Anthropic have signed a deal for multiple gigawatts of next generation TPUs
Anthropic locks in massive compute power for future Claude models, securing its AI infrastructure for years.
Anthropic has inked a landmark, multi-year partnership with Google and Broadcom to secure a dedicated pipeline of next-generation Tensor Processing Units (TPUs), the custom AI chips powering Google's cloud services. The deal, reportedly valued at over $3 billion, guarantees Anthropic access to a massive supply of compute power measured in the gigawatts—a unit that reflects the immense energy required to run cutting-edge AI data centers. This strategic move locks in the critical hardware infrastructure Anthropic needs to develop future versions of its Claude AI models, such as the anticipated Claude 4, without being constrained by the global GPU shortage.
This partnership represents a deepening of the existing relationship between Anthropic and Google Cloud, which is also a major investor in the AI startup. By collaborating directly with Broadcom, the semiconductor designer of Google's TPUs, Anthropic is securing a more direct and reliable supply chain for the specialized silicon. The gigawatt-scale commitment underscores the staggering computational demands of frontier AI models and provides Anthropic with a long-term competitive moat. It ensures the company can execute its ambitious research roadmap and scale its model inference to meet growing user demand, solidifying its position as a top-tier AI lab alongside OpenAI and Google DeepMind.
- Anthropic's multi-year deal with Google and Broadcom is valued at over $3 billion.
- The agreement secures a supply of next-gen TPUs, with compute capacity measured in multiple gigawatts.
- This guarantees the infrastructure needed to train future Claude models and scale inference operations.
Why It Matters
Secures Anthropic's AI development pipeline against chip shortages, ensuring it can compete with OpenAI and Google at the frontier.