Microsoft unveils its third-gen in-house AI chip, the Maia 300, built on TSMC's N2P process node and aimed at serving as its "primary AI inference system" (Tom Warren/The Verge)
Microsoft says Maia 300 will become the main inference engine for its own products and services starting next year.
The chip is Microsoft’s third-generation in-house AI accelerator and is built on TSMC’s N2P process. Satya Nadella said Microsoft will scale the Maia family to multi-gigawatt deployments. Microsoft says Maia 300 delivers 40% more inference tokens per megawatt than leading GPU-based systems and up to 7x the rack-scale token throughput of Maia 200. Techmeme's note
The chip is Microsoft’s third-generation in-house AI accelerator and is built on TSMC’s N2P process. Satya Nadella said Microsoft will scale the Maia family to multi-gigawatt deployments. Microsoft says Maia 300 delivers 40% more inference tokens per megawatt than leading GPU-based systems and up to 7x the rack-scale token throughput of Maia 200. Techmeme's note
score 7