Microsoft shares rise to a year high as reports say it will boost AI chip output

Microsoft’s stock rose to its highest level this year as market reporting said the company is planning to increase production of its custom AI chips. The update also points to a broader shift in Microsoft’s approach to building and supplying the compute needed for AI workloads, with an emphasis on its own silicon and careful sourcing of manufacturing capacity.

According to the report circulated by Yahoo Finance, Microsoft is preparing to unveil its Maia 300 AI chip in the fall. Maia 300 is part of Microsoft’s “custom silicon” strategy, meaning the company designs chips tailored to its AI and cloud infrastructure rather than relying entirely on processors built by other companies.

The same reporting said Microsoft is also working to lock in additional manufacturing capacity at TSMC, one of the world’s leading semiconductor foundries. Securing foundry output is a key operational lever for AI chipmakers, because advanced manufacturing capacity can be limited and can drive costs when demand outpaces supply.

The report further suggested the capacity steps are intended to reduce reliance on expensive Nvidia hardware. For Microsoft, this matters because AI training and inference at scale require large amounts of specialized compute, and the economics of those systems can hinge on chip pricing, availability, and lead times for procurement.

Beyond the chip-specific claims, the market reaction underscores investor focus on whether major cloud and AI platforms can stabilize the supply of high-performance chips as demand continues to grow. Even when companies do not disclose detailed production schedules or procurement volumes, reports about ramp plans and manufacturing arrangements can move expectations for future AI infrastructure costs and capacity.

Microsoft’s push for its own accelerators has been positioned as a way to optimize performance and efficiency for workloads that run on its cloud. Custom silicon can also give a platform owner more control over system-level design, including how chips interact with the rest of its data center stack, such as networking and storage layers.

Still, key details were not provided in the market report. It did not specify target production quantities for Maia 300, the timing of chip volume shipments, or the exact terms and capacity commitments with TSMC. It also did not quantify how much Nvidia usage Microsoft expects to displace or over what time frame, leaving the magnitude of the potential cost benefit uncertain.

Going forward, investors and customers will likely watch for more concrete disclosures around Maia 300, including any announcements tied to deployments on Microsoft’s cloud infrastructure. They will also look for additional indicates on how quickly Microsoft can translate confirmed manufacturing access into actual data center capacity for AI services.