NVIDIA chief executive Jensen Huang has ample reason to actively court new customers by investing in various emerging cloud service providers and AI companies. Despite the company's immense success, NVIDIA's sales revenue is increasingly concentrated among a small group of buyers, some of whom may reduce their purchasing volumes in the future. This risk becomes clearly apparent when reviewing NVIDIA's public disclosure documents from recent years, which list the number of customers contributing more than 10% of total revenue. In the first half of the current fiscal year ending in July, three customers accounted for 44% of total revenue. In the prior fiscal year, two customers contributed 36% of total revenue. Going back to fiscal 2023, NVIDIA did not have a single customer exceeding the 10% revenue threshold.

As the business experiences explosive growth, NVIDIA's dependence on a handful of major clients continues to deepen. Since fiscal 2023, NVIDIA's data center segment—which tracks AI chip sales to data centers—has surged from $15 billion to $193.7 billion last year, and revenue for this business is poised to double again this year. This issue has already raised concerns among investors. Michael Burry, renowned for accurately predicting the 2008 subprime mortgage crisis, has in recent months flagged NVIDIA's rising customer concentration as a significant risk.

Of course, analyzing NVIDIA's customer concentration is inherently complex. The three major customers mentioned in the latest earnings report likely include Dell or Hon Hai (Foxconn). These companies purchase NVIDIA chips to assemble servers, which they then resell to a vast array of downstream businesses. For instance, Dell recently disclosed that its "AI-optimized server" revenue doubled year-over-year to $16.4 billion in the second quarter, driving overall company revenue up 58%. Hon Hai also stated that its first-half revenue grew 35% year-over-year, fueled by a surge in AI device sales.

But NVIDIA is also heavily reliant on certain cloud providers. In February, NVIDIA chief financial officer Colette Kress stated that the top five cloud providers and hyperscalers—likely including Meta and SpaceX—collectively account for more than 50% of data center revenue. It remains unclear how many of these five companies appear on NVIDIA's top-three customer list, which may include SpaceX, Meta, or Microsoft, all of which are aggressively building AI data centers powered by NVIDIA chips. Google and Amazon also purchase NVIDIA chips in large volumes, but both simultaneously deploy their own custom AI silicon at scale, making it less likely they rank among NVIDIA's largest customers.

Microsoft, Meta, and SpaceX are all advancing their own chip development efforts, which means they could potentially reduce their procurement from NVIDIA in the future. This explains why NVIDIA is so determined to cultivate new clients, expanding its customer pool by investing in chip buyers such as CoreWeave and Nebius, both emerging cloud service providers. NVIDIA is also directing its customer acquisition efforts overseas. For example, on Wednesday, NVIDIA brought several Australian companies into its "AI factory" initiative, supplying chips and networking equipment to firms that build and operate data centers. This year, NVIDIA has also pursued similar projects in India and Armenia, seeking to nurture more sovereign project clients and new cloud providers beyond its core U.S. customer base.

Payment Term Concerns

Despite NVIDIA's efforts to diversify its customer base, the risk of over-concentration may actually continue to rise. NVIDIA disclosed in July that 70% of its accounts receivable—outstanding payments for product sales—came from five customers. By comparison, in January, three customers accounted for 56% of total receivables; at the end of fiscal 2025, only two customers represented 33% of receivables. Notably, Dell recently disclosed that its accounts payable to suppliers surged 48% to $49.7 billion in the first half of the year through July, with a substantial portion of those obligations likely owed to NVIDIA.

Meanwhile, at least part of the growth in NVIDIA's accounts receivable stems from the company offering customers extended payment terms. At the end of August, NVIDIA disclosed that receivables had jumped 64% to $63 billion in the first half, driven by "large multi-quarter agreements with certain investment-grade customers featuring extended payment schedules." Last quarter, NVIDIA already lengthened its payment period from 45 days to 60 days. In securities filings submitted during the same period, the company indicated it may further extend these terms in the future.

"Customers generally pay shortly after product delivery. For large purchase orders from investment-grade customers, we have provided—and will continue to provide—extended payment conditions ranging from a minimum of 90 days up to one year, depending on the scale of data center construction projects, in order to support our customers' large-scale data center builds," NVIDIA stated in the filing. Relaxing payment terms carries risks: on one hand, it compresses NVIDIA's operating cash flow on a quarterly basis; on the other hand, the chip maker's business performance becomes increasingly tied to the financial health of those customers enjoying deferred payment privileges.