Anthropic's $30B Run Rate and Massive Google TPU Deal Raises Eyebrows at Broadcom

Anthropic, the AI startup backed by Google and Amazon, has revealed it has surpassed a $30 billion annual revenue run rate—up from approximately $9 billion at the end of 2025—as it secures access to 3.5 gigawatts of next-generation Google Tensor Processing Units (TPUs) through a deal with Broadcom.

Broadcom's Dual Announcements Signal Major AI Infrastructure Expansion

The chipmaker's regulatory filing, which opened with two significant items, revealed that Google has entered into a long-term agreement for Broadcom to develop and supply custom TPUs for future generations of Google's AI accelerators. This partnership builds on years of collaboration between the two companies, with Broadcom CEO Hock Tan recently predicting the company's chip business will generate over $100 billion in AI chip revenue by 2027 alone.

"Working on next-gen TPUs for Google will presumably help to make that prediction a reality," noted industry observers following the announcement.

Anthropic's Massive Compute Commitment

The second major revelation in Broadcom's filing was a supply assurance agreement to provide networking and other components for Google's next-generation AI racks through 2031. Under this agreement, Anthropic will access approximately 3.5 gigawatts of next-generation TPU-based AI compute capacity starting in 2027.

This represents a substantial portion of the "multiple gigawatts" of AI compute capacity committed by Anthropic across its various partnerships. The scale is staggering—3.5 gigawatts is equivalent to the power consumption of a small city, highlighting the enormous computational demands of frontier AI development.

Financial Concerns Lurk Beneath the Surface

Despite Anthropic's impressive revenue growth claims—doubling its number of customers spending over $1 million annually from 500 to more than 1,000 in under two months—Broadcom's filing includes a notable cautionary statement: "The consumption of such expanded AI compute capacity by Anthropic is dependent on Anthropic's continued commercial success."

The filing further notes that "in connection with this deployment, the parties are in discussions with certain operational and financial partners," suggesting Broadcom is taking steps to mitigate financial risk associated with such a massive infrastructure commitment.

Multi-Cloud Strategy for AI Workloads

Anthropic appears to be pursuing a multi-cloud strategy to support its AI development. While securing the massive Google TPU commitment, the company emphasized it also uses AWS's Trainium AI chips and Nvidia GPUs, allowing it to "match workloads to the chips best suited for them."

Google's announcement of the deal highlighted that Anthropic is not only a significant TPU customer but also a major Google Cloud customer, suggesting a deepening partnership between the AI startup and the search giant.

Industry Context and Implications

The deal comes amid intense competition in the AI infrastructure space. Amazon recently built a massive AI supercluster for Anthropic called Project Rainier, while other tech giants are racing to secure their positions in the AI chip market. Broadcom's prediction of $100 billion in AI chip revenue by 2027 underscores the scale of investment flowing into this sector.

For Anthropic, the combination of rapid revenue growth and massive compute commitments positions it as a serious contender in the AI race. However, the cautionary language in Broadcom's filing serves as a reminder that even well-funded AI startups face significant execution and financial risks as they scale their operations to meet the enormous computational demands of frontier AI development.

The next few years will be critical in determining whether Anthropic can successfully convert its massive infrastructure investments and growing customer base into sustainable profitability in the increasingly competitive AI market.