US Tech Giants' Hidden AI Debt Surges 90% in Six Months to $3.2 Trillion

According to a report by Nikkei Asia, US tech giants' "hidden debt" in the AI sector has surged 90% over the past six months, reaching a total scale of $3.2 trillion. The figure sparked a discussion on Hacker News with 276 points and 122 comments, signaling that the market is growing increasingly sensitive to the financial risks behind the AI arms race.

Where the $3.2 Trillion in "Hidden Debt" Comes From

So-called "hidden debt" typically refers to liabilities that do not appear directly on the main columns of the balance sheet but substantively constitute corporate financial commitments — such as long-term leases, purchase commitments, joint venture guarantees, and financing arrangements conducted through special purpose entities.

In the AI infrastructure race, tech giants need to procure computing chips at scale, build data centers, and lock in long-term power supply. These investments are often locked in through multi-year contracts or off-balance-sheet structures. As the pace of AI capital expenditure accelerates, such commitments have nearly doubled in just six months, climbing to $3.2 trillion.

Off-Balance-Sheet Liabilities is the core accounting concept for understanding this topic. Under the disclosure frameworks of US GAAP and IFRS, the main balance sheet only lists liabilities that meet specific recognition criteria, while operating leases, unfulfilled purchase commitments, letter-of-credit guarantees, and most obligations of variable interest entities (VIEs) are typically disclosed only in the notes. In recent years, tech giants have held data centers through "build-lease-leaseback" arrangements and joint-venture SPVs with infrastructure funds, keeping enormous capital commitments in the notes rather than on the main statements. This makes the same investment "invisible" in commonly used metrics such as leverage ratios and net debt/EBITDA — yet as long as the contract is irrevocable, its economic substance is no different from debt. When analyzing such companies, lease commitments and purchase obligations must be discounted to present value and folded into adjusted net debt to restore the true leverage level.

The "Off-Balance-Sheet" Tendency of AI Capital Expenditure

Shifting AI infrastructure investment into off-balance-sheet structures has practical motivations. Data center construction has long cycles and large individual amounts; by sharing costs through leases or cooperative entities, companies can maintain expansion speed while easing pressure on current-period financial statements, and avoid pulling down free cash flow metrics all at once.

But the cost of such arrangements is reduced financial transparency. The statements investors see may understate a company's true long-term solvency obligations — and once AI demand growth slows or financing costs rise, these commitments will become visible all at once.

Free Cash Flow (FCF) refers to the balance of operating cash flow minus capital expenditure, and is a key metric the market uses to gauge a tech company's "self-generated cash capability" and valuation reasonableness. If data center construction spending is counted directly as capital expenditure, FCF would be significantly depressed, in turn hitting valuation models based on FCF discounting and shareholder return expectations. By undertaking construction through long-term leases or joint venture entities, companies can convert one-time heavy capital expenditure into annual rent or capital contribution commitments, making current-period FCF appear healthier. But this treatment does not eliminate cash outflows — it only changes the timing and location of reporting. Investors who look only at FCF and ignore lease and purchase commitments in the notes can easily overestimate a company's actual cash-generating ability.

Why the Market Is Starting to Focus on This Bill

The high level of discussion on Hacker News reflects a widespread concern: whether the valuation expansion driven by the AI narrative can sustainably detach from cash flow reality. When hidden debt inflates at a 90% growth rate, the nature of the question shifts from "how aggressive is capital expenditure" to "is the debt structure manageable."

Worth watching in the discussion are: the maturity structure of these liabilities, their interest rate sensitivity, and the chain reactions if AI investment returns fall short of expectations. For tech giants, the real test may not be whether they can keep buying chips, but whether they can find a balance between debt costs and AI revenue.

Maturity structure and interest rate sensitivity are two technical dimensions for assessing the risk of this hidden debt. Maturity structure refers to the distribution of liability due dates: if a large number of commitments mature within the next 2–3 years while AI revenue materializes later, refinancing pressure and cash flow mismatches will emerge; if maturities are longer and locked in at fixed rates, the impact of short-term rate fluctuations is limited. Interest rate sensitivity measures the effect of rising financing costs on debt-servicing capacity — in a high-rate environment, renewing leases or project financing could significantly raise interest expenses. For off-balance-sheet commitments, it is also important to watch whether contracts contain price adjustment clauses, take-or-pay obligations, and early termination penalties — these terms determine whether a company can exit at low cost when demand slows.

Implications for the Industry and Investors

The $3.2 trillion in hidden liabilities is a reminder that evaluating AI companies requires more than revenue growth and model capabilities — one must look through off-balance-sheet structures to see true leverage.

For upstream players in the supply chain, this means counterparty risk on long-term contracts for computing power and electricity needs to be repriced; for secondary market investors, it means reading notes and commitment disclosures more carefully. The winner of the AI race may ultimately depend on who can hold out at a lower cost of capital until commercialization pays off.

One-line Summary

US tech giants' hidden AI debt surged 90% in six months to $3.2 trillion, raising market alarm over off-balance-sheet leverage risk.

Paragraph Summary

Nikkei Asia reports that US tech giants' "hidden debt" in AI surged 90% over the past six months to $3.2 trillion, sparking heated discussion on Hacker News. Such liabilities largely stem from long-term contracts, leases, and off-balance-sheet financing arrangements for data centers, computing chips, and power supply — motivated by easing pressure on current-period statements and free cash flow, at the cost of reduced transparency and understated true leverage. Market concerns center on debt maturity structure, interest rate sensitivity, and chain reactions if AI investment returns fall short. The article argues that evaluating AI companies requires looking through off-balance-sheet structures to true leverage, and that supply chain counterparty risk and financing costs will be key to competition.