Microsoft has announced the development of two custom-designed AI chips, the Azure Maia 100 AI Accelerator and the Azure Cobalt 100 CPU, in response to the global shortage of GPUs. The company aims to optimize its data centers and maximize performance by creating its own infrastructure to support AI innovation. The Maia 100 chip, containing 105 billion transistors, will be used to train and run AI models, while the Cobalt 100 chip is designed for general purpose workloads. Both chips will be rolled out to Azure data centers next year and will initially power Microsoft AI services such as Copilot and Azure OpenAI Service. Microsoft plans to continue developing second-generation Maia and Cobalt hardware. The announcement comes after reports that Nvidia’s AI cards are sold out until 2024, and the CEO of chipmaker TSMC suggested that the shortage could extend into 2025.
Microsoft Breaks Free from GPU Constraints with Innovative Custom AI Chips
