Microsoft is accelerating its shift toward using its own AI chips, reducing its long-standing dependence on Nvidia's graphics processing units (GPUs). The company unveiled its third-generation Maia AI accelerator, signaling a move to rely more on homegrown silicon. Microsoft's Azure cloud has been a cornerstone of its AI strategy, powering models for clients and supporting services like Microsoft 365 Copilot and GitHub Copilot. While the company develops in-house AI models, it has yet to make a significant impact in a market dominated by OpenAI, Anthropic, and Google.
The demand for advanced processors and the market's desire for alternatives have created an opportunity for Microsoft's AI chips. Nvidia's GPUs have been the industry standard, but Microsoft is now embracing a variety of silicon solutions. The company already uses its own central processing units (CPUs) alongside silicon from Intel and AMD. Finance chief Amy Hood noted that using its own chips has contributed to margin improvements over the past decade.
Microsoft introduced its first AI chip, the Maia 100, in 2023, followed by the Maia 200 in January. CEO Satya Nadella highlighted that the Maia 200 delivers 30% better performance per dollar compared to the latest hardware in their fleet. Production of the Maia 300 will begin in 2027, with Microsoft targeting multi-gigawatt deployments. The new version can handle up to seven times more AI tokens per rack and offers 40% more tokens per megawatt compared to GPUs.
Despite the shift, Nvidia will remain a key partner. Microsoft recently announced a Surface laptop containing an Nvidia processor. Evercore ISI analysts estimate that Maia chips will represent about 10% of Microsoft's spending on AI chips by the first quarter of 2027. However, Microsoft's accelerators are not available for rent to other companies, limiting their external testing.