TL;DR — Four customers account for 61% of Nvidia’s revenue, and Nvidia won’t name them in its filings. That concentration is the real risk sitting underneath Nvidia’s growth numbers, not the chip roadmap.
Nvidia’s most recent quarterly SEC filing lists four customers, each buying more than 10% of the company’s total revenue. Combined, they account for 61% of $57 billion in quarterly sales. Nvidia calls them Customer A, Customer B, Customer C, and Customer D. It does not say who they actually are, and it isn’t required to. That’s a real business generating real revenue, sold to a buyer base so concentrated that four unnamed accounts now move the majority of the company’s top line.
The Concentration Is Getting Worse, Not Better
Trace it quarter by quarter through fiscal 2026. In the first quarter, four unnamed customers accounted for 54% of $44.1 billion in sales. By the second quarter, Customer A alone represented 20% of revenue and Customer B another 15%. By the third quarter, the top four — Customer A at 22%, B at 15%, C at 13%, D at 11% — accounted for 61% of $57 billion in quarterly sales. Revenue concentration among Nvidia’s biggest buyers has climbed from roughly 36% to 61% over the course of a single year. This isn’t a company whose growth is slowing and leaning harder on a few loyal customers to prop up the top line. Nvidia closed fiscal 2026 with $216 billion in revenue, up 65% year over year. The business is accelerating and getting more concentrated at the same time, which is the less comfortable version of that combination.
Why Nvidia Won’t Name Names
Standard accounting disclosure rules require a company to report when a single customer accounts for more than 10% of revenue. They don’t require naming that customer unless a separate legal or contractual obligation forces it. Nvidia is following the letter of the rule, not hiding something illegal. But the practical effect is that the market has to guess. The working assumption among analysts is that Customers A through D are some combination of the largest hyperscalers — the same companies whose capex guidance we’ve tracked all season, Microsoft, Meta, Amazon, and Alphabet among the most likely candidates given who’s actually spending at this scale. None of that is confirmed by Nvidia. It’s inference built on public capex numbers from a different set of earnings releases.
What Concentration Actually Means for Risk
If four buyers represent 61% of revenue, a meaningful slowdown from just one of them moves Nvidia’s total revenue in a way that’s disproportionate to what “one customer” usually implies. And the exposure doesn’t stop at Nvidia. We’ve already written about how Nvidia holds roughly 60% of TSMC’s advanced packaging capacity — which means a slowdown at the hyperscaler level doesn’t just hit Nvidia’s income statement, it ripples straight through to the company building the chips underneath it. Concentration risk compounds across the supply chain rather than staying contained at any single link.
That chain matters more now than it did earlier this year. If financing conditions stay tighter for longer than the market originally priced in — the scenario we laid out after the Fed’s July meeting left the door open to a rate hike rather than a cut — hyperscaler capex is exactly the kind of large, discretionary, debt-and-equity-funded spending that gets re-examined first when the cost of capital rises. A pullback at any one of Nvidia’s four largest customers wouldn’t just dent one earnings report. It would show up simultaneously in Nvidia’s revenue, TSMC’s packaging bookings, and the broader AI infrastructure story this entire earnings season has been built around.
The Read for Investors
None of this means Nvidia’s growth is fabricated or the AI buildout is a mirage. Concentration among a handful of buyers is close to structurally inevitable when the market for a technology is a small number of hyperscalers with the balance sheets to spend at this scale. But 61% of revenue sitting with four unnamed accounts is a real single-point-of-failure profile, distinct from and additive to any separate question about whether AI capex keeps growing at all. Treating Nvidia’s growth rate and its customer concentration as two different risks, rather than one, is the more honest way to model what actually happens if any single hyperscaler changes its mind about the pace of spending.
Sources: NVIDIA Corp Form 10-Q filings, fiscal year 2026, Q1 through Q3; SEC EDGAR customer concentration disclosures. Third Pole Markets holds no position in Nvidia as of publication. This is not investment advice — see our About page for our full disclosure policy.
Nvidia’s customer concentration disclosures are in its 10-K filings on SEC EDGAR, alongside Nvidia’s investor relations page.
