Nvidia's quiet transformation from chipmaker into the AI industry's financier is facing its first real market stress test. As neocloud operators — the GPU-rental firms that sprang up to satisfy insatiable demand for AI compute — absorb a sharp market hit, investors and ratings agencies are scrutinizing the web of guarantees, equity stakes, and purchase commitments Nvidia has extended across the sector, CNBC reported Friday.
The stakes are no longer hypothetical. Research firm New Constructs estimates the net present value of Nvidia's off-balance-sheet liabilities reached more than $153 billion in fiscal 2026 — an increase of over $100 billion from the previous year, according to CNBC. Moody's says guarantees are the fastest-growing category of off-balance-sheet financing arrangements in 2026. For more context on this story, see our ongoing artificial intelligence updates.
The Neocloud Credit Squeeze
The pressure point is CoreWeave, the largest of the neoclouds and one of Nvidia's closest commercial partners. CNBC reports CoreWeave carries roughly $51.6 billion in total debt, including about $35 billion in notes, bonds, and loans — and that yields on its bonds maturing in 2031 and 2032 have topped 15%. In credit markets, double-digit yields on investment-grade-adjacent tech debt signal genuine doubt about repayment.
Nvidia's exposure to CoreWeave takes several forms: an equity investment plus a commitment to buy up to $6.3 billion of CoreWeave's compute capacity if the company cannot find another taker. CNBC notes Nvidia's backstops across the industry can also include guaranteed minimum resale prices or minimum hourly offtake prices for chips if a lender has to repossess assets, and revenue-sharing agreements with operators like Sharon AI.
"Nvidia doesn't want to be in the position of insurer, but they're willing to play that role while the market and the industry catches up," Warren Hosseinion, co-founder of AI infrastructure startup Compute Labs, told CNBC.
Portfolio manager Brian Stutland of Equity Armor Investments framed the worry plainly: neoclouds "will sign contracts into the future, and that all looks fine and dandy in their earnings reports, but is that contract going to be stable? Is it going to break at some point?"
The $105 Billion Ohio Backstop
Nvidia's single largest guarantee is attached to OpenAI's planned facility in Pike County, Ohio — a site that could become the largest data center in the world. CNBC reports the deal requires the use of Nvidia hardware for at least 20 years.
The mechanics, as laid out by CNBC: if OpenAI misses payments or becomes insolvent, Nvidia can take over the lease, find a new tenant, or ask the landlord, SB Energy, to sell the facility — with Nvidia guaranteeing the difference up to $105 billion if the lease or sale falls below a minimum price. Nvidia could instead pay SB Energy a lump sum. The guarantee dissolves if OpenAI posts its own collateral or achieves a sufficiently strong credit rating.
The scale is the story. A $105 billion contingent liability tied to a single customer — however remote the downside scenario — would have been unthinkable for a chip company a few years ago. It is the kind of arrangement banks structure, and it explains why analysts increasingly describe Nvidia as the AI buildout's lender of last resort.
A Balance Sheet Growing Off the Books
What makes the situation unusual is how quickly it materialized. CNBC notes Nvidia's credit backstops mostly emerged over the past six to twelve months, and that the company's last annual report, published in February, made no mention of AI cloud agreements at all. In its most recent quarterly filing, Nvidia defended the strategy: "We believe these AI cloud partners have strong customer demand and robust sales pipelines but are constrained by the large-scale infrastructure that is required to meet that demand."
Ratings agencies are playing catch-up. "These things are advanced structures that are unique, and so there is no blanket real approach you can take," David Gonzales, a vice president at Moody's, told CNBC, adding that the agency is "constantly monitoring" arrangements it does not count as debt. Gonzales said chips will likely retain "values in excess of contractual residual value amounts" near term, but conceded that how much exposure investment-grade AI players can safely absorb is "the question we're evaluating at all times every day" and "the center of all of our discussions right now."
Huang's Defense and the Bull Case
Jensen Huang pushed back this week, telling CNBC that CoreWeave does not need Nvidia's backstop and that the neocloud's cost of capital is decreasing. The bull case is straightforward: Nvidia's financial scale makes it a uniquely credible guarantor. "They'll do $700 billion in revenue next year," Gene Munster of Deepwater Asset Management told CNBC last month. "That affords them the ability to finance some of their customers."
That logic has a circular quality critics are quick to flag — Nvidia finances customers to buy chips that generate the revenue that lets Nvidia finance more customers — but it has held so far, through an extraordinary run of vendor-financed data center deals.
Insurers, Not Panaceas
One signal of how seriously Nvidia takes the risk: it is shopping for help. CNBC reports the company is actively seeking insurers to bear part of the potential loss in GPU value across its financing structures. "Insurers are not yet a complete panacea to this problem," said Tom Hoad, executive director of product innovation at insurance group Howden, "but they are probably going to turn up and be helpful in the right situations, and therefore open up some commercial transactions."
What to Watch
The neocloud wobble arrives just ahead of hyperscaler earnings, with Wall Street parsing every datapoint for cracks in AI infrastructure demand — CNBC describes the week's stock slumps, pulled IPOs, and widening credit default swap spreads as setting the stage. CoreWeave's bond prices, the pace of Nvidia's guarantee disclosures, and whether insurers actually take GPU residual-value risk are the markers to watch.
Nvidia bet that seeding its ecosystem with capital would lock in demand for its chips. The next few quarters will show whether that bet behaves like smart industrial policy — or like the opening chapter of a credit cycle.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
