The race for AI supremacy isn't just about algorithms and models; it's fundamentally a race for compute. And Anthropic, a leading force in frontier AI, is making an unprecedented move to secure its position. The company has just announced a monumental expansion of its partnership with Google and Broadcom, locking in 'multiple gigawatts' of next-generation TPU capacity expected to come online starting in 2027.
Fueling Frontier AI: Why This Deal Matters
This isn't merely an incremental upgrade; it's a strategic, long-term play to ensure Anthropic's Claude models can continue to define the frontier of AI development. As Krishna Rao, CFO of Anthropic, stated, "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."
Unprecedented Growth Demands Unprecedented Compute
The need for this immense compute power is evident in Anthropic's staggering growth metrics:
- Revenue Surge: Run-rate revenue has now surpassed an astounding $30 billion, a sharp increase from approximately $9 billion at the end of 2025.
- Customer Expansion: The number of business customers spending over $1 million on an annualized basis has more than doubled in less than two months, skyrocketing from over 500 to exceed 1,000.
These figures underscore the rapid enterprise adoption of Claude and the urgent necessity for Anthropic to scale its infrastructure to meet this accelerating demand.
A Strategic Bet on American AI Infrastructure
A significant portion of this new compute capacity will be sited in the United States. This commitment directly supports Anthropic's November 2025 pledge to invest $50 billion in strengthening American computing infrastructure. It's a move that not only future-proofs Anthropic's operations but also contributes to the nation's strategic AI capabilities.
The partnership deepens existing relationships, building on Anthropic's increased TPU capacity announcement with Google Cloud last October and its ongoing work with Broadcom.
The Power of Platform Diversity
One of Anthropic's key strategic advantages lies in its diverse hardware approach. The company trains and runs Claude on a range of AI hardware, including:
- AWS Trainium
- Google TPUs
- NVIDIA GPUs
This platform diversity allows Anthropic to match specific workloads to the chips best suited for them, leading to enhanced performance and greater resilience for customers. While Amazon remains Anthropic's primary cloud provider and training partner (working closely on Project Rainier), the expanded Google and Broadcom collaboration ensures a robust, multi-faceted compute strategy.
Furthermore, Claude remains unique in its availability across all three of the world's largest cloud platforms:
- Amazon Web Services (Bedrock)
- Google Cloud (Vertex AI)
- Microsoft Azure (Foundry)
This broad accessibility offers unparalleled flexibility and choice for developers and enterprises integrating frontier AI into their applications.
Looking Ahead
This compute expansion is a clear signal of Anthropic's aggressive posture in the AI landscape. By securing 'multiple gigawatts' of cutting-edge TPU capacity, Anthropic is not just responding to current demand but proactively building the foundation for the next generation of AI breakthroughs. For developers and businesses, this means continued access to more powerful, reliable, and accessible Claude models capable of tackling increasingly complex challenges. The compute race is indeed heating up, and Anthropic is clearly in it to win.