The concentration has been building in Nvidia’s books for years, and the latest filing makes the trend hard to ignore. In the first six months of fiscal 2027, Nvidia’s three largest direct customers accounted for 16 percent, 15 percent, and 13 percent of revenue respectively, a combined 44 percent. Two years earlier, no single customer reached 10 percent. The numbers describe a company whose fortunes now rest on a shrinking circle of buyers.
The shift reflects how the AI market has consolidated. When Nvidia’s chips were bought by a broad base of gamers and small businesses, its revenue was spread across thousands of accounts. The AI buildout changed that. A handful of hyperscalers and AI labs now place orders large enough to move the company’s quarterly results, and their decisions have become the single most important variable in Nvidia’s outlook.
The accounts receivable data tells the same story with a sharper edge. As of July 26, 2026, Nvidia’s five largest direct customers accounted for 70 percent of accounts receivable, up from 56 percent on January 25. The jump means the company is increasingly dependent on a small group for both orders and payment. That is a different kind of exposure, one that investors are only beginning to price.
The figures appeared in a filing reported on September 14, and they arrive at a moment when the durability of AI spending is the market’s central question. Nvidia has been the primary beneficiary of the buildout, and its valuation rests on the assumption that the hyperscalers keep buying. A disclosure that highlights how few buyers matter puts that assumption under a spotlight.
The risk is not abstract. Some of Nvidia’s largest customers are simultaneously developing their own AI chips, a hedge against the very dependence the filing describes. If those efforts succeed, the customers that drive Nvidia’s revenue could become competitors, or at least buy less. The company has acknowledged the shift in its public statements, framing it as a normal part of a maturing market.
Analysts said the concentration is a double-edged sword. On one side, it means Nvidia’s biggest customers are committed enough to place enormous orders, which is a vote of confidence. On the other, it means a change of heart at any one of them would ripple through the income statement immediately. The margin for error shrinks as the customer list does.
Nvidia’s defense has been its lead in performance and software. The company argues that its chips and the ecosystem around them remain ahead of what customers can build themselves, and that the cost of switching is high. The counterargument, made by some investors, is that the largest customers have both the capital and the incentive to try anyway.
The accounts receivable concentration is the detail that worries credit analysts in particular. A company whose top five customers owe 70 percent of its receivables is exposed to payment risk in a way a diversified one is not. For Nvidia, whose customers are mostly well-capitalized technology giants, that risk is muted, but it is not zero, and it is now quantified in the filings.
The disclosure is also a window into who is buying. Nvidia does not name the customers, but the profile, large orders for data center accelerators, points to the usual suspects: the cloud providers and frontier labs that dominate AI. Their names are not in the filing, but their fingerprints are all over the revenue line.
The customer concentration also has implications for how Nvidia is valued. A company with a diversified customer base can be priced on its own fundamentals. A company whose revenue depends on a handful of buyers is priced on their fundamentals too, which means Nvidia’s multiple is partly a bet on the cloud providers and AI labs that dominate its order book.
The self-driving chip efforts among those customers add a temporal dimension to the risk. Even if those efforts take years to bear fruit, their existence changes the negotiating dynamic. A customer that is building its own silicon has a credible fallback, and credible fallbacks shape prices and terms. Nvidia’s dominance is real, but it is now contested at the top of its customer list.
For investors, the filing is a prompt to look past the revenue line. Nvidia’s growth is extraordinary, but the quality of that growth now depends on a handful of relationships. The company’s job is to keep those relationships intact while the market figures out what they are worth.
What happens next depends on the customers’ own arithmetic. They buy Nvidia chips because the chips make their AI businesses possible. If those businesses keep growing, the orders keep coming, and concentration becomes a sign of strength. If they stall, the same concentration becomes a vulnerability. The filing gives investors the numbers to make that judgment. It does not tell them which way the judgment will go.
