Anthropic has entered into a massive infrastructure agreement with Google and Broadcom to secure next-generation compute capacity, a move designed to sustain the explosive growth of its Claude AI models. The partnership focuses on expanding access to Tensor Processing Units (TPUs), the specialized AI accelerators developed by Google, to ensure the startup can meet surging global demand.
The deal represents a significant scaling of Anthropic’s technical foundation. According to a company announcement on April 6, 2026, the new agreement provides for multiple gigawatts of next-generation TPU capacity, which is expected to come online starting in 2027 Anthropic News. This expansion is a critical component of Anthropic’s strategy to maintain its position at the frontier of artificial intelligence development.
The scale of the computing power involved is immense. Reports indicate that the deal gives the AI startup access to approximately 3.5 gigawatts of computing capacity drawing on Google’s AI processors CNBC. Under the terms of the broader arrangement, Broadcom has also agreed to produce future versions of Google’s artificial intelligence chips, acting as a key hardware partner in the supply chain.
Scaling for Explosive Revenue Growth
The rush to secure more compute is driven by a dramatic surge in Anthropic’s financial performance and user base. The company revealed that its run-rate revenue has now surpassed $30 billion, a staggering increase from approximately $9 billion at the end of 2025 Anthropic News. This growth reflects the rapid integration of Claude into enterprise workflows and consumer applications.
Enterprise adoption has been particularly aggressive. In February, Anthropic noted that over 500 business customers were spending more than $1 million on an annualized basis. As of early April, that number has more than doubled, now exceeding 1,000 business clients spending over $1 million annually CNBC.
This commercial momentum was further bolstered in February when the Claude app became the top free U.S. App in Apple’s App Store, a spike in popularity that coincided with a public dispute between Anthropic and the Pentagon CNBC.
A Diversified Hardware Strategy
While the new deal centers on Google’s TPUs, Anthropic is pursuing a “hardware-agnostic” approach to avoid vendor lock-in and maximize resilience. The company currently trains and runs Claude across a diverse range of AI hardware, including Google TPUs, NVIDIA GPUs, and AWS Trainium Anthropic News.
This diversity allows Anthropic to match specific AI workloads to the chips best suited for them, which the company says translates to better overall performance and greater reliability for customers relying on Claude for critical business operations. The financial commitment to this infrastructure is vast; some reports suggest Anthropic will spend hundreds of billions of dollars on Google’s chips and cloud services over the long term Financial Times.
“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,” said Krishna Rao, CFO of Anthropic.
Investment in American Computing Infrastructure
A significant portion of the new compute capacity will be sited within the United States. This move is part of a larger strategic effort to strengthen domestic AI capabilities. The current partnership serves as a major expansion of a commitment made in November 2025 to invest $50 billion in strengthening American computing infrastructure Anthropic News.
By anchoring its infrastructure in the U.S., Anthropic aims to secure its supply chain and ensure that the development of its frontier models remains supported by robust, local hardware ecosystems. This approach aligns with broader industry trends where AI labs are increasingly concerned with the physical location and security of the compute clusters powering their models.
Key Takeaways of the Anthropic-Google-Broadcom Deal
- Compute Capacity: Access to approximately 3.5 gigawatts of next-gen TPU capacity CNBC.
- Timeline: New capacity is expected to turn into operational starting in 2027 Anthropic News.
- Financial Growth: Run-rate revenue has surged to over $30 billion, up from $9 billion at the end of 2025 Anthropic News.
- Enterprise Scale: More than 1,000 business customers now spend over $1 million annually CNBC.
- Domestic Focus: Expands a $50 billion commitment to invest in U.S. Computing infrastructure Anthropic News.
The next major milestone for this partnership will be the phased rollout of the new TPU capacity beginning in 2027. As Anthropic continues to scale its infrastructure and revenue, the industry will be watching to see how this massive influx of compute translates into the capabilities of future Claude models.
Do you think the race for compute capacity is the biggest bottleneck for AI development today? Share your thoughts in the comments below.