Amazon steps up its in-house AI chip push, adding pressure to Nvidia’s data-center dominance

The race to build out artificial intelligence data centers is forcing major cloud operators to rethink how they buy compute. A report published by Yahoo Finance through 247Wallst argues that Amazon is taking a more direct approach by selling its own AI chips, a move portrayed as a competitive response to Nvidia’s position at the center of the modern AI compute supply chain.

The report places the urgency in broader infrastructure spend. It says the AI boom has driven one of the largest infrastructure buildouts in technology history, with the largest cloud providers reportedly on pace to spend more than 750 billion dollars on AI infrastructure in 2026. That scale matters because chip demand is typically bundled with the rest of the data-center build, from networking and power delivery to rack-level systems.

Within that context, the report frames Amazon’s chip strategy as more than a procurement decision. By “selling its own AI chips,” Amazon is depicted as offering an alternative path to the market that can reduce customers’ dependence on third-party accelerators, at least in some deployments where the chip, software stack, and cloud services are integrated.

The story’s competitive thrust is aimed at Nvidia’s data-center hardware. Nvidia’s accelerators are widely used to train and run AI models, and the report implies that when a cloud provider builds and commercializes its own silicon, it can capture more of the economics and control more of the technical roadmap inside its platform.

What the report does not spell out, however, is critical detail. It does not identify the specific Amazon chip product names, the customers or contract terms associated with “selling,” or whether the chips are positioned primarily for Amazon’s own workloads, for broader third-party cloud use, or for an external marketplace. It also does not quantify how much of Amazon’s AI compute demand is expected to shift away from Nvidia in the near term.

Even so, the underlying dynamic highlighted by the report aligns with a broader pattern in the technology sector: as AI infrastructure spending climbs, the largest buyers gain leverage. A cloud operator that controls chip design and deployment can tailor performance and cost to its workloads, potentially improving margins and insulating itself from supply constraints that can affect vendor ecosystems.

Still, the exact impact on Nvidia’s revenue mix remains unclear based on the information presented. The report offers a directional claim about intensified competition, but without disclosed financial figures, market-share estimates, or timelines for any substitution of Nvidia hardware with Amazon silicon.

For investors and industry watchers, the next indicates to watch are concrete: whether Amazon expands the availability of its AI chips beyond its own platform, whether it publishes performance and pricing comparisons versus mainstream accelerator offerings, and whether customers publicly describe any migration to in-house silicon for training or inference workloads.