Google unveils Gemini Enterprise Agent Platform and new TPUs as AI token volumes surge

Google used its Cloud Next ‘26 event on April 22, 2026, to put a sharper commercial frame around its AI strategy, unveiling Gemini Enterprise Agent Platform and new eighth-generation Tensor Processing Units as it reported accelerating use of its AI products across the cloud business.

Google’s pitch shifts from chatbots to enterprise agents

The company said nearly 75% of Google Cloud customers are now using its AI products, and that 330 customers processed more than a trillion tokens each over the past 12 months. Google also said its models are now processing more than 16 billion tokens per minute via direct API use, up from 10 billion last quarter.

That scale matters because Google is no longer presenting AI as a feature layered onto cloud services. Instead, it is positioning agentic systems as the next operating model for enterprise software, with Gemini Enterprise Agent Platform intended to help businesses build, scale, govern and optimize autonomous agents.

New TPUs are built for heavier training and inference loads

Alongside the software push, Google announced its eighth-generation TPUs, describing them as two specialized chips designed for the agentic era. The company said the hardware will support both training and inference, signaling that the infrastructure buildout is being matched to the rising demand created by larger models and more frequent API-driven usage.

For customers, the practical implication is straightforward: the AI stack is becoming more industrialized. The competitive question is moving away from whether a company can add generative AI at all, and toward whether it can run those workloads reliably, at scale and with enough governance to bring agents into production systems.

Cloud Next shows Google leaning into commercialization

The announcements also show how aggressively Google is tying AI progress to monetization. By highlighting customer adoption, token throughput and new dedicated chips in the same launch cycle, the company is emphasizing that its AI business is not just a research story or a consumer-product story, but an infrastructure and enterprise execution story.

Google said the new platform and hardware are part of the roadmap for helping organizations move into what it calls an agentic enterprise. The immediate significance is that the company is now selling the tools, compute and operational controls it expects enterprises to need as AI systems take on more of the work inside existing workflows.