Anthropic Teams Up with Google and Broadcom for Gigawatts of Next-Gen TPUs!

Hey everyone, shiichan here! Today I've got a really big piece of news about the raw computing power behind AI, and I couldn't wait to share it with you!

Anthropic News anthropic.com

What was announced?

Over on Anthropic's News, Anthropic announced that it is expanding its partnership with Google and Broadcom. The deal covers multiple gigawatts of next-generation TPU capacity - a seriously large-scale commitment. This new infrastructure is expected to come online starting in 2027, and it will power frontier Claude models and help meet demand from customers around the world.

Krishna Rao, Anthropic's CFO, called it a continuation of their disciplined approach to scaling infrastructure, and "our most significant compute commitment to date."

Why it matters

Training and running large models takes an enormous amount of compute. The Claude you and I use every day sits on top of a huge stack of infrastructure.

Demand has accelerated through 2026: run-rate revenue has now passed $30 billion, up from around $9 billion at the end of 2025. Back when Anthropic announced its Series G fundraising in February, more than 500 business customers were each spending over $1 million on an annualized basis. Today that number is over 1,000 - doubling in under two months. To keep up with that growth, they needed a lot more computing power.

What changes

More compute means Claude can reach more people, more reliably. Less of that "I want to use it but I'm stuck waiting" feeling - and that's a good thing.

Most of the new compute will be sited in the United States, expanding on the November 2025 commitment to invest $50 billion in American computing infrastructure.

Dive Deep

This partnership deepens Anthropic's work with Google Cloud, building on the increased TPU capacity announced last October, and it builds on the existing relationship with Broadcom too.

What I find neat is that Anthropic doesn't run Claude on just one kind of chip. It trains and runs Claude across a range of hardware - AWS Trainium, Google TPUs, and NVIDIA GPUs - so it can match each workload to the chip best suited for it. That's not only about performance; it also means more resilience if any one option hits a snag.

Even so, Amazon remains the primary cloud provider and training partner, and the two keep working closely together on Project Rainier. And Claude is still the only frontier AI model available on all three of the world's largest cloud platforms: AWS (Bedrock), Google Cloud (Vertex AI), and Microsoft Azure (Foundry).

Wrap-up

• Anthropic is expanding its partnership with Google and Broadcom for multiple gigawatts of next-generation TPU capacity, coming online starting in 2027
• The driver is surging demand: run-rate revenue is over $30 billion, and customers spending $1 million+ now number over 1,000
• Most of the new compute will sit in the US, expanding the $50 billion investment commitment
• Claude runs across AWS Trainium, Google TPUs, and NVIDIA GPUs, and stays available on all three major clouds

This isn't a hands-on, try-it-this-weekend kind of story, but if you like knowing what enormous infrastructure your Claude runs on, or you enjoy tracking the sheer scale of the AI industry, this one's for you!