The companies that made Nvidia the most valuable name in artificial intelligence are increasingly trying to build their way around it.
According to a report published by Stocktwits and carried by Yahoo Finance, titled "From OpenAI To Google To Amazon: Nvidia's Biggest Customers Are Becoming Its Biggest Rivals — The Custom Silicon Race Explained," three of Nvidia's largest buyers — OpenAI, Google and Amazon — are pushing into custom silicon of their own.
The dynamic is unusual. Nvidia's graphics processors have been the default engine for training and running large AI models, and the firms buying them in bulk are the same ones now designing chips tailored to their own workloads. That makes them customers and competitors at once.
The logic behind the shift is straightforward. Buying general-purpose chips from a single dominant supplier means paying that supplier's prices and waiting in that supplier's queue. Designing your own silicon — tuned to the specific models you run — offers a path to lower costs, more predictable supply, and hardware shaped around your software rather than the other way around.
The two source items here are the same report distributed through different outlets, so the underlying reporting is singular rather than corroborated across independent newsrooms. Neither item, as summarized, provides chip specifications, timelines, or volume figures.
What is clear is the direction of travel: the customer base that underwrites Nvidia's dominance has an incentive to erode it, and several of its biggest members are acting on that incentive.
This matters because the cost and availability of AI chips shape what AI products get built, who can afford to build them, and how much of the industry's spending flows to a single company.