enterprise: First, delivering a frictionless journey for your customers; and second, helping your employees be more effective. Gemini Enterprise not only offers a unified system of agents that can orchestrate and automate entire business processes, but also it seamlessly integrates with all of your existing apps and investments.

Earlier this year at NRF, we launched Gemini Enterprise for Customer Experience, a platform for retailers that brings together fragmented search, commerce and service interactions into one personalized, intelligent customer journey. It includes:

Shopping and Food Ordering agents that handle everything—from discovery to checkout—entirely through natural language. With the Food Ordering Agent, Papa John’s is building a hyper-personalized system that remembers customer preferences to get dinner to your door faster.
Customer Support agents that provide a frictionless journey for customers—from search to service—turning every interaction into a revenue opportunity.
Conversational Insights delivers an analytics platform that identifies call drivers and performs automated quality evaluation on 100% of interactions.
Agent Assist augments human agents in real-time by providing contextual knowledge, smart replies, and automated call summarization. Best Buy is guiding shoppers through complex tech specs, issue resolution, and appointment scheduling—autonomously.
Omnichannel Gateway provides a single entry point for WhatsApp, SMS, RCS, and Voice with consistent cross-channel history.
Universal Consumer Context tracks interactions across channels and turns every consumer interaction into a continuation of a single, long running conversation.
Human-Like Voice capabilities Low-latency, audio stream-in and stream-out preserves human tone and multilingual switching for highly natural voice interactions.

In India, retail giant Reliance is transforming retail for millions of its customers with a Shopping Agent. Customers simply type “Plan a birthday party,” and the agent guides their journeys and assembles their carts, driving greater revenue per session. Gemini-powered catalog enrichment supports this, analyzing millions of images in minutes instead of months.

Further empowering your workforce with AI is Google Workspace—including popular apps like Gmail, Docs, Drive, and Meet. Workspace’s AI features have improved employee collaboration and productivity for more than a decade at companies like Colgate-Palmolive, Compass Real Estate, Korean Airlines, and Natura.

Moving beyond a traditional suite of apps, today we are announcing Workspace Intelligence—a semantic unifying layer that breaks down information and context silos for you and your agents. By leveraging the deep semantic context of your digital workflows that span across meeting notes, emails, files, and more, it creates an intelligence layer grounded in your unique context that can fundamentally change the way you work.

Today we are showcasing how Workspace Intelligence extends across your apps, bringing agentic power to every surface:

AI Inbox and AI Overviews in Gmail uses Gemini to create a personal, proactive inbox assistant.
Ask Gemini in Google Chat allows you to instantly synthesize information, surface insights, and query projects from across Workspace directly from your Chat window. It also lets you take immediate action—such as scheduling a meeting on your calendar or creating a Google Doc to develop a pre-meeting brief—turning your conversations into momentum without having to switch tabs.
Reimagined content creation in Docs, Sheets, and Slides instantly synthesizes information from Drive, Gmail, and the web to transform ideas into professionally formatted drafts in Docs that mimic your exact voice and style. In Docs, Gemini now acts as a collaborative partner that can generate fully formatted first drafts by synthesizing your specific emails, chats, files, and the web. In Sheets, you can transform raw data into interactive dashboards, or instantly generate custom project trackers, plans, and organizational structures, all from a simple prompt. And in Slides, your ideas come to life with an agentic builder that creates and edits full, on-brand presentation decks with a single prompt.
Google Drive Projects is a new, intelligent space that instantly organizes your team's files and emails to manage workflows, generate content, and deliver specific answers based on rich project context. In addition to newly added AI Overviews and Ask Gemini, Projects is another way we’re transforming Drive from a storage tool into an active collaborator to provide insights about your data.
Workspace agent in Gemini Enterprise can execute complex, multi-step tasks across Google Workspace apps without having to leave Gemini Enterprise.

To help organizations immediately unlock this new era of productivity, we are also introducing Rapid Enterprise Migration. Now migrating your entire organization—including complex legal and finance teams—from Microsoft 365 to Google Workspace is up to five times faster.

Our commitment to openness and our ecosystem

At Google Cloud, we understand that openness is a requirement for our customers, and we remain committed to choice at every layer of the stack. Not only does this allow you to ensure your new investments in AI are interoperable with existing investments, but it also gives you the freedom to choose what is best for your business—whether it comes from us or one of our many partners. With optionality at every layer, you can choose from a wide selection of models, AI accelerators, and open ML software, as well as agents and applications from ISVs, and implementation services from leading systems integrators.

At Next, we’re highlighting new ways that our partners can help you access Google Cloud AI across more of your applications and interfaces, including:

Gemini momentum with ISVs: Leading companies like Salesforce, SAP, ServiceNow, Workday, and hundreds more are deeply integrating Gemini into their products, bringing more AI-powered features and agents to their users.
More enterprise-grade agents from partners: Already, thousands of agents built by our partners are available in Google Cloud Marketplace. Now, you can access select agents directly through Gemini Enterprise, making it easier for you to find the right agent for your team’s needs from partners like Accenture, Adobe, Atlassian, Deloitte, Lovable, Oracle, Palo Alto Networks, Replit, S&P Global, Salesforce, ServiceNow and Workday.
Additional expertise and forward deployed engineers (FDEs) through systems integrators: Today, every major systems integrator offers customers access to highly skilled teams of consultants and AI implementation experts—together, totalling more than 330,000 consultants trained on Google Cloud AI technologies around the world. We are now sending teams of Google Cloud forward-deployed engineers to partner closely with teams from Accenture, Capgemini, Cognizant, Deloitte, HCLTech, PwC and TCS, to accelerate customers’ agentic journeys. We’ll also begin rolling out a new early model access program for a select group of partners, including Accenture, BCG, Deloitte and McKinsey, who will be able to preview and begin building with pre-release versions of upcoming Google Deepmind models.

And for our customers in regulated industries like financial services and healthcare, In fact, at Next, we are announcing support for the latest Gemini 3 models in both air-gapped and connected environments.

Customer success across every industry

As the cloud market evolves in the agentic era, we are seeing the transformative impact our products are having with customers across every industry. At Next this year, we will be sharing the stories of more than 500 organizations from around the world using our AI and agentic tools, including:

Financial Services: Citadel Securities, Citi Wealth, CME Group, DBS, Macquarie Bank, Revolut, Starling Bank
Retail, Restaurants & Consumer Goods: Alpura, Authentic Brands Group, Cia Hering, Colgate, Fairprice, Kingfisher, Kroger, Lowe’s, Macy’s, Mars, Papa Johns, PepsiCo, The Home Depot, Ulta, Unilever, Woolworths
Media & Entertainment and Games: 10six Games, Antstream Arcade, Avid, CANAL+, Dreamlands, ElevenLabs, Formula E, MLB,Parallel, Quick Play and Gray Media
Telecommunications: Amdocs, Deutsche Telekom, Liberty Global, Nokia, Vodafone
Travel & Hospitality: Latam Airlines, Virgin Voyages
Security & Tech: Atlassian, Bending Spoons, TCS
Healthcare & Life Sciences: Merck, CVS Health, Humana, Quest Diagnostics, Waystar
Manufacturing: Bosch, EcoVadis, GE Appliances, Tata Steel
Governments:, German Armed Forces, Maryland, NATO, UK Department for Transport, U.S. Department of Energy, U.S. Department of Transportation, U.S. Food and Drug Administration, SENAI Brazil
ISVs: Lovable, Oracle, Salesforce, SAP, ServiceNow , UKG, Workday
Systems Integrators: Accenture, BCG, Capgemini, Cognizant, Deloitte, HCLTech, IBM, Infosys, Kyndryl, KPMG, McKinsey, PwC, TCS
Startups: Emergent AI, Jusbrasil, Notion, Parallel Web Systems, Reve, Safe Superintelligence, Thinking Machines Lab, Unico
SMBs: Apex Leaders, B3Networks, Constructora Las Galias, OneUnited Bank, Papel Semente, Premji Invest, Thrive Restaurant Group
And many, many more showcased on our website.

Your Agentic Enterprise

Transformation to an Agentic Enterprise is the future of every organization, and Google Cloud uniquely offers a unified system that can take you there. Our AI Hypercomputer provides scalable foundation, our Agentic Data Cloud provides the context, our Agentic Defense provides the protection, our agents deliver new levels of engagement and productivity, and our Agent Platform provides the orchestration.

And while technology can provide transformational power, you must still provide the vision: With all this innovation at your fingertips, what will your Agentic Enterprise build?

Cross-cloud infrastructure innovation for the agentic enterprise

Wed, 22 Apr 2026 12:00:00 +0000

The era of agentic AI is accelerating from human- to machine-speed operations, while also creating profound stress on legacy technology infrastructure. This new reality pushes foundational systems to their limits: agents generate thousands of internal messages and complex queries, spawning more agents, all of which can rapidly overwhelm traditional networks and databases, and expose new security vulnerabilities.

Unlocking AI's full potential in the era of agents requires a secure, adaptive foundation. We call it cross-cloud infrastructure for the agentic enterprise – and at Google Cloud Next ‘26, we’re launching a powerful set of new innovations across four areas:

What’s new:

Fluid compute: Google Compute Engine and Kubernetes services work together to enable cost-effective, high-speed AI agents and enterprise workloads with new compute and orchestration capabilities.
Secure cross-cloud connectivity: Agent Gateway, Cloud Armor, and other tools deliver a secure, governed, and simplified networking foundation for AI agents, including observability of agentic traffic across clouds.
Unified data layer: Smart Storage, Knowledge Catalog, and other innovations transform passive data archives into dynamic reasoning engines, giving AI agents the context they need to execute.
Digital sovereignty: Confidential External Key Management and new features in Google Distributed Cloud bring Google’s leading models and AI enablers wherever your data lives.

Let’s take a closer look at all the news for each of these four areas.

Fluid compute

Agentic workloads are dynamic and unpredictable, impacting both traditional enterprise applications and the AI agents themselves. Fluid compute is enabled by Google Compute Engine and Google Kubernetes services working together to dynamically adapt and shift weight in real-time, enabling cost-effective, high-speed AI agents and operational enterprise workloads for all customers.

While our AI Hypercomputer delivers raw power for large-scale AI model training, fluid compute addresses the needs of operational workloads and agents. As agents move toward reasoning and reinforcement learning, CPUs are reclaiming a central role, excelling at the "branchy" logic, complex control flows, and secure execution sandboxes (like those for agentic orchestration, RL, SLM inference, and RAG) that agent workflows demand. CPUs also provide the critical isolation needed for secure agent execution, complementing the parallel processing strength of GPUs and TPUs used in training.

We are introducing new CPU families, GKE capabilities, and Hyperdisk block storage capabilities to run traditional workloads and AI agents securely at scale, including:

Google C4N Series: These VMs help ensure your enterprise workloads don't slow down under the demands of agentic AI by processing up to 95 million packets per second, up to 40% faster than other leading hyperscalers. This eliminates I/O bottlenecks for demanding workloads like security appliances, streaming media, and open source databases, even when utilizing smaller instance sizes.
Google M4N Series with Hyperdisk Extreme: M4N removes data pipeline bottlenecks and eliminates overprovisioning to deliver industry leading per-core IOPS and throughput required to handle massive data I/O from agents, analytics, and mission-critical databases. M4N provides 26.57 GB of RAM per vCPU, allowing you to scale mission-critical workloads cost-effectively on fewer cores. For example, M4N with Hyperdisk Extreme reduces Oracle workload total cost of ownership by over 20% compared to leading hyperscale clouds.
GKE Agent Sandbox: This solution secures agents with trusted gVisor isolation and handles demand spikes, launching up to 300 sandboxes per second, per cluster. Backed by the only managed sandbox technology available among leading hyperscale clouds, it achieves up to 30% better price-performance than competitors when running AI agents on GKE Agent Sandbox with Google Axion N4A.

“Wayfair's AI strategy is built on years of systematic infrastructure modernization on Google Cloud — migrating our core eCommerce engine and databases off legacy systems, decomposing monolithic services into cloud-native architecture, and unifying our data and analytics platform. That foundation is what makes everything else possible. Today, Gemini Enterprise Agent Platform is powering everything from catalog enrichment to generative shopping experiences that help customers create a home that's just right for them — and it's the same foundation preparing us for the agentic era, where AI doesn't just assist but actively drives discovery, personalization, and commerce across every customer touchpoint and across our business.” - Fiona Tan, Chief Technology Officer, Wayfair

Explore all our latest compute innovations in this blog.

Secure cross-cloud connectivity

Agentic AI replaces predictable human requests with autonomous “reasoning loops,” in which agents call other agents that, in turn, call LLMs, triggering massive, sudden surges in compute and machine-to-machine traffic. This shift creates unique challenges for network predictability and security of non-human identities. Optimized for agentic AI, our Cross-Cloud Network moves data across diverse environments, connecting employees, customers, and agents with visibility and security. New in Cross-Cloud Network are:

Agent Gateway: Governs and orchestrates your enterprise agentic traffic as the “air traffic controller” in Gemini Enterprise Agent Platform. It natively understands agent protocols like MCP and A2A to inspect and govern every agent interaction. By integrating with Google and third-party identity and AI safety services, it enables deep inspection to verify access, block attacks, and protect sensitive data, maintaining compliance across your core business.
Cloud Network Insights: Delivers broad visibility across your hybrid and multi-cloud infrastructure to drive faster troubleshooting and network resolutions. Continuously monitor your end-to-end agent, network and web performance across Google Cloud, AWS, Azure, data centers, internet applications, and agentic workloads. Using synthetic traffic analytics, Cloud Network Insights provides hop-by-hop network path visibility to help you pinpoint the source of degradations and is coupled with AI-powered insights from Gemini Cloud Assist to deliver more autonomous operations.
Enhanced Cloud Next Generation Firewall (NGFW) and Cloud Armor: Provides machine-speed, AI-powered protection to combat the rapid explosion of AI-generated polymorphic malware and zero-day exploits. Cloud NGFW advanced malware sandbox delivers real-time inline prevention of AI-generated threats, while Cloud Armor managed rules provides automated protection against both known and unknown Common Vulnerabilities and Exposures (CVEs). Together with Model Armor, these services analyze the intent and content of AI agent communications.

Discover more about how we optimized networking for AI in and outside of the data center.

Unified data layer

AI agents are only as powerful as the data they can access and the context they’re given. More applications and platforms are using structured and unstructured data, but it can be difficult to catalog, find, and act on that data at scale, leading to less effective agent interactions. To close the gap, your agents need all of your data brought together into a cohesive, queryable knowledge engine, or unified data layer. This way, your agents can identify and access accurate sources. At Next ‘26, we’re enhancing the unified data layer with:

Smart Storage: This solution transforms dark data into a powerful knowledge asset for AI agents and training by embedding new semantic intelligence directly into your data objects. With new Google Cloud Storage capabilities like automated annotation, entity extraction, and semantic search, your agents can instantly find and use the specific data they need — whether it's hidden in spreadsheets, PDFs, or other unstructured formats across your entire organization. This significantly speeds up the development and deployment of your AI solutions. Learn more about storage innovations to accelerate your AI workloads.
Knowledge Catalog: Knowledge Catalog maps business meaning across your entire data estate, providing a grounded source of truth so agents can deliver the most accurate results. This foundation enables AI training and inferencing and doesn’t require you to migrate your data; your agents interact with it directly, wherever it lives, with full context and governance, making modernization easier.

Part of our Agentic Data Cloud, Smart Storage and Knowledge Catalog can take your data from a passive archive into a dynamic reasoning engine.

“AI is critical to making our customers’ smart home and security solutions more intelligent and convenient. By leveraging Google Cloud’s Smart Storage, we auto-annotate rich metadata delivered in BigQuery. We’ve scaled and accelerated our data discovery and curation efforts, speeding up our AI development process from months to weeks, continuously delivering innovations that build trust and enhance the overall home experience.” - Brandon Bunker, VP of Product, AI, Vivint

Digital sovereignty

In the agentic era, digital sovereignty is a fundamental requirement for public sector and enterprise customers looking to accelerate innovation — without sacrificing control. There’s no one-size-fits-all solution, which is why we’ve designed a comprehensive set of offerings to meet different sovereign AI needs anywhere: public cloud, on-premises, or hybrid. New capabilities in our sovereign AI portfolio include:

Confidential External Key Management: Organizations can use Confidential External Key Management to maintain complete possession, custody, and control of your encryption keys and the policies that govern them. Confidential External Key Management leverages Confidential Compute to host the key management endpoint in a tamper-proof environment within Google Cloud. You are in control and determine where your keys are stored, who can access them, and under what circumstances. Even highly privileged Google administrators cannot access your keys without authorization, which you can revoke at any time. Your data, your control.
Gemini on Google Distributed Cloud: With Gemini on GDC, companies can securely deploy Gemini in sensitive environments, while meeting data sovereignty needs. Your choice of deployment models includes managed software on your connected hardware or a fully disconnected, air-gapped solution. You can now scale with Google's leading AI capabilities even in the most restricted, high-security environments — from powerful Gemini models to advanced coding, search, and other agentic capabilities.

In addition, Google Distributed Cloud supports an end-to-end AI stack, combining our latest-generation AI infrastructure with Gemini models to accelerate and enhance all your sovereign AI workloads. This stack includes:

NVIDIA Blackwell GPUs: NVIDIA Blackwell (NVIDIA HGX B200) and NVIDIA Blackwell Ultra platforms (NVIDIA HGX B300) GPUs accelerate AI performance, leveraging fifth-gen NVIDIA NVLink to deliver data-center scale bandwidth directly to your environment.
New VM families: New A4 family offerings provide the ability to handle the most demanding inference tasks, delivering a 2.25x increase in peak compute. Memory-Optimized M2 and M3 brings the high memory-to-vCPU ratios needed for massive ERP and data analytics workloads on-premises.
Enhanced storage: Eliminate storage bottlenecks with 6x storage capacity per zone and a 10x performance boost, giving you the ability to do AI reasoning on-premises. Now, your data infrastructure moves at the speed of AI reasoning.

"Our customers demand high-performance, private AI inference without the risks of multi-tenancy. Google Distributed Cloud allows us to provide dedicated, low-latency environments that meet strict sensitive data requirements. With the ability to run Gemini on B200s and B300s, we can significantly increase inference speeds and provide the token throughput our clients need to scale." - Dave Driggers, CEO & Co-founder, Cirrascale Cloud Services

Transforming vision into reality

When these product areas converge, your infrastructure evolves into a high-performing, secure, adaptive foundation for the agentic era. We're not just offering tools; we're providing the architectural blueprint to enable enterprises and the public sector to rapidly embrace the full power of AI and agents with confidence.

To learn more about key industry trends for AI Infrastructure, read our State of Infrastructure in the Agentic AI Era report.