Microsoft unveils Maia 300 AI chip to cut reliance on Nvidia GPUs

Microsoft announced its third-generation Maia 300 chip, which will become the primary AI inference system for its products and services. Built on TSMC's N2P process with a chiplet architecture, it promises 40 percent more tokens per megawatt than leading GPU-based systems, with production rollout starting in 2027.

• Maia 300 delivers 40 percent more tokens per megawatt than GPU-based systems
• In a 96-accelerator configuration it offers up to 7x the token throughput of Maia 200
• Chiplet architecture on TSMC N2P separates compute, memory controller and I/O
• Production rollout is scheduled to begin in 2027