Anthropic doubles down on compute with Google and Broadcom, revenue hits $30B

Anthropic just announced a new agreement with Google and Broadcom that locks in multiple gigawatts of next-generation TPU capacity, expected to come online starting in 2027. That’s a lot of compute for Claude, and frankly, it’s about time.

The company’s CFO, Krishna Rao, described it as their “most significant compute commitment to date.” He’s not wrong. This isn’t just incremental—it’s a multi-year bet on keeping Claude competitive at the frontier while serving a customer base that’s growing faster than I think even they anticipated.

Let’s talk numbers, because they’re worth pausing on. Anthropic’s run-rate revenue has crossed $30 billion. That’s up from about $9 billion at the end of 2025. In case you’re doing the math, that’s more than 3x growth in roughly a year. And the enterprise side is accelerating hard: in February, they said over 500 business customers were spending over $1 million annually. Today that number is over 1,000—doubled in less than two months. That’s not just growth, that’s a land grab.

Most of the new compute will be sited in the US, which builds on their November 2025 pledge to invest $50 billion in American computing infrastructure. I appreciate the domestic focus, even if it’s partly driven by geopolitical realities around chip supply chains. Still, it’s a concrete move.

This also deepens their existing work with Google Cloud (they already announced increased TPU capacity last October) and Broadcom. But here’s the part I find interesting: Anthropic isn’t putting all its eggs in one basket. They train and run Claude on AWS Trainium, Google TPUs, and NVIDIA GPUs. That’s three different hardware platforms. The stated reason is that they can match workloads to the best-suited chips, which translates to better performance and resilience. I’d also wager it gives them leverage in negotiations and protects against supply constraints from any single vendor.

Amazon remains their primary cloud provider and training partner—Project Rainier is still on—and Claude is still the only frontier AI model available on all three major clouds: AWS Bedrock, Google Vertex AI, and Microsoft Azure Foundry. That’s a nice differentiator, though I suspect Microsoft’s inclusion was a strategic necessity rather than a first choice.

This is a big deal. Not just for Anthropic, but for the broader AI infrastructure race. Everyone is scrambling for compute, and locking in multi-gigawatt TPU capacity years in advance is how you stay in the game. The question is whether the revenue growth can keep pace with the spending. $30B is impressive, but building out this kind of compute isn’t cheap. I’ll be watching their next earnings closely.