Nvidia’s latest move has turned its chips into a financing story as much as a hardware story. The company has signed memorandums of understanding with six heavyweights — Apollo Global Management, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs and KKR — to help mobilize more than $500 billion in third-party capital for AI infrastructure. The pitch is simple and audacious: use debt and private capital to pay for data centers and Nvidia hardware, then let the chips hold value long enough to secure the financing. Shares of some of the firms moved sharply after the announcement, while critics warned the setup looks uncomfortably like the kind of leverage that can go bad fast.
The platform is aimed at financing data centers and Nvidia hardware purchases for hyperscalers, frontier AI labs and enterprises, giving customers a way to borrow without hitting their own balance sheets. That is the heart of the story: Nvidia is not just selling processors, but helping build a market structure around them. In the company’s view, the chips are no longer throwaway equipment that gets written down quickly. Instead, they can be treated more like assets that help produce revenue over time.
Jensen Huang, Nvidia’s founder and chief executive, framed the idea in unusually bold terms. “This is really the first time that technology chips have become an investable asset class. These are revenue-generating assets now. They’re productive, they’re long-lived, they’re fungible, they’re flexible,” he said on CNBC. That language matters because it tries to reprice the entire AI buildout. If chips can be financed like income-producing equipment, the market for AI infrastructure gets bigger, more liquid and, potentially, more speculative.
The core bet is on durability. Nvidia and its partners are effectively wagering that its GPUs will retain value longer than traditional hardware that depreciates fast. In that model, compute becomes collateral in a way more familiar to commercial real estate or toll roads than to ordinary electronics. That is why the financing structure has drawn such intense attention. It is not only about who wants the chips today, but whether lenders and investors believe they will still matter enough years from now to support the loans tied to them.
The headline figure is huge, but it is easy to overread it. Only memorandums of understanding have been signed so far. The $500 billion is an aggregate potential over time, not a committed fund, and no individual commitments, interest rates or deployment timetable were disclosed. That leaves plenty unresolved about how much money actually moves, how quickly it moves and what kinds of projects will qualify.
There is also uncertainty around Nvidia’s own exposure. One report says the company may provide residual-value guarantees covering up to 25% of the value of its chips in individual financing transactions, and that those would be evaluated project-by-project. Other coverage framed the idea more broadly, but the figures do not line up cleanly, so the safest conclusion is that Nvidia is potentially putting meaningful balance-sheet support behind the loans without disclosing a standardized cap across the platform. That is enough to make this more than a marketing exercise, but not enough to map the risk precisely.
For Goldman Sachs, the pitch is especially notable because it ties Wall Street’s credit machinery directly to the AI trade. David Solomon called it “We’re in a pivotal moment of a historic AI investment cycle… we’re excited for the new opportunity to create a market for credit backed by NVIDIA compute.” That is the kind of sentence that sounds like both conviction and careful positioning. It suggests the bank sees a fresh lending market forming around AI infrastructure, but it also underscores that this is still an opportunity, not a completed financing empire.
The firms on the MOU list are not casual partners. Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR all have deep experience in private credit, infrastructure and structured finance. That mix tells you what Nvidia is trying to assemble: a capital stack large enough to support the next wave of AI buildout without forcing every buyer to fund the whole thing upfront. In practical terms, that could accelerate the pace at which hyperscalers and enterprises deploy more compute.
The market seemed to like the structure at first glance. KKR and Apollo shares rose roughly 4% each the day after the announcement, while Brookfield and Blackstone gained about 3%. BlackRock rose under 2%, and Goldman Sachs was roughly unchanged. Nvidia itself rebounded about 1.1% intraday after sliding nearly 3% on the announcement day. The price action suggests investors are still sorting out whether this is a clever way to extend the AI boom or a sign the boom needs more financial engineering to keep going.
That uncertainty is part of why the story has spread so quickly. The AI trade already depends on enormous capital expenditure, and this proposal adds a new layer of leverage on top of it. If the chips prove durable and demand stays hot, lenders could have a valuable new asset class. If the hardware cycle turns faster than expected, the residual-value assumptions become much more fragile. The whole structure depends on time: enough time for the chips to earn back their financing, and enough time for the market to trust that they still have worth.
Michael Burry, who built his reputation on betting against housing before the financial crisis, wasted no time attacking the plan. “That $500 billion NVDA Wall Street stunt involves Nvidia taking 25% stakes & providing residual value guarantees on purchase of its chips. All filtered through Private Equity’s Private Credit schemes… Meet the new Boss. Same as the old Boss,” he wrote. His point was not subtle: he sees the arrangement as a reboot of familiar leverage dynamics, only this time wrapped around AI infrastructure instead of mortgage bonds.
Whether or not that warning proves right, it captures the central risk. The financing structure only works if the market believes Nvidia chips can behave like long-lived assets and not like expensive equipment that loses value the moment the next generation arrives. That is a high bar in a business where product cycles move fast and the technology race rarely pauses. The more capital that flows into AI buildout, the more tempting it becomes to smooth over that risk with credit. That is exactly what makes the current setup so marketable — and so vulnerable.
The next real test arrives with Nvidia’s fiscal second-quarter results on Aug. 26, 2026. That report will give investors a cleaner read on data-center demand and, by extension, whether this financing platform is supporting a genuine wave of purchases or simply helping stretch the timing of them. For now, Wall Street has a new question to price: are Nvidia’s chips the next great collateral, or just the latest asset class finance is trying to invent on the fly?
