Parts/Accelerators/Microsoft Maia 200
Accelerators · Maia · 2026
Microsoft Maia 200
Microsoft's inference accelerator on TSMC 3nm: over 10 PFLOPS FP4, over 5 PFLOPS FP8, 216 GB HBM3e at 7 TB/s, 750 W.
Why this part
Announced 2026-01-26 and deployed in Azure US Central. Microsoft states the compute figures as 'over'.
Systems using this part
No systems in the current dataset record this part. It is carried in the taxonomy so the graph is complete, and so edges can attach to it as coverage grows.