Microsoft has unveiled Maia 300, the third generation of its custom-built AI chip, and plans to deploy potentially more than one million of the processors next year to power its own internal AI services.
The announcement, made on Monday, marks the company's most aggressive move yet to reduce its dependence on Nvidia's graphics processors. Microsoft CEO Satya Nadella called the effort "our biggest bet yet on our own silicon" in a post on X, while Azure chief Scott Guthrie said in a blog post that Maia will become "the primary AI inference system for Microsoft products and services."
40% more work per watt
According to Microsoft, the Maia 300 delivers about 40% more useful computing per megawatt than what it described as "leading GPU-based systems," without naming the comparison point. A full rack of the new chips, the company said, will produce up to seven times as much AI output as the current generation.
The chip takes a different technical path from Nvidia's accelerators. Its central compute processor is built from chiplets — specialized silicon bonded together — using 2-nanometer technology from Taiwan Semiconductor Manufacturing Co for key computing functions, while cheaper 3-nanometer chiplets handle data movement inside the system.
Maia 300 processors will ship in racks holding 96 chips that draw about 167 kilowatts of power, compared with 142 kilowatts for Nvidia's most recent GB300 AI server racks, which carry 72 GPUs.
Loosening Nvidia's grip
The launch arrives amid a flurry of custom chips from cloud-computing rivals and AI firms such as OpenAI, all aimed at loosening Nvidia's hold on the AI computing market. Microsoft designs the silicon itself and pays TSMC to manufacture it, rather than buying finished chips from Nvidia — a shift that could save the company billions of dollars.
Maia chips are designed for the workloads growing fastest right now: AI agents that carry out multistep jobs, models handling long documents or conversations, and coding assistants. Microsoft's predecessor chip is already running the company's in-house AI models as well as OpenAI models delivered on the Azure cloud.
Partners, for now
Despite the push, Microsoft and Nvidia remain close partners. Nadella and Nvidia CEO Jensen Huang shared a stage last week to launch Windows PCs built on Nvidia chips. Nadella said earlier this year that Microsoft will be among the first cloud providers to deploy next-generation systems based on Nvidia's Vera Rubin and AMD's Helios.
Large-scale Maia 300 deployments are expected to begin in 2027. With power supply now one of the biggest constraints on AI expansion, the race to build the most efficient chip — not just the fastest — is becoming the industry's next battleground.
By Hamza Faisal
