Big Tech is spending at a scale that would make a Vegas casino blush.
And the real number isn’t what shows up on the balance sheet.
A Financial Times investigation just pulled back the curtain on a financial structure that lets Silicon Valley bet hundreds of billions on AI without telling investors the whole story.
The Guarantee Game Nobody Is Talking About
Technology companies issued as much as $300 billion in guarantees tied to AI data centers and chips over the past year, according to a Financial Times analysis.
These aren’t conventional loans. They’re something called residual value guarantees, and the way they work is almost elegant in how much they hide.
A technology company promises that chips, data centers, or other infrastructure will retain a minimum value in the future. If those assets get sold or leased for less than the guaranteed amount later on, the tech company that issued the guarantee could be on the hook for covering part of the gap. But because the debt technically belongs to a separate financing vehicle rather than the tech company itself, the full exposure never lands on the company’s own balance sheet.
In other words, the liability is real. The accounting treatment makes it disappear.
S&P Global Ratings warned that special-purpose vehicles, vendor financing, backstop agreements, and residual value guarantees can “increase debt-like exposure, obscure risk, and create financial interdependencies.”
That’s a credit ratings agency telling investors the picture they’re looking at is incomplete.
Who Is Actually Doing This and How Much
Meta moved first and at scale, providing a roughly $28 billion guarantee supporting its Hyperion data center joint venture with Blue Owl in Louisiana. The Hyperion project is expected to span roughly 4 million square feet and eventually consume enough electricity to power about 1.5 million homes. That guarantee helped the project raise roughly $27 billion in debt at borrowing costs only modestly above Meta’s own bonds, while leaving relatively little exposure directly recorded on Meta’s balance sheet.
Then Broadcom joined the party.
Broadcom took on roughly $29 billion in exposure in June as part of a deal involving chips that will ultimately be leased to Anthropic. The financing arrangement developed alongside Google as part of a broader effort to supply computing infrastructure to the AI company. Broadcom told investors it believed the guarantees were unlikely to be triggered because of the profitability of major AI developers and the expected value of the assets backing the financing.
Nvidia has increasingly embraced similar arrangements. The chipmaker said it could provide residual value support covering as much as 25% of certain infrastructure financing deals being assembled alongside Goldman Sachs and other major investment firms seeking to mobilize more than $500 billion for Nvidia-powered AI projects. Nvidia also provided roughly $105 billion in guarantees to SB Energy, a SoftBank subsidiary developing a large Ohio data center campus for OpenAI, according to the Financial Times. That agreement allows Nvidia to avoid immediately recording a liability tied to the project. OpenAI’s leases are expected to begin in 2028, while the data center campus is expected to use Nvidia hardware for decades.
And these structures specifically lower borrowing costs because lenders get some protection if rapidly advancing technology causes chips or data centers to lose value faster than expected.
Which sounds like a reasonable deal — until the technology does exactly that.
What Happens When the Boom Slows Down
Technology companies could face greater exposure if demand for computing power disappoints, companies build more data center capacity than customers actually need, or newer chips make existing hardware obsolete faster than expected.
Analysts at CreditSights identified the specific danger: Nvidia, by issuing residual value guarantees, is effectively “writing a put.” Their report put it plainly: “This is pro-cyclical and exacerbates boom-bust potential.” The guarantee is nearly costless while the boom runs hot. But if the market turns and customers start defaulting while hardware values fall, the exposure becomes very real, very fast.
S&P Global analysts estimate how much the guaranteed value of an asset exceeds what the infrastructure could fetch during a distressed sale, and they can add that difference to a company’s adjusted leverage. Credit rating agencies are already attempting to account for some of that risk. But the accounting is catching up to the exposure, not leading it.
The structures allow Big Tech companies to support massive financing without immediately recording the entire obligation as debt. Those entities own the chips or data centers, while the tech companies supply the guarantee and the credibility to make the debt marketable.
It is a brilliant system for concentrating risk inside entities that are opaque to ordinary investors while the good times roll. The same companies that built their reputations on rock-solid balance sheets are now running contingent liabilities that dwarf what appears on their own financial statements.
Big Tech has done this before with data, with user attention, with market dominance. Now these companies are doing it with debt. They borrow the credibility of their own credit ratings to make off-balance-sheet obligations cheap to finance, and then they tell investors the guarantees are unlikely to be triggered.
Broadcom said exactly that. Investors should pay attention to the word “unlikely.” It is not “impossible.” It is not “zero.” It is the financial equivalent of “trust us.”
And the same companies issuing these guarantees — Meta, Google, Nvidia, Broadcom — are the ones that spent years censoring conservatives over COVID policy, the 2020 election, and the events of January 6. These are not neutral actors operating purely in the public interest. They are corporations that have demonstrated a willingness to use their platforms and power to benefit one political coalition over another. Now they want the financial system to take their word for it that hundreds of billions in contingent obligations are nothing to worry about.
Anthropic CEO Dario Amodei has warned separately that AI could wipe out half of all entry-level white-collar jobs and drive unemployment to levels not seen in generations. That warning — from inside the industry — deserves more attention than it gets. The same infrastructure buildout that generates these hidden financial guarantees is the physical foundation for AI systems that will automate the jobs of working Americans. The electricity bills, the grid strain, the financial complexity, the job displacement — all of it flows downstream to ordinary people. The profits flow to coastal headquarters.
The question investors and lawmakers should be asking is simple: if these guarantees are as unlikely to be triggered as Big Tech claims, why does S&P Global feel the need to add them to adjusted leverage calculations? Why does a credit ratings agency issue a formal warning about debt-like exposure being obscured? Why do analysts at CreditSights describe the structure as pro-cyclical and boom-bust?
Because the risk is real. The exposure is real. The only thing that isn’t fully real — yet — is the accounting treatment.
Big Tech built a $300 billion off-balance-sheet liability for AI infrastructure and buried it in the footnotes. The Federal Reserve’s long history of enabling exactly this kind of financial complexity — cheap money creating the conditions for creative accounting and leverage accumulation — is part of the backdrop here. When rates were near zero, borrowing was cheap and the incentive to obscure debt was lower because debt itself was cheap. Now the structure of these deals hides the true cost. And when the AI boom eventually slows, the guarantee becomes the most expensive paper in the portfolio.
The Enron comparisons are already circulating in financial press. That should tell you something.
But Big Tech will keep building, keep guaranteeing, and keep describing the risk as remote — right up until it isn’t.
Sources: Daily Caller, “Big Tech’s AI Boom Is Hiding Hundreds Of Billions In Financial Risk,” September 25, 2026; Financial Times analysis cited therein; S&P Global Ratings; CreditSights analyst report; Broadcom investor communications.
