On the surface, Nvidia has been generating enormous amounts of cash by selling the chips driving the artificial-intelligence boom to customers that are clamoring for them. The company generated $48.6 billion of free cash flow in the last quarter, according to a recent securities filing. Saudi Aramco, the Saudi oil monopoly, and Apple, the iPhone maker, are the only other companies in the world capable of producing so much cash.
But a close look at Nvidia’s financials suggests that a portion of that cash may be coming from Nvidia itself, analysts say, adding that Nvidia could need to increasingly fund its own customers to help them buy chips.
“Our intention is to prioritize R&D and strategic investment. Both will enable us to cultivate our ecosystem, drive market growth, and strengthen our market position,” CFO Colette Kress said on Nvidia’s earnings call this week.
Cape Fear Advisors CEO Greg Collins wrote in a recent Substack post that Nvidia has been clear that it’s investing in private AI companies to help them buy Nvidia chips. “The company replaced an independent demand market with one it is partly financing into existence — and on the call, named the pattern a competitive advantage,” Collins wrote. “The demand-source mix has moved toward categories whose growth depends on Nvidia’s continued capital deployment.”
As the kingmaker of the AI boom, Nvidia has invested in its own customers, strategic partners and competitors. It has stepped into the role of venture capitalist by participating in funding rounds for some of the hottest AI startups. Recent deals include the right to purchase up to $2.1 billion in shares of neocloud Iren and up to $3.2 billion in shares of fiber-optics company Corning.
The move has bolstered Nvidia’s full-stack AI portfolio and put the company’s cash to work. But writing these checks is costly and fuels a circular investment cycle that makes it difficult to determine whether demand for Nvidia’s chips is economically organic or inflated by its own capital, according to analysts. One line item on Nvidia’s statement of cash flows in particular could be relevant to Nvidia’s available capital: purchases of nonmarketable equity securities. These purchases have been steadily growing over the past few quarters.
In its last quarter, Nvidia used $18.58 billion of cash to purchase nonmarketable equity securities — illiquid assets not traded on exchanges— a securities filing showed. In the same quarter a year earlier, Nvidia spent just $649 million on purchases of nonmarketable equity securities. Nvidia’s rapid increase over the last year of such purchases resulted in the company owning $42.3 billion on nonmarketable equity securities at the end of the April quarter, up from just $3.2 billion a year ago. As a result, Nvidia’s capital intensity may be much higher than what it looks like on paper, according to Collins.
While Nvidia didn’t specifically say in its regulatory filings which private companies make up its portfolio of nonmarketable equity securities, the company disclosed in its annual filing from January that it invested $17.5 billion in private companies and infrastructure funds, including AI model makers, and that “many of these investments are illiquid and non‑marketable.”
Collins believes the line item encompasses private companies Nvidia has invested in, such as Anthropic, OpenAI and xAI, which was acquired by SpaceX. The value of these investments is held at cost basis, a securities filing said, unless an observable event such as a new financing round results in a price change.
Representatives from Nvidia did not immediately respond to a MarketWatch request for comment.
Since the $18.58 billion that Nvidia used to purchase stakes in private companies is not classified as capital expenditures, Nvidia’s headline free cash flow of $48.6 billion for the quarter presents as if the money is still available for distribution or organic reinvestment, Collins told MarketWatch over email. However, in reality, the billions that Nvidia has invested in its AI ecosystem “will only convert back to cash if and when those companies IPO or get acquired at favorable marks,” Collins added. Nvidia reported $1.8 billion in capital expenditures in its last quarter, which means that for every dollar the company spends on its own assets, it deploys roughly 10 dollars into the overall AI ecosystem, Collins pointed out.
The private companies that raised $18.58 billion from Nvidia in its last quarter are using that large amount of cash for something, like buying Nvidia chips.
With OpenAI, Anthropic and SpaceX all planning to go public sometime this year, that could require Nvidia to reclassify its nonmarketable equity securities as marketable securities at fair value — essentially liquid assets that can fluctuate wildly in price. Nvidia has benefited from making itself a central part of the AI ecosystem, but that means that the company has tethered its own financial performance to that of its partners and customers.
Seaport analyst Jay Goldberg sees Nvidia’s extensive investments as part of the company’s strategy to shield itself from growing competition. He initially flagged this trend following the company’s third-quarter results in November, when the company reported $8.2 billion of nonmarketable equity securities. “This is fairly small in the context of its operations, but it grew considerably this quarter, tripling over the past 9 months, and is likely to grow even more next year,” Goldberg wrote at the time. Goldberg is the only analyst tracked by FactSet with a sell rating on Nvidia’s stock.
In an email to MarketWatch, Goldberg also highlighted the disclosure of $30 billion in multiyear cloud service agreement commitments, which represent the backstops Nvidia provides to neoclouds like CoreWeave. Last September, Nvidia signed a deal with CoreWeave, for example, agreeing to purchase any unsold capacity through 2032, providing credit support for CoreWeave to borrow money from Wall Street lenders and continue growing its business.
Comments