Microsoft Expands Production of In-House AI Chips, Reportedly Secures Over 300,000 Orders from TSMC

According to a Monday (10th) report by The Information, Microsoft (MSFT-US) plans to significantly increase production of its in-house artificial intelligence (AI) chips.

Citing sources familiar with the matter, the report states that Microsoft could launch its Maia 300 chip as early as next month. It adds that Microsoft has been in discussions with semiconductor foundry TSMC (2330-TW) to place orders for over 300,000 Maia 300 chips, with delivery expected by 2027.

Andrew Wall, General Manager of Azure Maia at Microsoft, told Barron’s, “Microsoft continues to invest in custom chips as part of our long-term AI infrastructure strategy. We don’t disclose production volumes, but the numbers reported do not reflect the scale of our overall plans.”

Wall further stated that Microsoft expects “Azure Maia deployments to support AI workloads measured in gigawatts.”

On Monday, Microsoft’s stock rose 1.1%, while TSMC’s ADR closed down slightly by 0.37%.

Microsoft first introduced the Maia chip in November 2023, followed by the Maia 200 in January this year. With soaring demand for AI computing and ongoing supply constraints and rising costs, Microsoft’s move to expand in-house chip production is unsurprising.

Global tech giants are investing hundreds of billions of dollars to build infrastructure supporting AI computing. In this high-capital-expenditure environment, developing in-house chips to reduce costs has become crucial—especially as these companies currently rely heavily on a single semiconductor supplier, Nvidia (NVDA-US), for AI chips.

Google Cloud, under Alphabet (GOOGL-US), has developed its own Tensor Processing Unit (TPU), while Amazon (AMZN-US)’s Amazon Web Services (AWS) offers the Trainium AI accelerator.

Jim Hines, IDC’s Research Director for Computing Systems Platforms and Technologies, said, “Microsoft’s development of in-house chip design capabilities reflects a broader trend of vertical integration among hyperscale cloud providers. Facing rising costs, power constraints, and supply limitations, companies are increasingly integrating the entire technology stack of AI infrastructure.”