Microsoft to rely more on homegrown AI chips, reducing dependence on Nvidia

Microsoft will ramp up use of its homegrown artificial intelligence chips, ending years of heavy reliance on Nvidia's graphics processing units, according to CNBC. The shift is aimed at bolstering Microsoft's AI stack and expanding its margins.

The anchor fact: Microsoft is moving from buyer to builder. By leaning more on in-house silicon, the company reduces how much of its AI spending flows to Nvidia, the supplier it has depended on for GPUs.

Interpretation: if Microsoft can run more AI workloads on its own chips, it controls more of its cost structure. Every workload shifted away from purchased GPUs is spending that stays inside Microsoft's own stack, which is the mechanism behind the margin argument CNBC cites. That is analysis, not a reported figure, and no timeline or volume for the transition has been quantified.

For Nvidia, the read is more cautious. A major customer committing to its own chips signals a potential long-term dent in demand from a key buyer. Again, this is interpretation of the direction of travel; the reported facts do not size the impact on Nvidia's revenue.

The setup pits two names against each other. Microsoft's story is margin expansion and a more self-sufficient AI platform. Nvidia's is a customer that is increasingly also a competitor.