Microsoft plans to rely much more on its custom Maia chips than on Nvidia GPUs in its Azure data centers starting next year. The shift adds to signs that Nvidia's biggest customers are turning to alternatives, while Marvell Technology stands to benefit as Microsoft's custom chip partner.
Microsoft is planning a big change to its Azure data centers starting next year, relying much more on its custom-designed silicon than on leading GPUs such as Nvidia's. The company expects its next-generation Maia 300 chips to deliver significant improvements once deployment begins.
According to CEO Satya Nadella: "Starting next calendar year, we will scale our Maia family to multi-gigawatt deployments", making it the primary inference system for Microsoft's own products and services.
Nvidia's biggest customers build their own chips
Nvidia's customers, including hyperscalers and leading AI frontier labs, have developed custom solutions that can handle AI workloads, and that can deliver better price performance on specific tasks than Nvidia's GPUs. Still, GPUs offer far greater flexibility, which makes them valuable for companies selling compute capacity to third parties.
Microsoft is starting its Maia expansion with its own products and services, not with leasing Maia capacity to third parties. Alphabet, meanwhile, has seen strong adoption of its Tensor Processing Units among Google Cloud customers, and Anthropic is using gigawatts worth of TPUs to develop its large language models.
Nvidia can continue to grow sales extremely quickly, but other chipmakers' growth rates are likely to outpace it. That is reflected in its P/E ratio, which now sits below those of its biggest rivals.
Marvell stands to gain from Maia
At its 2026 investor day this month, Marvell said it expects to grow its share of the custom AI accelerator market over the next five years. Its work on the Maia 300 is likely a big reason why.
Microsoft will spend approximately $175 billion in capital expenditures on data centers over the next year, and roughly two-thirds of that will go toward servers and networking equipment. Marvell's revenue totaled just $9.5 billion over the last four quarters, so even a sliver of that spending would boost it.
Marvell believes its custom AI chips and networking chips will help it grow revenue tenfold over the next five years. It also signed a deal with Google earlier this year to design and sell inference-optimized versions of its TPU chips.
Source: Motley Fool
Trading involves risk.
Risk Disclaimer
