Microsoft Maia
Azure's custom AI accelerator for first-party and OpenAI workloads
Microsoft designs the Maia AI accelerator (Maia 100 launched late 2023) to run Azure AI and OpenAI workloads and reduce Nvidia dependence, paired with its Cobalt Arm-based server CPU. A next-gen Maia (Braga / Maia 200) has faced reported delays into 2026, so Azure still leans heavily on Nvidia and AMD GPUs in the interim. Maia is fabbed at TSMC and built with a custom liquid-cooling 'sidekick' rack design.
AI chip
Maia 100 (2023); Maia 200/Braga next
Companion CPU
Cobalt (Arm-based)
Status
Next-gen reportedly delayed to 2026
Cooling
Custom liquid-cooled sidekick rack
How it fits the stack
Microsoft Maia with what it depends on (above) and what it feeds (below). The figure renders as a crawlable diagram and upgrades to an interactive 3D graph as it scrolls into view.
Microsoft Maia in the AI stack. Microsoft Maia with its immediate upstream dependencies (top) and downstream dependents (bottom) in the AI value chain. Hover a node in 3D, or read the full relationships below.
Graph data (text) — 8 entities, 7 relationships
- Microsoft Maia —uses→ High-Bandwidth Memory (HBM)
- Microsoft Maia —uses→ Liquid Cooling
- Microsoft Maia —depends on→ TSMC (Taiwan Semiconductor Manufacturing Company)
- Microsoft Azure —used by→ Microsoft Maia
- OpenAI —used by→ Microsoft Maia
- Microsoft Maia —partners with→ Marvell Technology
- Microsoft Maia —competes with→ Nvidia
Depends on ↑ · 3
Context — capital, rivals, policy · · 2