Microsoft Rolls Out Maia 200 AI Chips | 30 Jan 2026

Why in News? 

Microsoft has launched its second-generation AI chip, the Maia 200, marking a significant step in the company’s strategy to develop custom AI silicon for large-scale AI workloads.

Key Points: 

• Fabrication: Manufactured using TSMC’s 3nm process technology, featuring over 140 billion transistors. 
• Performance: Delivers 10+ petaFLOPS in 4-bit precision (FP4) and ~5 petaFLOPS in 8-bit precision (FP8). 
• Memory Architecture: Equipped with 216GB of HBM3e (High Bandwidth Memory) with 7 TB/s bandwidth and 272MB of on-die SRAM to eliminate data movement bottlenecks. 
• Networking: Uses a two-tier scale-up design based on standard Ethernet instead of proprietary fabrics, supporting clusters of up to 6,144 accelerators. 
• Vertical Integration: Microsoft joins Google (TPU) and Amazon (Trainium) in designing custom hardware to reduce dependence on Nvidia and lower operational costs. 
• Economic Efficiency: The chip offers a 30% improvement in performance-per-dollar compared to current systems. 
• Software Ecosystem (The "Triton" Advantage): Microsoft released the Triton compiler (developed with OpenAI) as an open-source alternative to Nvidia's proprietary CUDA software, aiming to lower the barrier for developer adoption. 
• Strategic Significance:Represents Microsoft’s push to reduce reliance on third-party silicon providers and manage AI operating costs.