The scale of debt hidden off the books by five of the world’s largest technology companies is staggering. According to a study by Nikkei, Alphabet, Microsoft, Amazon, Meta, and Oracle collectively hold $1.65 trillion in off-balance-sheet liabilities, roughly eight times what they reported four years ago. This hidden figure now eclipses the $1.35 trillion they officially owe, raising fresh questions about the true financial health of the industry driving the artificial intelligence boom.
The Enron echo
The method used to mask this debt is not new. It relies on off-balance-sheet vehicles—special-purpose entities that allow companies to keep massive financial obligations from appearing on their ledgers. This is the same type of structure that Enron employed to conceal its debts before its dramatic collapse in 2001. Back then, it was fraud because Enron deliberately hid the existence of these vehicles. Today, the same tools are legal, provided they are disclosed in footnotes. As analyst Gil Luria told Bloomberg Law, "Enron’s crime wasn’t having special purpose vehicles. Enron’s crime was hiding them." The distinction shows how accounting rules have tightened, yet still leave room for substantial off-balance-sheet financing.
How it works
The mechanics are straightforward. The tech companies package debt incurred for building AI data centers—costs for chips, servers, power, and real estate—into separate legal entities, often joint ventures. Because these entities are technically independent, the debt never flows through the parent company’s balance sheet. Consider Meta’s Hyperion data center in Louisiana. Meta partnered with Blue Owl Capital, both contributing equity into a separate structure that took on $27 billion in debt. Meta is the sole tenant of the facility, yet it argues it does not have to record that debt because it is not responsible for finding replacement tenants if it leaves. This reasoning has been accepted under current accounting standards, though critics say it obscures the true leverage of these firms.
Oracle provides another stark example. The company has $260 billion in future lease commitments that will eventually land on its books when data centers become operational. Nvidia, a key supplier of AI chips, carries $119 billion in purchase obligations. Alphabet and Microsoft also keep vast vehicle-related liabilities off-book. The common thread is that all these obligations are tied to the rapid buildout of AI infrastructure, which the industry expects to cost more than $3 trillion through 2028.
The numbers behind the boom
Meta’s off-balance-sheet debt alone is roughly $420 billion, nearly triple its reported debt. Oracle’s hidden leverage has grown roughly thirtyfold in just four years. This explosion in off-book financing is driven by the race to dominate AI. Companies are spending aggressively on data centers and hardware, often financed through structured deals that keep the debt out of sight. The reported debt figures look manageable, but the hidden pile is much larger and more concerning to analysts who track the industry.
The timing is particularly notable. Four of the five companies are set to report earnings in the next two weeks. Their official debt numbers will appear tidy, while the $1.65 trillion lurking in footnotes will not make headlines. This selective visibility creates a distorted picture for investors and regulators alike.
Why it matters now
The risk lies in what happens when these data centers go live. Leases and other obligations move onto the balance sheet at once, triggering a sudden increase in reported debt. If AI demand falls short of the massive bets being placed, facilities could be marked down. The losses would then land squarely on the lenders and insurers who funded them—potentially causing ripple effects across the financial system. Some institutions are already sounding alarms. S&P has cut Oracle’s credit rating due to stretched leverage. Both Morgan Stanley and Moody’s have flagged the wider issue. Accounting consultant Tom Selling posed a pointed question: "What if one of these companies was a house of cards and was propping itself up with this accounting treatment?"
Regulators have not yet stepped in, partly because the disclosures exist in footnotes. However, the complexity and scale of these vehicles make it difficult for average investors to fully grasp the risk. The situation echoes the early 2000s, when Enron’s off-balance-sheet structures collapsed, causing billions in losses and triggering the Sarbanes-Oxley Act. Today’s environment is different—disclosures are mandatory—but the underlying principle remains: hidden debt can turn dangerous when market conditions change.
The honest version
None of these practices are illegal, and the companies insist that future earnings from AI will more than cover the bills. The disclosures are there, buried in dense footnotes, for anyone willing to dig through financial reports. But in a year when talk of an AI bubble is rampant, the quiet number—the $1.65 trillion off the page—may be the most important figure investors are not seeing. The industry is effectively betting that AI demand will justify the massive spending, and that the off-balance-sheet debt will be serviced without triggering a crisis. Whether that bet pays off will determine not just the fortunes of these five giants, but the stability of the broader tech finance ecosystem.
Source: TNW | Amazon News