Microsoft announced its latest in-house AI chip and set a target for a multi-gigawatt data center buildout.

Microsoft has announced Azure Maia 300, the latest generation of its in-house accelerator chips, designed to power AI inference across Microsoft Azure. Maia 300 is built for demanding AI workloads such as multi-agent systems and low-latency conversational interactions.

The chip uses TSMC's multi-chiplet, two-nanometre process node architecture for performance and power improvements, and High Bandwidth Memory 4 (HBM4) for higher data throughput. Agentic systems need additional memory to maintain context and continuity over long-horizon, multi-step tasks. Microsoft said Maia 300 has a peak memory bandwidth of 18 TB/s, more than double that of Nvidia's widely used GB300 compute platform.

According to Microsoft, these improvements allow Maia 300 to deliver 40 per cent more tokens per megawatt than Maia 200, and up to seven times the token throughput.

Microsoft already runs production AI workloads on its Maia chips across the Azure stack, including Microsoft Copilot, GitHub Copilot and its in-house AI models. Maia 300 will sit alongside Maia 200 in Microsoft data centres.

Microsoft has committed to scaling its Maia chips to multi-gigawatt deployments in 2027, at which point Maia 300 will become the primary system it uses for AI inference.

Rani Borkar, president, Azure Hardware Systems and Infrastructure, said: "The economics of AI are increasingly determined not by peak compute performance, but how efficiently an entire system can generate tokens at scale.

"So we designed Maia 300 end to end to be our most efficient, scalable and reliable inference platform yet. The opportunity ahead is not simply to build more compute. It is to make advanced intelligence more accessible, affordable and useful."

In-house chips have become a priority for many AI firms, which aim to optimise costs through power efficiency and reduce their reliance on third-party GPUs from suppliers such as Nvidia and AMD. Google has used its in-house tensor processing units for AI training and inference for more than a decade, and Amazon has used its Trainium and Inferentia chips for several years. OpenAI and Broadcom worked together on the Jalapeño chip, which the AI developer said delivers faster inference than its previous workhorse hardware. Meta and DeepSeek are reportedly developing their own chips.

Austin Lyons, founder and principal analyst at Chipstrat, said: "The advantage of custom silicon is that Microsoft can optimize for the latency its own workloads actually require.

"A meeting summary and a coding agent don't need the same response time. Maia targets more throughput at each of those latency points, translates to more useful tokens served per megawatt and lower cost passed for customers."