Tech Giants Have Hidden $3 Trillion in AI Liabilities Off Their Balance Sheets — WSJ
Alphabet, Meta, Microsoft, and other tech companies are entering into non-cancellable contracts for equipment and data centers without reflecting them in their financial statements
The liabilities of IT giants that are not reflected in their financial statements fall into two main categories. Photo: gguy/Shutterstock
Nine leading U.S. technology companies, including Alphabet, Meta, Microsoft, Amazon, and Oracle, have accumulated approximately $3 trillion in liabilities related to the development of artificial intelligence infrastructure that are not reflected on their current balance sheets. This was reported by The Wall Street Journal (WSJ), which analyzed the companies’ latest financial reports filed with regulators. These hidden liabilities are growing faster than traditional capital expenditures and already exceed the companies’ official debt from loans and operating leases by a factor of three, the newspaper reports.
The widespread concealment of expenses poses serious risks: corporations are making massive bets on future demand for AI by entering into non-cancellable contracts. If the technology fails to deliver the expected returns, these commitments will become a burden for investors, the WSJ notes.
What Make Up the Hidden Trillions?
The WSJ divides the IT giants’ off-balance-sheet liabilities into two categories. Under accounting rules, these liabilities may not be recorded on a company’s main balance sheet until the goods are delivered or the asset is put into use.
The bulkof the amount—about $1.9 trillion—consists of procurement commitments. These include long-term contracts for the purchase of equipment, such as Nvidia’s hard-to-find chips and memory, as well as agreements for the supply of electricity to data centers. At Alphabet, for example, such energy contracts extend as far as 2054, and the company’s total commitments of this kind reached $811 billion as of the end of June.
The remaining $1.2 trillion comes from lease agreements that have not yet taken effect, according to the WSJ. The total value of commitments for future lease payments by technology companies has increased roughly fourfold in just one year, the newspaper reports.
How It Works in Practice
As an example of how billions in liabilities disappear from financial statements, the WSJ examined Meta’s Hyperion data center project in Louisiana. Although Meta is the developer of the facility, neither the data center itself nor the $27 billion in debt raised for its construction appears on its balance sheet.
The structure is as follows: a majority stake in the complex is held by funds managed by Blue Owl Capital, while a third-party holding company raised funds through a bond offering. Meta is merely a minority partner and future tenant in this deal, the WSJ notes.
The company will begin making lease payments in 2029 for a term of up to 20 years and guarantees payments to bondholders. However, since the lease has not yet begun and the company itself assesses the likelihood of payments under the guarantees as “unlikely,” the total amount of the future lease, $12.3 billion, is not included at all in Meta’s current balance sheet.
In addition to Meta, The Wall Street Journal examined the commitments made by Alphabet, Amazon, Microsoft, Oracle, Nvidia, Broadcom, SpaceX, and Advanced Micro Devices.
Why Is This Important?
American "blue-chip" companies in the IT sector are spending enormous amounts of money based on the assumption that, within a few years, AI will have fully permeated all areas of business and consumer life. Optimists on Wall Street believe that future revenues will easily cover these massive commitments, but many analysts are sounding the alarm. Most of the signed lease agreements and equipment purchase contracts are non-cancelable. If demand for AI computing fails to live up to expectations, tech companies will be stuck paying for expensive infrastructure that cannot be used profitably, the WSJ points out.
Warning signs are already emerging: in their second-quarter reports, Alphabet and Amazon reported negative free cash flow. This means that their current capital expenditures are already exceeding the cash generated from their core operating activities.
“As these off-balance-sheet liabilities become increasingly common, large, and complex, it is becoming increasingly difficult for investors to assess companies’ aggregate potential debt burden,” Morgan Stanley analysts wrote earlier.
This article was AI-translated and verified by a human editor
