No More Relying on NVIDIA: Microsoft Secures TSMC Capacity for Massive Maia 300 Expansion

Microsoft (MSFT-US) is planning to significantly ramp up production of its next-generation in-house artificial intelligence (AI) chip and is seeking adoption from major cloud customers like Anthropic to reduce its dependence on NVIDIA's (NVDA-US) AI accelerators.

According to a report by tech publication The Information on Monday (10th), citing sources familiar with the matter, Microsoft is preparing to launch its new AI chip, Maia 300, in the fall of this year, with a possible unveiling as early as August and an official release expected in September. Microsoft has already entered discussions with TSMC (TSM-US) (2330-TW) to secure manufacturing capacity, aiming to produce over 300,000 Maia 300 chips, with deliveries expected by 2027.

The Maia 300 is an upgraded version of Microsoft's in-house AI accelerator series, primarily designed to meet the growing AI computing demands on its Azure cloud platform. The report indicates that Microsoft aims to accelerate the development of its proprietary AI chip ecosystem by scaling up production and onboarding large customers. While a Microsoft executive declined to comment on specific production plans, they noted the company's ultimate goal is to deploy Maia chips at a scale reaching several gigawatts.

However, market adoption of the current Maia 200 remains relatively slow. Despite Microsoft's substantial investment in developing dedicated AI chips, customers face challenges in transitioning their existing computing architectures and deploying at scale.

As generative AI drives a surge in computing demand, major cloud service providers are accelerating the development of in-house chips to reduce reliance on NVIDIA's GPU supply chain. In addition to Microsoft, Google under Alphabet (GOOGL-US) and Amazon (AMZN-US) have already launched their own AI chips, intensifying the chip race among tech giants.