Microsoft is reportedly preparing to unveil its next-generation Maia 300 artificial intelligence chip as early as September, marking another step in the company's effort to expand its custom silicon programme and reduce its dependence on Nvidia's processors.

According to a report by The Information, cited by Reuters, Microsoft could introduce the new AI accelerator as soon as next month. The company has not announced an official launch date, but its latest move underlines the growing competition among major technology companies to develop their own hardware for increasingly demanding AI workloads.

Microsoft entered the custom AI chip race in November 2023, when it introduced its first Maia processor. Since then, the company has been working to expand its in-house silicon capabilities as demand for AI computing continues to rise.

Maia 300 Could Mark Bigger Push

The reported Maia 300 launch comes as Microsoft seeks to scale up its custom-chip operations.

Microsoft's Azure cloud business relies heavily on advanced computing infrastructure to support AI services and large language models. Developing its own processors could give the company greater control over hardware availability, performance and costs while reducing exposure to the supply constraints and pricing of third-party accelerators.

The move also comes at a time when major cloud providers are increasingly developing specialised chips for artificial intelligence.

Google and Amazon Step Up

Microsoft is competing with technology giants including Alphabet's Google and Amazon, both of which have invested heavily in their own AI processors.

Google has developed its Tensor Processing Units (TPUs) for AI workloads and has expanded their use across its cloud operations. Amazon has similarly developed its Trainium family of processors as part of its strategy to offer alternatives to Nvidia hardware.

The growing investment in custom silicon reflects the enormous cost of operating AI data centres, where processors account for a significant portion of infrastructure expenditure.

TSMC Manufacturing Talks

Microsoft is reportedly in discussions with Taiwan Semiconductor Manufacturing Company (TSMC) to secure manufacturing capacity for the Maia 300.

According to the report, Microsoft has been seeking capacity for more than 300,000 chips, with deliveries potentially beginning in 2027.

The company is also reportedly looking to significantly increase production and encourage major cloud customers, including AI company Anthropic, to use the Maia platform.

Microsoft, however, has indicated that the production figures cited in the report do not represent the full scale of its custom-silicon programme.

"Microsoft continues to invest in custom silicon as part of our long-term AI infrastructure strategy," Andrew Wall, general manager for Microsoft's Azure Maia, said in a statement. He added that the company does not disclose production volumes.

Ambitious Production Target

Microsoft's longer-term objective could be considerably larger than the initial reported production plan.

According to The Information, the company ultimately aims to secure manufacturing capacity for more than 1 million Maia 300 chips. However, the availability of components and ongoing negotiations over manufacturing capacity could affect the final scale of production.

Securing sufficient semiconductor capacity has become increasingly important as technology companies compete for advanced chips needed to expand their AI infrastructure.

Built on Earlier Generations

The Maia 300 would follow Microsoft's earlier custom AI processors.

The company unveiled the Maia 200 in January, with the second-generation processor manufactured by TSMC using 3-nanometre technology.

The progression from Maia 100 to Maia 200 and the reported development of Maia 300 indicates Microsoft's intention to build a sustained custom-silicon roadmap rather than treating its chips as a one-off experiment.

Nvidia Still Dominates AI Chips

Despite Microsoft's efforts, Nvidia remains one of the most important suppliers of AI accelerators used by cloud providers and AI companies worldwide.

Microsoft's custom-chip programme is therefore not necessarily an immediate replacement for Nvidia hardware. Instead, the strategy could give the company another source of computing capacity while allowing it to optimise certain workloads for its own cloud infrastructure.

The development also highlights a broader shift in the technology industry. As AI becomes increasingly central to cloud computing, software companies and cloud operators are investing more heavily in semiconductor design to gain greater control over the underlying technology.

If Microsoft proceeds with the reported September unveiling, Maia 300 could become an important component of the company's long-term strategy to build a larger, more diversified AI computing infrastructure.