Microsoft calls Maia 300 its biggest silicon bet, multi-gigawatt deployment from 2027
pdamodaran · x · 2026-10-20
The thread maps how hyperscaler capex is shifting from third-party chips to internal silicon in 2027, answering why Amazon is up while other chip players are down:
Microsoft (Oct 12, 2026 announcements):
- Satya Nadella explicitly called Maia 300 "our biggest bet yet on our own silicon."
- Scott Guthrie said Azure Maia will scale to multi-gigawatt deployments starting 2027, becoming the primary AI inference system for Microsoft products and services.
- Maia is positioned as the main engine for Copilot, GitHub Copilot and Teams, with explicit language about reducing reliance on Nvidia GPUs; production and large-scale deployment of Maia 300 begin in 2027.
Meta: published a multi-generation MTIA roadmap with inference-focused chips MTIA 450 and MTIA 500 targeted for H1 2027 and later in 2027, framed as improving performance-per-watt and cost efficiency.
The author's takeaway: capex from MSFT, GOOG, AMZN and META is shifting heavily to internal chips in 2027.
Related event: Hyperscalers to Shift Capex to In-house Chips by 2027(2 posts)→
More from Companies & People
- Anthropic vows to keep Opus 3 forever as OpenAI sunsets 4o with two weeks' notice — repligate · 2026-10-20
- PyTorchCon North America kicks off tomorrow with free keynote livestream — PyTorch · 2026-10-20
- Calling Astra's launch AGI mocked as the year's most embarrassing moment — karmay007 · 2026-10-20
- Gensyn co-founder to discuss auditable training, AI free of few-company control — benfielding · 2026-10-20
- ReForge Podcast Ep.039 hosts benefits expert Nick Lease on worker health plans — MatthewChang · 2026-10-20
- UCSB's REAL AI lab hiring postdocs to build autonomous labs spanning AI, chemistry and robotics — ninamiolane · 2026-10-20