Alphabet Inc.’s Google has entered into a landmark agreement with artificial intelligence startup Anthropic, committing to supply up to one million of its custom designed Tensor Processing Units (TPUs). The deal, valued in the tens of billions of dollars, marks one of the largest infrastructure commitments in the AI sector to date and is expected to bring over a gigawatt of computing capacity online by 2026.
TPUs are specialized chips developed by Google to accelerate machine learning workloads. Anthropic, known for its Claude family of AI models, will use this expanded compute power to train and deploy next-generation systems, including Claude Sonnet 4.5, which has gained recognition for its coding capabilities.
The deal strengthens Google’s position as both a strategic investor and infrastructure provider. Google has already invested approximately $3 billion in Anthropic, while Amazon remains a major backer with up to $8 billion committed and continues to supply custom chips via AWS. Anthropic’s diversified compute strategy spans Google’s TPUs, Amazon’s Trainium chips, and Nvidia’s GPUs, allowing the company to optimize workloads across platforms for cost, performance, and energy efficiency.
The startup recently closed a $13 billion Series F funding round led by Iconiq Capital, Fidelity, and Lightspeed Venture Partners, tripling its valuation to $183 billion. The capital will support global expansion, safety research, and infrastructure scaling to meet surging enterprise demand. With over 300,000 business customers and a projected revenue run rate approaching $7 billion, Anthropic is positioning itself as a formidable competitor to OpenAI and other generative AI leaders.
