One of the first things financial experts recommend investors do before they choose to buy individual stocks is look at fundamental metrics such as earnings, revenue, free cash flow and debt to get a sense of the underlying company’s financial health. The healthier a company’s operations and balance sheet are, the more comfortable investors should be paying a higher price for the company’s stock.
[Sign up for stock news with our Invested newsletter.]
However, a new study by Nikkei found five leading U.S. artificial intelligence hyperscalers have kept more than $1.6 trillion in debt off their balance sheets and relegated it to the footnotes of their financial filings. While this practice is perfectly legal under U.S. Securities and Exchange Commission rules, it may be setting the stock market up for a massive hidden debt crisis down the road:
— What is the hidden debt crisis?
— What are the risks of the hidden debt crisis?
— Have companies hidden debt in the past?
— Should investors be worried?
What Is the Hidden Debt Crisis?
The Nikkei study focused on the five largest AI infrastructure hyperscalers: Alphabet Inc. (ticker: GOOG, GOOGL), Meta Platforms Inc. (META), Microsoft Corp. (MSFT), Amazon.com Inc. (AMZN) and Oracle Corp. (ORCL). These AI hyperscalers are cloud computing specialists that are each investing hundreds of billions of dollars in building and managing massive data centers that will be necessary to run and scale intense AI workloads.
The problem with spending hundreds of billions of dollars on infrastructure that may not generate returns for years is that it can spook investors and make corporate balance sheets look really ugly. All of the AI hyperscalers are incredibly successful, profitable companies. But the sheer massive size of their AI data center investments is enough to weigh heavily on their overall financial ratios.
Fortunately for these hyperscalers, there are perfectly legal accounting tricks and loopholes that are compliant with generally accepted accounting principles (GAAP) that allow them to “hide” certain debt by keeping it off their main balance sheets.
One example of this type of loophole is a special purpose vehicle, or SPV. An SPV is typically a partnership between hyperscalers and private equity that involves creating separate corporate entities to borrow money to build data centers. Instead of taking on the debt themselves, the hyperscalers will take a minority ownership stake in the SPV or sign a long-term service or lease agreement with the SPV. Because the hyperscalers don’t hold a controlling interest in the SPVs, they are not required to disclose the SPV’s debt as line items of their main balance sheets. Instead, they must only disclose their contractual commitments to the SPVs in the footnotes and annotations of their quarterly financial filings.
The Nikkei study found hyperscalers are exploiting these types of financial loopholes to a potentially troubling degree. Nikkei reported Alphabet, Meta, Microsoft, Amazon and Oracle have a combined $1.65 trillion in off-balance-sheet debt compared to roughly $1.35 trillion in combined on-balance-sheet debt. In other words, these hyperscalers have more debt stashed in the footnotes of their financial filings than they do on their main balance sheets.
What Are the Risks of the Hidden Debt Crisis?
The good news for investors is that each of the five hyperscalers operates healthy, successful businesses that are in no danger of collapsing any time soon. For example, Alphabet recently reported a staggering $112.1 billion in net income and $40.7 billion in operating income in the second quarter of 2026.
The major danger is that the stock prices of these five companies are pricing in a degree of financial health that is misleading at best. An investor may look at the debt-to-equity ratio of the five hyperscalers and deem them to be safe investments. If that debt-to-equity ratio were to include these companies’ off-balance-sheet debt, it would more than double. Now that same investor might see these companies and their stock prices in a different light.
In addition, this hidden debt could threaten the companies’ investment-grade credit ratings. Moody’s Ratings recently warned investors it considers off-balance-sheet debt when issuing credit ratings to hyperscalers. If this debt ultimately leads to one or more hyperscalers losing their investment-grade credit rating, investors who were unaware of the off-balance-sheet debt could be blindsided by the downgrade.
Matt Twiford, managing director at Pegacorn Group LLC, says off-balance-sheet borrowing is perfectly legal, but it goes against the spirit of transparency GAAP accounting and SEC rules were created to enforce.
“When this much of a company’s true obligation shifts off the face of the balance sheet and is buried in the footnotes where most people never look, it undercuts the whole purpose of those rules, even if it follows the letter of them,” Twiford says.
“Many retail investors don’t have the sophisticated software, the financial models or the time to dig into the footnotes and quantify these off-balance-sheet obligations.”
[Read: 6 of the Best AI ETFs to Buy for 2026]
Have Companies Hidden Debt in the Past?
Off-balance-sheet debt is not new to Wall Street, but the market’s history with it may not be particularly reassuring to investors.
In the late 1990s, Enron notoriously relied heavily on SPVs to hide its toxic assets and debt from creditors and investors. During the subprime mortgage bubble of the early 2000s, Citigroup Inc. (C), HSBC Holdings PLC (HSBC) and other commercial banks relied on similar entities called structured investment vehicles (SIVs) to hold risky mortgage-backed securities off their main balance sheets and bypass regulatory capital requirements.
Neither situation ended well for investors.
But while there are parallels between the off-balance-sheet accounting practices of the past and today’s hyperscalers, there are also key differences. The biggest difference is that there is no evidence hyperscalers are taking advantage of these accounting loopholes to circumvent regulatory rules or commit outright fraud. The firms are seemingly building all the AI infrastructure they claim to be building and spending all the money they claim to be spending with the full intention of operating massive-scale AI data centers in the future. In the meantime, they are simply trying to keep their balance sheets as clean as possible in the eyes of investors.
Frankie DiAntonio, founder and CEO of Lexington Capital Holdings, says legality and compliance are key distinctions between the current off-balance-sheet borrowing by tech hyperscalers and more nefarious accounting in the past.
“Unlike Enron’s outright fraud or the toxic subprime assets of 2008, Big Tech’s off-book data center leases and SPVs are GAAP-compliant, legal structures supported by hundreds of billions in real organic cash flow from core advertising and cloud operations,” DiAntonio says.
Should Investors Be Worried?
When investors see some of the most valuable Big Tech companies in the world are using similar accounting practices as Enron did during its heyday and big banks did during the subprime mortgage bubble, it’s understandable they might be worried.
The risks to the market in the current situation may not be as extreme as the risks to Enron and big bank investors in the past, but it doesn’t mean they should be ignored.
Bank of America analyst Savita Subramanian says she doesn’t yet see any reason to believe the hyperscalers risk insolvency in the near future. However, Bank of America estimates the five firms have a combined $2.3 trillion in footnote debt, even more than Nikkei estimates.
“Capex shifting from hard asset buildout to off-balance-sheet lease obligations obfuscates true leverage,” Subramanian says.
She says adding this $2.3 trillion in off-balance-sheet debt to the balance sheets of the hyperscalers would push the S&P 500’s overall debt-to-capital ratio from 19% to 21%.
“Given the strong relationship between the equity risk premium and leverage, this might warrant further de-rating in earnings multiples,” Subramanian says.
An earnings multiple, or price-to-earnings ratio, is a fundamental stock valuation metric that reflects the price investors are willing to pay for one dollar of earnings per share. If a company’s earnings per share remain constant but its stock’s earnings multiple contracts 30% because investors become spooked about off-balance-sheet debt, the stock’s share price would also drop by 30%.
DiAntonio says investors shouldn’t be worried about imminent defaults of these hyperscalers, but valuation shock and margin compression are legitimate concerns.
“If AI revenue monetization stalls or takes significantly longer than expected, those fixed off-balance-sheet obligations will inevitably roll onto income statements as margin drag or asset write-downs, leaving retail investors exposed to sudden valuation adjustments,” DiAntonio says.
More likely than not, those “sudden valuation adjustments” would come in the form of a severe sell-off in tech sector stocks that could easily spread to the rest of the market and could even ultimately spiral into an economic recession.
More from U.S. News
5 Best Photonics Stocks to Buy for 2026
Donald Trump Stocks: 8 Stocks Owned by the President
7 Best Long-Term ETFs to Buy and Hold
Big Tech’s AI-Fueled $1.65 Trillion Hidden Debt Crisis originally appeared on usnews.com
