Google Cloud presents the enterprise platform that unifies the creation, scaling and orchestration of agents

Google Cloud has presented the new platform with which it intends to support its clients’ artificial intelligence (AI) projects, which enables the creation, scaling and orchestration of agents, along with developments in data management, security against new threats and productivity in the era of agentic companies.

75 percent of Google Cloud customers already use its AI products to power their businesses. In the last year alone, the company has delivered the seventh generation of its Ironwood TPUs and Axion processors, in addition to launching the Gemini 3.1 Pro and Nano Banana 2 (Gemini 3.1 Flash Image) models.

This Wednesday, the company announced its latest advances in corporate AI within the framework of its annual Google Cloud Next’26 event, with which it aims to help its clients become agentic companiesas the natural step in AI with the creation and deployment of agents.

To this end, it has announced the new Gemini Enterprise Agent platform, an evolution of the Gemini Enterprise platform which offers a unified solution to create, scale and orchestrate AI agents.

This new platform includes Vertex AI, features such as Agent Studio, Agent-to-Agent Orchestration, Agent Registry, Agent Identity, Agent Gateway and Agent Observability for agent integration and security.

In it, companies will be able to access their own models such as Gemini 3.1 Pro, Nano Banana 2 and Lyria 3, and others from third parties, including Claude Opus, Sonnet and Haiku.

Also includes the Gemini Enterprise application for bring agents into the company’s workflow and to employees. It has been updated with new features, which help you execute complex tasks, monitor agent activity, add shortcuts, and more.

THE INFRASTRUCTURE THAT DRIVES AI

Google has presented the latest developments in what it calls ‘AI hypercomputing’, the infrastructure that allows the most demanding artificial intelligence workloads to run and scale. These include the new eighth generation TPUsstorage expansions and new features in Compute Services.

The eighth generation of arrives with two models. The first, TPU 8t, aimed at accelerating training, uses Inter-Chip Interconnect (ICI) technology to scale up to 9,600 TPUs and 2PB of shared bandwidth memory in a single superpod.

The second, TPU 8iit has been designed to drive inference with better cost and less latency. Uses Boardfly topology to connect 1,152 YPUs into a single pod.

Google Cloud has also confirmed that they will be among the first to offer the Nvidia Vera Rubin NVL72 artificial intelligence platform.

UPDATE OF AGENTS PLATFORMS

To accelerate and scale AI agents, Google Cloud has announced Agentic Data Cloud, a new native AI architecture that allows agents act with the data and context of the companies without being limited in data silos.

And to protect enterprises’ artificial intelligence and cloud environments from sophisticated new attacks using AI, Agentic Defense, which combines insights from Google Threat Intelligence and the expertise of Google Security Operations partners, including the recent addition of Wiz, to prevent, detect and respond to threats.

These works are reinforced with three new agents that Google has incorporated into Google Security Operations: the threat hunting agent, which proactively searches for new attack patterns and stealthy adversary behaviors; the Detection Engineering Agent, which automates the detection of security breaches; and the third-party context agent, which brings third-party contextual data to workflows.

Likewise, and as part of the Agentic Taskforce, Google has included new capabilities in Gemini Enterprise for Customer Experience and Workspace with the aim of improving productivity. One of the new features is Workspace Intelligencewhich eliminates silos between the worker and agents when it comes to the semantic context of the documents and files they work with within Google productivity applications.