|
Neubauer Artists LLC
Getting your Trinity Audio player ready...
|
PALO ALTO, California — U.S. tech giants’ off-balance-sheet debt from artificial intelligence-related obligations continues to balloon, almost doubling in six months to about $3.2 trillion, with investors increasingly alarmed by seemingly unending spending.
Nikkei compiled data from the latest earnings disclosures and estimates from Google owner Alphabet, Microsoft, Amazon, Meta and Oracle.
Tech groups procure chips, servers and power from suppliers to build AI data centers, often signing contracts spanning several years to secure access amid tight supply. Some also lease equipment or reserve computing resources from cloud service startups instead of building their own facilities.
Unfulfilled purchase commitments and lease agreements that have not yet started are generally treated as off-balance-sheet items. Although fully compliant with accounting rules, they are seen as hidden debt that is likely to be booked in the future.
Such hidden debt has expanded to about $3.2 trillion across the five companies. Roughly 60% consist of purchase commitments and 40% of lease obligations that have yet to begin. The total already equals about 80% of their combined assets, and could soon exceed total assets if this pace continues.
Google has the most hidden debt at $1.05 trillion as of Sept. 30. The figure is about 160% higher than at the end of March, though direct comparison is difficult because the company appears to have changed part of its calculation methodology. Meta has the next highest figure, followed by Microsoft, Amazon and Oracle.
Google is directing more cash into data centers and other capital expenditures than it generates from its core business. Capital investment cash outflows reached $71.3 billion in the July-September quarter.
Combined with operating cash flow, the figure was about $29 billion in the red, marking a third consecutive quarterly deficit. The company plans to further boost spending in 2027 to meet robust demand.
Meta’s off-balance-sheet obligations totaled about $810 billion at the end of September, up 90% from the end of March.
The company is increasing computing capacity using outside investor capital. In July, it announced a joint venture with U.S. investment firm BlackRock and signed an agreement to lease data centers.
Meta also agreed in April to purchase about $21 billion worth of cloud capacity through 2032 from U.S. startup CoreWeave.
Part of these companies’ hidden debt has also served to bolster the creditworthiness of startups like CoreWeave. Even with weaker balance sheets, these companies can attract financing from banks and investors because they have future sales backed by contracts with the tech giants.
But signs of change are now emerging in this unprecedented infrastructure investment boom.
CoreWeave’s bond prices plunged in October, spreading concerns about the financial health of AI-focused cloud startups, sometimes called neocloud providers. Many such companies have limited revenue sources beyond cloud services, leaving them vulnerable if rising interest rates make additional fundraising more difficult and jeopardize continued investment in data centers.
Investor concerns are also intensifying over the massive AI spending of highly creditworthy tech companies themselves. Following earnings announcements last week, shares of Alphabet and Amazon fell sharply after both signaled plans to continue expanding investments in 2027 and beyond.
Oracle — which has the least hidden debt of the five companies — faces particularly strong concerns about its finances. Its shares have fallen more than 60% from their peak about a year ago.
To reassure investors, the company said last week that it would return free cash flow, which reflects the balance between core earnings and capital expenditures, to positive territory in the period from December 2027 to May 2028.
Today’s AI industry is built on a web of long-term purchasing agreements among tech giants, chipmakers, AI developers and neocloud providers. Investment, financing and guarantees reinforce one another, enhancing creditworthiness and allowing companies to raise vast sums from financial markets.
But all of the credit is still tied to a few tech giants. If AI investment continues at its current pace and profits fail to meet expectations, the market’s trust in those companies could come under pressure, with repercussions potentially spreading across both the tech sector and the broader financial system.
