TechnologyTracking
Microsoft making Maia primary AI inference from 2027: impact on Nvidia/AMD and chip market?
Microsoft's plan to make Maia its primary AI inference system from 2027 with multi-gigawatt deployments—how will this reshape its procurement relationship with Nvidia and AMD, and what are the broader implications for the AI chip market?
Watching: 微软在2026年11月Ignite大会及后续财报中披露的Maia部署细节、对英伟达和AMD采购量的变化、以及竞争对手的回应。
Current read
Microsoft released Maia 300, planning to make it the primary inference system by 2027 with multi-gigawatt deployment, while continuing to purchase Nvidia and AMD chips, maintaining a dual-track strategy.
Microsoft released the Maia 300 chip in October 2026 to reduce dependence on Nvidia and AMD, planning to deploy it as the primary inference system at multi-gigawatt scale by 2027. However, CEO Nadella explicitly stated continued purchases from Nvidia and AMD, emphasizing partnerships. This indicates a dual-track strategy of partial substitution rather than full replacement. The evidence does not provide specific procurement ratios or deployment scale post-2027, so the impact on supplier relationships and market competition cannot be quantified.
Key developments
- The assessment representation was corrected; evidence and earlier records are retained.
How we read it
Microsoft released Maia 300, with plans for it to account for a significant share of AI chip spending by Q1 2027, indicating accelerated in-house deployment.
CEO Nadella committed to continued purchases from Nvidia and AMD, emphasizing partnerships, indicating partial substitution rather than full replacement.
Evidence lacks specific procurement ratios or deployment scale post-2027, so long-term impact on supplier relationships or market competition cannot be quantified.
Earlier updates
- — Public tracking started; no independent probability estimate yet.; Starting public context: Microsoft announced the Maia 300 AI chip and plans to scale its Maia silicon family to multi-gigawatt deployments, positioning Maia as the primary inference system for its own products and services starting in 2027. Maia 200 is already serving production workloads including Microsoft Copilot, with GitHub Copilot and Microsoft Foundry to follow. Microsoft claims Maia 300 delivers 40% more tokens per megawatt than leading GPU-based systems, part of a push to reduce reliance on Nvidia and AMD and cut costs.
- — Microsoft released Maia 300, planning to make it the primary inference system by 2027 with multi-gigawatt deployment, while continuing to purchase Nvidia and AMD chips, maintaining a dual-track strategy.; Microsoft released the Maia 300 chip in October 2026 to reduce dependence on Nvidia and AMD, planning to deploy it as the primary inference system at multi-gigawatt scale by 2027. However, CEO Nadella explicitly stated continued purchases from Nvidia and AMD, emphasizing partnerships. This indicates a dual-track strategy of partial substitution rather than full replacement. The evidence does not provide specific procurement ratios or deployment scale post-2027, so the impact on supplier relationships and market competition cannot be quantified.