Nvidia is at the center of the artificial intelligence boom, and its latest earnings report confirms this in spectacular fashion.

On Wednesday, the chipmaker reported a record-breaking revenue of $46.7 billion for its second quarter—a staggering 56% year-over-year increase. However, a deeper look into the company’s filing with the Securities and Exchange Commission (SEC) reveals a surprising and potentially risky concentration of this success: nearly 40% of that massive revenue came from just two unidentified customers.

The two mystery customers driving Nvidia’s growth

The SEC filing is clear but intentionally vague. It states that a single customer, referred to as “Customer A,” represented 23% of total Q2 revenue. Another entity, “Customer B,” accounted for 16% of Q2 revenue. Together, these two mega-buyers were responsible for 39% of Nvidia’s sales in the quarter. The filing also notes that four other customers contributed 14%, 11%, another 11%, and 10% of revenue, respectively, highlighting a dependency on a small group of major players.

So, who are these mystery customers?

The filing specifies that these are all “direct” customers—such as original equipment manufacturers (OEMs), system integrators, or distributors—who purchase chips directly from Nvidia. This means it is unlikely that a big cloud provider like Microsoft, Oracle, Amazon, or Google is secretly Customer A or Customer B.

However, that doesn’t mean the cloud giants aren’t behind this spending. These indirect customers purchase Nvidia chips from the direct customers. Nvidia’s Chief Financial Officer, Nicole Kress, shed light on this by stating that “large cloud service providers” accounted for 50% of Nvidia’s data center revenue, which in turn represented 88% of the company’s total revenue.

In simple terms, while a distributor or system integrator might be the “direct customer” on paper, they are likely buying these billions of dollars worth of chips on behalf of a hyperscaler like Microsoft or Google to build out their AI data centers.

Concentration risk vs. guaranteed growth

This heavy reliance on a small number of buyers presents a classic business dilemma. Gimme Credit analyst Dave Novosel told Fortune that this “concentration of revenue among such a small group of customers does present a significant risk.” If one of these key customers were to reduce their spending or switch suppliers, the impact on Nvidia’s bottom line would be immediate and severe.

However, there’s a powerful counter-argument, which Novosel also points out. The good news for Nvidia is that “these customers have bountiful cash on hand, generate massive amounts of free cash flow, and are expected to spend lavishly on data centers over the next couple of years.”

The customers driving this demand—the Microsofts, Googles, and Amazons of the world—are locked in an existential AI arms race. They cannot afford to slow down their spending on AI infrastructure if they want to remain competitive. This provides Nvidia with a relatively stable and predictable stream of massive orders, mitigating the traditional risks associated with customer concentration.

A narrow foundation for a booming industry

Ultimately, Nvidia’s customer concentration is a direct reflection of the current state of the AI industry itself. While AI technology is spreading everywhere, the race to build the foundational models and the massive data centers required to run them is being fought by a very small number of incredibly wealthy “hyperscaler” companies.

Nvidia’s record-breaking success is undeniable, but the revelation that so much of it is built on the spending of just a handful of entities is a crucial piece of context. It highlights both the incredible scale of the AI boom and the surprisingly narrow foundation upon which it is currently being built. For investors and industry watchers, the health and spending habits of these few mystery customers will be the most important factor in determining Nvidia’s future prospects.

Featured image