Nvidia CEO Jensen Huang has spent years telling investors his chips are indispensable to the future of computing.
On Aug. 10, he tried something bolder, convincing some of the most powerful money managers on Wall Street to treat those same chips like an entirely new kind of investment. One that could be bought, sold, and borrowed against much like real estate or infrastructure.
The pitch worked, at least on paper.
Several of the biggest names in finance signed on within days, and the numbers involved are large enough to reshape how the AI buildout gets paid for going forward, at a moment when questions about who ultimately foots the bill have only gotten louder.
Jensen Huang says Nvidia AI chips are now an investable asset class
Nvidia signed memorandums of understanding with Apollo Global Management, Blackstone, BlackRock, Brookfield Asset Management, Goldman Sachs, and KKR to build financing platforms aimed at mobilizing more than $500 billion in third-party capital for hyperscalers, AI labs, and enterprises building data centers, according to CNBC.
Huang told CNBC he approached only those six firms about the idea, and none of them turned him down, framing the deal as proof that chips have become a genuine asset class for the first time, calling them revenue-generating, long-lived, and fungible, according to Benzinga.
Huang kept coming back to one analogy. A GPU rack, he said, is not a PC. It generates revenue. It lasts years. It can be moved from one customer to another. That is what makes it financeable the way a building or a toll road is financeable.
Related: BofA sends strong message on Nvidia’s weak spot
Whether that argument holds up over a 30-year bond depends entirely on whether AI demand stays as strong as the room full of asset managers in that CNBC interview clearly believes it will.
Here is how the money actually moves. A customer wants Nvidia chips but does not want to put $5 billion on its balance sheet. A financing platform buys the chips, leases them back to the customer, and collects payments tied to the compute revenue those chips generate. Nvidia backstops a portion of the residual value in some arrangements, which lets the lender charge a lower rate.
The customer gets the hardware. The lender gets a long-dated asset-backed return. Nvidia books a sale. Everybody gets what they want, as long as the revenue holds.
Goldman Sachs, BlackRock, and Blackstone CEOs back the $500B Nvidia deal
Goldman Sachs CEO David Solomon called the moment a pivotal point in a historic AI investment cycle. He said the firm’s involvement reflects confidence in Nvidia’s leadership and a chance to help create a credit market backed directly by Nvidia compute, according to CNBC.
Blackstone President Jon Gray compared the opportunity to how mortgage lenders evaluate homes, arguing AI compute will increasingly be viewed as a financeable asset in its own right, reported CNBC. He pointed to usage at Blackstone’s own portfolio companies surging sevenfold this year as evidence demand is outrunning supply.
More Wall Street:
- Wall Street’s AI trade faces its biggest valuation test
- The next Wall Street shift is already underway
- Wall Street sends strong 4-word verdict on the stock market
BlackRock CEO Larry Fink went further, describing the project as the start of the next era of financial engineering, comparable to the creation of mortgage-backed securities in the 1970s, and said BlackRock had already raised some funds with plans to raise substantially more, according to CNBC.
Nvidia shares closed down 2.86% at approximately $217.55 on the day of the announcement, even as the asset managers themselves surged. Blackstone advanced 4%, while Apollo and KKR both rose more than 6%, according to CNBC.
Markets appeared to read the consortium as confirmation that balance-sheet constraints are becoming a real ceiling on organic demand growth rather than as straightforwardly bullish news for the chipmaker itself.
Nvidia circular financing, debt concerns and the BIS warning
The numbers behind the debt story are worth knowing. Morgan Stanley is projecting nearly $570 billion in global AI-linked bond issuance in 2026, more than double last year.
Nvidia joined that wave on June 15, raising $25 billion in its first bond sale since 2021. The deal was targeting about $20 billion. Investors threw $85 billion at it, so the company took more, according to Bloomberg. The chipmaker at the center of the AI trade is now borrowing like everyone else building it.
Skeptics have raised a related concern for months, arguing that some of Nvidia’s financing arrangements risk becoming circular, where Nvidia effectively helps fund the same customers who then use that money to buy Nvidia chips, complicating how cleanly outside observers can separate genuine demand from demand Nvidia itself is subsidizing.
The Bank for International Settlements flagged the resemblance to pre-2008 credit structures in its June 2026 Annual Economic Report. It specifically warned that the circular financing patterns in AI infrastructure could unwind faster than the 2008 banking crisis if sentiment reverses, as TheStreet reported.
Not every analyst views that structure as a red flag. Investor Louis Navellier has argued Nvidia is simply asserting the market dominance that comes with superior pricing power and technology, and said he has no plans to sell his position despite the circular financing narrative surrounding the company, as TheStreet reported.
The scrutiny extends beyond this one announcement. Nvidia’s separate financing talks with OpenAI, reportedly worth roughly $250 billion as a backstop for a 10-gigawatt data center in Ohio, have drawn similar circular financing criticism from investors trying to trace where the money actually originated and where it ends up, according to CNBC.
Michael/Getty Images
What the $500B Nvidia deal means for NVDA stock investors
Nvidia has increasingly positioned itself as a financier as much as a chipmaker, a shift also visible in its July 27 deal with AI lab Safe Superintelligence, in which Nvidia invested approximately $5 billion and provided priority access to its Vera Rubin compute platform in exchange for research insights that will feed into Nvidia’s future hardware development, as CNBC reported.
The demand side of the story still looks strong on paper. Goldman Sachs Research has repeatedly revised its hyperscaler capital spending estimates higher through the year, a trend Huang has pointed to directly when defending Nvidia’s growth on AI infrastructure investments.
Whether Nvidia’s chips genuinely behave like a durable, bankable asset class or simply extend the runway on an already leveraged buildout may not be settled for years.
For now, Wall Street’s biggest lenders have decided the bet is worth taking, and their willingness to underwrite it may end up mattering to the AI trade as much as any single earnings report.
Related: Nvidia CEO doubles down on AI and stock market verdict