Microsoft to scale custom AI chips in multi-gigawatt push to cut costs

What happened: Microsoft (NASDAQ:MSFT) plans to deploy custom AI processors across its datacenters at multi-gigawatt scale starting in 2027. The Azure Maia 300 will power internal AI apps including GitHub Copilot and Microsoft Foundry. Key numbers: Maia 200 delivers up to 1.7 times higher Why it matters: The move aims to cut reliance on third-party hardware vendors and lower operating costs as cloud providers face rising AI expenses and GPU supply shortages. Google, Amazon, and Meta have also accelerated proprietary chip programs, challenging Nvidia's datacenter accelerator dominance. Market: +0.4% by next close ($543.45 on Oct 13), +13.6% YTD

Microsoft (NASDAQ:MSFT) plans to deploy custom AI processors across its datacenters at multi-gigawatt scale starting in 2027. The Azure Maia 300 will power internal AI apps including GitHub Copilot and Microsoft Foundry.

The move aims to cut reliance on third-party hardware vendors and lower operating costs as cloud providers face rising AI expenses and GPU supply shortages. Google, Amazon, and Meta have also accelerated proprietary chip programs, challenging Nvidia's datacenter accelerator dominance. Market: +0.4% by next close ($543.45 on Oct 13), +13.6% YTD