From the source
Microsoft announced Maia 300, its latest custom AI accelerator, which will begin deployment in 2027 and become the primary inference system for its own products and services.
The company says Maia 300 will deliver 40% more AI tokens per megawatt than leading GPU-based systems and is designed to minimize unnecessary data movement between chips and memory.
Microsoft's move is part of a broader industry shift as major AI players including Google, OpenAI,
Amazon,
Anthropic, and
Meta develop or explore their own
Nvidia alternatives.