Anthropic has signed a new agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity, with the first systems expected to come online in 2027. The company disclosed the deal alongside fresh financials: run-rate revenue has surpassed $30 billion, up from roughly $9 billion at the end of 2025. Anthropic says the buildout is its largest compute commitment to date and is designed to keep pace with what CFO Krishna Rao called "unprecedented growth."
The customer numbers moved nearly as fast as the revenue line. When Anthropic announced its Series G in February, it counted more than 500 business customers each spending over $1 million on an annualized basis. That figure now exceeds 1,000 — a doubling in under two months.
The vast majority of the new compute will be sited in the United States, making the Google and Broadcom deal a direct extension of the $50 billion commitment Anthropic made in November 2025 to invest in American computing infrastructure. It also builds on the expanded Google Cloud TPU capacity the two companies announced last October.
Key facts
- 01Anthropic signed a new deal with Google and Broadcom for multiple gigawatts of next-generation TPU capacity, starting in 2027.
- 02Run-rate revenue has surpassed $30 billion, up from approximately $9 billion at the end of 2025.
- 03Customers spending $1M+ annualized on Claude now exceed 1,000, doubling from 500+ in February.
- 04The buildout extends Anthropic's November 2025 pledge to invest $50 billion in US computing infrastructure.
- 05Claude remains the only frontier model on AWS Bedrock, Google Vertex AI, and Microsoft Azure Foundry.
"This groundbreaking partnership with Google and Broadcom is a continuation of our disciplined approach to scaling infrastructure: we are building the capacity necessary to serve the exponential growth we have seen in our customer base while also enabling Claude to define the frontier of AI development," Rao said in the announcement.
“Anthropic's run-rate revenue has crossed $30 billion, up from roughly $9 billion at the end of 2025, with $1M+ customers doubling past 1,000 in under two months.”— Jaeden Schafer
The hardware strategy is deliberately plural. Anthropic trains and serves Claude across AWS Trainium, Google TPUs, and NVIDIA GPUs, matching workloads to the silicon best suited for them. The company frames that diversity as a resilience argument for enterprise customers, not just a cost play — if one platform has a capacity crunch or a hardware bug, Claude keeps running.
Amazon Web Services remains Anthropic's primary cloud provider and training partner, and the two continue to build out Project Rainier, the AWS-hosted training cluster disclosed last year. The new Google and Broadcom capacity stacks on top of that relationship rather than replacing it, which matters for AWS investors watching how much of Anthropic's growing compute spend flows through Bedrock versus Vertex AI and Azure Foundry.
Claude is still the only frontier AI model available on all three of the largest cloud platforms — AWS Bedrock, Google Cloud Vertex AI, and Microsoft Azure Foundry. That tri-cloud distribution is becoming a commercial moat in its own right, particularly for regulated customers that refuse to standardize on a single hyperscaler.
Rao framed the deal as continuity rather than a pivot. "We are making our most significant compute commitment to date to keep pace with our unprecedented growth," he said. The language tracks with Anthropic's recent cadence of stacking multi-year supply deals — AWS, Google, now Broadcom — rather than betting on one partner.
The obvious caveat is that multi-gigawatt capacity starting in 2027 is a forward commitment, not shipping silicon. Power, siting, and Broadcom's custom-chip production schedule all sit between today's press release and the first watt delivered. Anthropic has also not disclosed the dollar value of the new agreement, which makes it hard to size against OpenAI's and Meta's publicly announced infrastructure deals.
Still, the revenue math is the story. A jump from $9 billion to over $30 billion in run-rate inside a single fiscal year, paired with a doubling of $1M+ customers in under two months, is the kind of curve that forces a company to pre-buy gigawatts years in advance. For the rest of the market, Anthropic's appetite is another data point that frontier AI demand is not cooling — and that the bottleneck is increasingly power and fabs, not model quality.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
