What happened
Anthropic has gained a competitive advantage over OpenAI by securing cheaper and more efficient AI chips through a multibillion-dollar deal with Google. This deal allows Anthropic to use Google's custom Tensor Processing Units (TPUs), which offer a superior price-performance ratio compared to the GPUs OpenAI uses from Microsoft and Nvidia. This efficiency focus could be critical as AI training costs continue to rise.
Why it matters
- Cheaper, more efficient chips reduce Anthropicâs infrastructure costs, potentially enabling more affordable AI model training.
- Anthropicâs hardware-software codesign partnership with AWS and Google may accelerate innovation in AI scalability and performance.
- Cost efficiency could influence the competitive dynamics of AI adoption and investment in the industry.
Key facts
- 1. Anthropicâs deal with Google involves expanding use of custom TPUs to secure compute capacity at a superior price-performance ratio.
- 2. As of Q2 2025, Anthropicâs cloud spending is over twice smaller than OpenAIâs, reflecting its cost discipline.
- 3. Anthropic projects revenue growth from $1 billion in 2024 to $9 billion annualized by the end of 2025, with a target of $26 billion by the end of 2026.
