Three Customers Account for 44% of Revenue, Nvidia Grows More Dependent on Key Clients

Nvidia's customer concentration is rising rapidly. According to reports, in the first half of this fiscal year, three customers together contributed 44% of Nvidia's total revenue, whereas two years ago no single customer accounted for more than 10%. Analysts note that these customers may include Dell or Hon Hai Technology.

Nvidia's revenue has ballooned over the past two years, but at the same time, its customer base has quietly narrowed.

On Sept. 13, The Information reported that in the first half of the current fiscal year ending in July, three Nvidia customers each contributed more than 10% of total revenue, with the three together accounting for 44%. In the previous fiscal year, the figure was two customers at 36%. Going back further to FY2023, no single Nvidia customer reached 10%.

Behind this shift is the explosive growth of Nvidia's data center business—revenue soared from about $15 billion in FY2023 to $193.7 billion in the previous fiscal year, and is on track to double again this year. The bigger the business, the deeper the reliance on top customers.

Who Are the Three Big Customers?

Nvidia has not named the three customers in public filings, but The Information's analysis suggests they may include Dell or Hon Hai Technology—both of which integrate Nvidia chips into servers and resell them to other enterprises.

Dell recently disclosed that its "AI-optimized servers" revenue doubled year over year to $16.4 billion in the second quarter, driving overall revenue growth of 58%; Hon Hai's first-half revenue also grew 35% year over year, with AI equipment sales as the main driver.

Meanwhile, Nvidia CFO Colette Kress said in February that the top five cloud providers and "hyperscale" customers—a term that may encompass Meta and SpaceX—together accounted for more than 50% of data center revenue.

Microsoft, Meta, and SpaceX are all considered major Nvidia buyers, as all three are building AI data centers at scale. Google and Amazon also purchase large volumes of Nvidia chips, but because both also heavily use their own in-house AI chips, they are unlikely to rank among Nvidia's largest customers.

The In-House Chip Concern

Rising customer concentration is itself a risk, and the fact that these large customers are also developing their own chips makes the issue more complex.

Microsoft, Meta, and SpaceX are all advancing in-house AI chip programs. Once these chips mature, these customers' demand for Nvidia products may gradually decline.

This also explains why Nvidia CEO Jensen Huang has in recent years continued to invest in emerging cloud service providers (neoclouds) and AI enterprises—CoreWeave, Nebius, and others are both Nvidia investment targets and important buyers of its chips.

Short-selling guru Michael Burry—famous for accurately predicting the 2008 subprime crisis—has in recent months publicly cited Nvidia's rising customer concentration as one of the major risks.

Accounts Receivable Surge, Payment Terms Extended

The customer concentration issue is also reflected in another dimension of Nvidia's financial data.

Nvidia disclosed in July that 70% of its accounts receivable balance came from five customers. By comparison, in January this year the figure was three customers at 56%, and at the end of FY2025 it was two customers at 33%.

The accounts receivable balance itself is also climbing rapidly. Nvidia disclosed at the end of August that its accounts receivable balance jumped 64% in the first half to $63 billion, due to "extended payment terms on large multi-quarter agreements with certain investment-grade customers."

Nvidia has extended payment terms from 45 days in the previous quarter to 60 days. But in securities filings during the same period, the company indicated it may extend them further in the future:

"Customer payments are typically due shortly after product delivery. In certain cases, for purchases by investment-grade customers, we have provided and may in the future provide longer payment terms, ranging from 90 days to up to one year, to assist customers with large data center builds."

Longer payment terms mean Nvidia receives less actual cash in a single quarter, while also tying its financial health more closely to the creditworthiness of its large customers.

Expanding to Overseas and Emerging Customers

Facing pressure from rising concentration, Nvidia is accelerating its expansion to new customers.

This week, Nvidia brought several Australian companies into its "AI Factory" program, providing chips and networking equipment to enterprises that build and operate data centers. Earlier this year, Nvidia also advanced similar projects in India and Armenia.

These moves are part of Nvidia's active cultivation of sovereign customers and emerging cloud service providers beyond its top U.S. buyers.