Microsoft plans to introduce Maia 300 AI chip in 2027

what happened

Seeking Alpha reports that Microsoft said on Monday it intends to introduce the Maia 300 AI chip in 2027. The company is expanding its internally developed hardware to meet rising demand for AI services. The accelerator is built for inference, or running models after training, including applications such as chatbots, coding assistants and systems that perform tasks in several steps.

Microsoft said Maia 300 uses an inference-first architecture aimed at long-context, memory-intensive and decode-heavy workloads. The design combines several chiplets, with its computing components produced using Taiwan Semiconductor Manufacturing Co’s N2P process and eight stacks of high-bandwidth memory.

A rack equipped with 96 Maia 300 accelerators could provide up to seven times the token throughput of a Maia 200 rack, according to Microsoft. The company also expects the new chip to produce 40% more tokens per megawatt than leading graphics processing unit-based systems. Microsoft plans to scale Maia deployments to multi-gigawatt capacity in 2027. Maia 200 is already handling production workloads, including Microsoft Copilot, while support for GitHub Copilot and Microsoft Foundry is planned by the end of 2026.

why it matters

The announcement highlights Microsoft’s effort to build more of its own AI infrastructure and reduce reliance on Nvidia’s costly processors. Maia 300 is not planned until 2027, while Microsoft has previously lagged peers in scaling its Maia hardware.

instruments