Microsoft Maia 300
Microsoft’s third Maia moves to 2nm-class chiplets and HBM4 for multi-gigawatt Azure deployments from 2027.
Silicon from Microsoft
3 chips on record, newest first.
-
2026
Microsoft Maia 300
TSMC N2P (compute chiplets) Inference accelerator, eight HBM4 stacks, 950 W TDP, 96 per rack
up to 2.5x more performance per watt compared to Maia 200
Relative only -
2026
Microsoft Maia 200
TSMC 3nm Inference SoC, 272 MB on-chip SRAM, 750 W
over 10 petaFLOPS in 4-bit precision (FP4) per chip
Absolute -
2023
Microsoft Maia 100
TSMC 5nm First-gen Azure AI accelerator, about 820 mm2, 500 MB of L1/L2
Peak Dense Tensor POPS 6bit: 3, 9bit: 1.5, BF16: 0.8
Absolute