Microsoft Announces New AI Chip, Maia 300
Microsoft has announced its third-generation in-house AI chip, Maia 300, designed to serve as the primary system for AI inference across Microsoft's products and services. The new chip is expected to improve efficiency by delivering up to 40% more tokens per megawatt compared to leading GPU-based systems like those from Nvidia. Maia 300 will begin rolling out in production workloads starting in 2027, allowing Microsoft to reduce its reliance on third-party AI hardware providers such as Nvidia and AMD. The chip is built using TSMC’s N2P process node with a modular architecture that enhances compute performance while optimizing efficiency. Microsoft aims to scale Maia deployments across multiple gigawatts by 2027, supporting advanced workloads like long-context reasoning and multi-step agentic tasks. This move aligns with Microsoft's broader strategy of developing in-house AI infrastructure to compete more effectively against rivals such as Google DeepMind, OpenAI, and Anthropic.
Topics: AI chips cloud computing Microsoft Maia 300 Nvidia GPUs
Read the full article at The Verge
Part of the story Microsoft Launches New AI Decision Model (8 articles)