DEVOURED
Microsoft Prepares to Put Gigawatts of Its Chips in Data Centers
Microsoft plans to deploy gigawatts of its proprietary Maia 300 inference chips across its data centers starting next year.
What: The Maia 300 is an AI inference chip that Microsoft claims delivers 40% more tokens per megawatt than current GPU-based systems. Microsoft does not currently intend to offer these chips as a rentable cloud service to customers.
Why it matters: Building proprietary silicon allows hyperscalers to optimize energy efficiency and throughput specifically for their own internal large-scale AI workloads, reducing long-term dependence on merchant silicon suppliers like Nvidia.
Decoder
- Inference: The process of running a trained machine learning model to produce predictions or outputs based on new input data.
- Hyperscaler: Large cloud service providers such as AWS, Microsoft Azure, and Google Cloud that operate massive data centers.
Original article
Microsoft's newest iteration of its Maia chips will start rolling out next year. The Maia 300 is designed to handle inference. Microsoft expects to stock multiple gigawatts of the chips at its server farms. The Maia 300 delivers 40% more tokens per megawatt than leading GPU-based systems. Microsoft isn't currently planning to rent the underlying computing power of Maia chips directly to its cloud customers.
Explore all 41 articles in today's digest
Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi