Day 1 at Google Cloud Next ‘26 recap

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 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!

Posted in

• Google Cloud Next
• AI & Machine Learning
• Application Development
• Data Analytics