Google has updated its chips for AI and opened access to the agent platform - ForkLog: cryptocurrencies, AI, singularity, the future

# Google has updated AI chips and opened access to the agent platform

Google has introduced a set of tools for creating AI agents designed to help companies automate tasks. Simultaneously, a fund of $750 million has been launched, and new AI processors have been announced.

Have a great #GoogleCloudNext!

Watch select sessions and your favorite moments from the opening keynote on demand → https://t.co/Ss9VorZkOvhttps://t.co/VnbGXrIjvi

— Google Cloud (@googlecloud) April 22, 2026

AI Agent Software

At the annual conference in Las Vegas, the company’s cloud division showcased a set of tools for creating AI agents and monitoring their operation within companies. Among the new features is a special mailbox where virtual assistants send reports on completed work.

The Gemini Enterprise Agent Platform will receive two new modules: Memory Bank and Memory Profile. They will allow agents to save interaction histories with users. Agent Simulation will give developers the ability to test tools more thoroughly before launch.

Google announced updates to the Workspace application suite and described a scenario where AI agents radically change the daily life of an ordinary worker.

The company stated that employees will be able to use Gemini Enterprise to create virtual assistants without writing code.

The company also announced Projects — a platform for collaborative work designed for employees to interact with each other or support operators.

The tool consolidates information from various sources such as Workspace, Microsoft OneDrive, and corporate chats, enabling work with the necessary context.

New Fund

Google Cloud’s division announced a fund of $750 million aimed at helping consulting firms like McKinsey, Accenture, and Deloitte implement agent-based AI for clients.

DeepMind’s laboratory will provide select companies early access to Gemini AI models — before their official release.

Funds will be allocated to help train engineers, develop AI agents through the corporate platform, and co-finance projects and pre-sales activities.

“Consulting firms are at the center of some of the biggest transformations happening with clients. They understand the situation and bring unique expertise in specific industries and business processes,” — said Kevin Ichpuran, head of Google Cloud’s global partner ecosystem.

New Chips

Google Cloud’s division introduced a new generation of custom-designed tensor processors (TPU). These chips are intended to accelerate and reduce the cost of AI computations.

The lineup includes two versions:

• TPU 8t — designed for AI development;
• TPU 8i — better suited for inference.

Google holds strong positions among manufacturers of proprietary AI chips, competing with Nvidia. In recent months, TPUs have seen increased demand in Silicon Valley.

The new processors store more information, reducing response latency.

“It’s about how to ensure the minimal possible response delay at the lowest possible cost per operation,” — said Mark Lomeier, vice president of Google’s computing infrastructure.

Systems capable of quickly processing large data sets—finding connections and patterns that are then expressed mathematically—are used to create and deploy AI services. Computations, program launches, and services run on processors with large amounts of embedded memory.

This approach allows AI to respond almost instantly, as it does not need to fetch data from external sources.

TPU 8t can be clustered into groups of 9,600 units. At such scales, power consumption becomes a key factor. The new chips provide 124% higher performance per watt compared to the previous generation. TPU 8i — 117%.

Recall that Google is discussing with the U.S. Department of Defense the possibility of integrating Gemini into Pentagon systems across all levels of information access — from open to highly classified.