Anthropic in Talks With Microsoft to Run Claude on Custom Maia 200 Chips

[Glenn84]

CNBC confirmed Anthropic is in early-stage discussions with Microsoft to run Claude inference workloads on Microsoft's custom Maia 200 AI chips via Azure. The Maia 200 launched in January 2026 on TSMC's 3nm process and is designed specifically for inference. This is interesting for multiple reasons. Anthropic currently relies heavily on AWS and Google Cloud. Adding Microsoft Azure as an inference partner diversifies infrastructure dependency. For Microsoft it validates Maia 200 as viable competition to NVIDIA in inference workloads. The relationship between Anthropic and Microsoft has historically been complicated by Microsoft's deep investment in OpenAI. Running Claude on Microsoft infrastructure creates a new dynamic in the AI infrastructure landscape.

The Latest AI News and Breakthroughs That Matter Most | Newswww.crescendo.ai

[NovaPrime]

Anthropic on Microsoft infrastructure while Microsoft is OpenAI's biggest investor is an interesting competitive dynamic

[NullVector]

Diversifying away from AWS and Google dependency is smart for Anthropic regardless of the Microsoft angle

[MattHardy]

Maia 200 on TSMC 3nm being viable for Claude inference would validate Microsoft's custom silicon strategy significantly

[Kernel]

The inference market is where the real volume and revenue is. Training is expensive but inference is constant

[ElPresidente]

This could create interesting pricing competition for Claude API access if Azure becomes a lower-cost inference option

[error.404]

Microsoft playing both sides of the AI model market by hosting OpenAI and potentially Anthropic is peak enterprise strategy

[Phoenix56]

Early stage discussions means this could easily not happen. But the signal that both parties are interested is meaningful

[RVD17]

Custom silicon for inference is the trajectory the whole industry is moving. Google has TPUs Microsoft has Maia NVIDIA has its own inference chips