Nvidia Wants Wall Street to Treat Computer Chips Like Toll Roads

Nvidia announced on August 10 that it signed non-binding memorandums of understanding with six major financial institutions, Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR, aiming to mobilize more than $500 billion in third-party capital for AI infrastructure over time, according to Nvidia's own announcement. CEO Jensen Huang called it a milestone: "We began by building chips; today, we are helping create a new class of productive, investable infrastructure: AI factories."

The mechanism is designed to let outside investors, not Nvidia itself, fund the data centers, power infrastructure and other buildout AI companies need, keeping most of that exposure off Nvidia's own balance sheet. "This is really the first time that technology chips have become an investable asset class," Huang told CNBC, arguing that because Nvidia hardware is broadly adopted and transferable across customers, lenders can underwrite it more like durable infrastructure than typical fast-depreciating computer equipment.

What makes this worth understanding in real estate and infrastructure terms specifically is exactly who signed on. These are the same firms that have spent the past decade building deep expertise underwriting real estate and infrastructure debt, evaluating whether a building's rental income or a toll road's traffic revenue can reliably support decades of loan payments, factoring in tenant credit quality and what the asset will still be worth once the loan comes due. Applying that same framework to semiconductors is a genuine leap. Real estate assets typically hold or appreciate in value over long horizons; GPU chip generations, historically, have depreciated fast as newer, faster hardware supersedes them every few years. KKR's co-CEOs, Joe Bae and Scott Nuttall, described the logic behind their participation in terms that make the analogy explicit: compute "has become a critical infrastructure asset," combining "Nvidia's accelerated computing platform with KKR's long-term capital, infrastructure expertise, and capital markets capabilities," according to BigGo Finance's coverage of the deal.

Credit markets haven't fully embraced that framing yet. Reported stock reactions to the announcement varied depending on exactly when they were measured, but the more telling signal came from Nvidia's five-year credit default swap spread, a market-based gauge of perceived credit risk, which rose to roughly 77 to 82 basis points following the news, according to both BigGo Finance and analysis from explainx.ai. That's not a default signal on its own, but it reflects something genuinely new: credit markets increasingly view Nvidia's own creditworthiness as entangled with the financial health of the customers buying its chips, a dynamic that didn't exist when Nvidia was simply a company that sold hardware for cash.

That entanglement sits at the center of a broader "circular financing" concern that's followed Nvidia's dealmaking all year, and this new platform isn't happening in isolation. Nvidia has separately committed up to $100 billion in investment to OpenAI, expanded a partnership with South Korea's SK Group worth more than $500 billion in mutual business, and made a substantial investment in Safe Superintelligence, according to Fortune. OpenAI itself has committed to $250 billion in cloud services from Microsoft while deploying billions of dollars in AMD chips and becoming one of AMD's largest shareholders. No single deal in that web is improper on its own, but together they describe a market where the same small group of companies are simultaneously customers, investors and suppliers to one another, making any individual company's credit risk harder to evaluate through traditional, standalone balance-sheet analysis.

Investor Michael Burry, known for his early public warnings ahead of the 2008 financial crisis, has been considerably sharper in his criticism, calling the financing push a "Wall Street stunt" and drawing comparisons to circular financing structures seen in past corporate collapses, according to Forbes. Nvidia pushed back directly in a company blog post addressing the concern head-on: "We are bringing independent, long-term institutional capital into the AI infrastructure market... The capital providers independently underwrite each project, including the customer, demand, utilization, cash flow and residual value. NVIDIA provides the platform; the investors make independent financing decisions."

One structural detail complicates that fully independent framing. Nvidia has reportedly agreed to provide a residual value support mechanism covering up to 25% of asset value in certain cases, according to BigGo Finance's reporting on the deal terms. That means Nvidia isn't entirely absent from the risk chain, even though the headline framing emphasizes third-party capital and independent underwriting; the company is still backstopping part of the downside if its own hardware ends up worth less than expected by the time loans come due, the same kind of guarantee a real estate sponsor might offer a lender to get a deal financed on better terms.

This deal also builds on, and substantially expands, an existing effort rather than starting from nothing. In 2024, BlackRock, Microsoft and the UAE's MGX investment vehicle formed the AI Infrastructure Partnership specifically to bankroll data centers, with Nvidia supporting that coalition from the sidelines. This week's announcement is best understood as a much larger version of that same underlying idea, applying real-asset financing logic to AI infrastructure at roughly five times the scale of the earlier effort.

It's worth being precise about what's actually committed here versus what's aspirational. The $500 billion figure describes a target for capital to be mobilized "over time," not money that's already been raised or deployed. The agreements themselves are non-binding memorandums of understanding; final agreements, specific terms for individual projects, and the actual pace of capital deployment remain to be worked out. Fortune's own reporting notes plainly that "details of the arrangements, like the extent of each deal, are still unknown."

For ordinary investors, the practical relevance runs deeper than headline technology news. Much of this new capital pool, given its source, banks, insurers and alternative asset managers that also run pension and retirement-linked investment vehicles, is likely to touch retirement savings and insurance products indirectly, even for people with no direct exposure to AI or technology stocks. Anyone invested in funds managed by these six firms now carries some degree of exposure to how well semiconductor-backed lending actually performs over the coming years, a genuinely untested proposition regardless of how confidently either side of the circular-financing debate states its case.

The clearest things to watch next are whether Nvidia and its six partners move from these preliminary memorandums to signed, binding final agreements, whether Nvidia's credit default swap spread and broader AI infrastructure debt pricing continue reflecting elevated skepticism or ease as more details emerge, and whether other major banks or asset managers beyond this initial group of six announce comparable AI compute financing platforms of their own.