Nvidia is trying to turn AI infrastructure into a financeable asset class. The company has signed letters of intent with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to mobilize more than $500 billion in third-party capital for data centers, chip factories, and power plants.
The structure matters because the AI buildout depends on expensive hardware whose long-term value is now a central debate. Nvidia says its chips can keep producing value for years. Critics argue that rapid upgrade cycles could leave lenders and investors exposed to hardware that loses value faster than expected.
What Nvidia is trying to finance
The letters of intent cover a broad infrastructure push rather than a single deal. Nvidia CEO Jensen Huang described the effort on X as a move away from one-off projects and toward repeatable financing platforms.
In Huang's framing, "AI factories" should be treated as productive infrastructure, comparable to power grids or transportation networks. The basic idea is that many AI companies need compute capacity but cannot access enough capital to build it at the scale they want.
Huang also clarified what the headline figure does and does not mean. The $500 billion is an aggregate target spread over years. It is not Nvidia revenue, not a single fund, and not a commitment to any one customer.
Nvidia did not provide terms, individual commitments, or a timeline. According to the Financial Times, which broke the deal, Nvidia's stock dropped about 1.4 percent afterward, wiping out more than $70 billion in market cap.
Why the chip guarantee is the key detail
To help the financing work, Nvidia plans to support individual projects with residual-value guarantees. If installed hardware is worth less than expected at the end of a financing term, Nvidia would cover part of the gap.
The guarantee can reach up to 25 percent of a given transaction, with each case reviewed project by project. In plain language, Nvidia is accepting some of the depreciation risk on its own products so outside capital providers can be more comfortable financing AI infrastructure.
Huang says that share is "significantly lower" than in other compute financing arrangements. He also says the credit assessment remains with the capital providers. That assessment includes the customer, demand, utilization, cash flow, and residual value.
This structure is also a response to criticism that Nvidia's financing of neoclouds and AI companies is circular. Nvidia regularly supports partners in taking on debt, and that borrowing can feed demand for Nvidia's own chips. The company is also negotiating a guarantee for a 10-gigawatt data center in Ohio leased to OpenAI.
The depreciation dispute behind the AI boom
The most direct challenge comes from investor Michael Burry, who has criticized hyperscalers' depreciation practices as "one of the more common frauds of the modern era." His argument is that GPUs become obsolete too quickly for five-to-seven-year useful lives because Nvidia has a two-to-three-year upgrade cycle.
Burry argued that depreciation would be understated by roughly $176 billion between 2026 and 2028 alone. That warning cuts to the core of AI infrastructure financing: if chips lose value faster than expected, the economics of debt-backed data centers become harder to justify.
Huang argues the opposite. He says the A100, launched in 2020, is still in commercial use six years later and that its economic lifespan stretches toward a decade. He also points to CUDA as a reason installed hardware can improve in value over time rather than simply age into irrelevance.
Rental pricing is another part of Huang's case. H100 annual contracts went from $1.70 per GPU-hour in October 2025 to $2.35 in March 2026, while B200 capacity runs between $5.30 and $7.05. For Nvidia, those figures support the argument that demand for existing and newer GPUs remains strong.
What investors and regulators are watching
The scale of AI infrastructure spending is now large enough to raise financial-stability questions. Morgan Stanley expects hyperscaler spending of $3.5 trillion between 2026 and 2028. Apollo president Jim Zelter puts the total investment need at over $8 trillion.
The Bank of England warned in its July Financial Stability Report that the pace is historically unprecedented. It also warned that a shock hitting highly leveraged AI companies could ripple through global financing conditions and trigger a credit crunch.
The report noted that banks and private credit firms have limited visibility into their indirect exposure. That point matters because AI infrastructure financing can connect chipmakers, cloud operators, data center developers, lenders, private credit firms, and large customers through several layers of obligation.
Nvidia's guarantee does not remove those risks. It changes who absorbs part of them. By backing up to 25 percent of the residual value of its own chips in selected transactions, Nvidia is signaling confidence in the long-term usefulness of its hardware while also making the financing more attractive to outside capital.
The result is a clearer picture of the AI boom's financial engine. Demand for compute is pushing companies toward vast data center, chip factory, and power plant projects. But the durability of the chips inside those projects has become one of the main questions determining how much capital the market is willing to provide.