Anthropic in Talks with Microsoft over AI Chips After $5 Billion Investment
Key point
Anthropic is in talks to adopt Microsoft's proprietary AI chip, **Maia**, to secure computing resources.
Details
Anthropic is in negotiations to adopt Microsoft's custom AI chip, Maia. Microsoft announced the processor last January, but has not yet officially made it available to Azure customers.
If this deal is finalized, Microsoft will gain an edge over rivals Amazon and Google in the field of special-purpose AI silicon supply. In particular, the next-generation Maia 200 processor is set to power OpenAI's GPT-5.2 model, and has improved token efficiency per dollar by more than 30% compared to before.
Anthropic has recently been struggling to secure computing resources due to the popularity of the Claude assistant and Claude Code. To address this, it is utilizing the following infrastructure:
- Nvidia GPUs (existing primary model)
- AWS Trainium (a 10-year contract worth more than $100 billion)
- Google TPU (adoption plan announced last October)
- SpaceX computing power (approximately $1.25 billion per month)
Earlier, Microsoft invested $5 billion in Anthropic, and Anthropic committed to spending $30 billion on Azure.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.