Microsoft to Scale In-House Maia AI Chips Starting in 2027

what happened

According to Stocktwits.com, Microsoft (MSFT) plans to deploy its Azure Maia systems at multi-gigawatt scale beginning in 2027. The company intends to make Maia the primary system for running AI models across its products and services. The chip was developed to run large AI models in Azure cloud data centers.

Microsoft will focus on Maia 200 and Maia 300. Introduced in January, Maia 200 already powers Microsoft Copilot, while GitHub Copilot and Microsoft Foundry are expected to follow by year-end. The rollout is currently under way in the U.S., Europe, and Australia. Microsoft says Maia 300 will produce 40% more AI-generated tokens per megawatt than “leading GPU-based systems,” while a system with 96 Maia 300 chips could deliver up to seven times the processing output of Maia 200.

CEO Satya Nadella described the project as the “biggest bet yet on our own silicon.” Microsoft is not abandoning Nvidia (NVDA) or Advanced Micro Devices (AMD): in July, it reaffirmed plans to deploy Nvidia’s Vera Rubin systems and AMD’s Helios infrastructure alongside Maia. MSFT shares have risen more than 55% from their 52-week low in June and about 14.5% this year, ahead of the Roundhill Magnificent Seven ETF (MAGS), up around 13.4%.

why it matters

The announcement highlights Microsoft’s effort to gain more control over AI infrastructure costs while continuing to use chips and systems from Nvidia and AMD. Stocktwits retail sentiment toward MSFT remained “bearish” over the past 24 hours.

instruments