The hidden AI-related liabilities of five major technology companies have climbed to about $3.2 trillion, a 90% increase since March 2026, according to an analysis by Nikkei Asia cited by t3n. The figure covers Alphabet, Microsoft, Amazon, Meta and Oracle.
The liabilities stem from “shadow borrowing”: separately established companies borrow money to build and operate data centers and other AI infrastructure, which the tech firms then rent or lease. Some commitments do not have to appear on the companies’ balance sheets, making their debt levels less apparent.
Nikkei Asia’s analysis says 60% of the total consists of purchase commitments, while 40% relates to lease agreements that have yet to begin. Together, the obligations amount to 80% of the five companies’ combined assets.
Alphabet accounts for the largest share, with $1.05 trillion in reported hidden liabilities, up 160%. However, Nikkei Asia cautioned that comparisons are difficult because Alphabet changed parts of its calculation method. The other companies, in descending order, are Meta, Microsoft, Amazon and Oracle.
Investors are increasingly questioning whether the companies’ enormous spending on AI infrastructure will generate returns soon enough. Share prices at many large technology companies fell after they reported third-quarter results in late October 2026, and the doubts are also weighing on the planned stock market debut of AI company Anthropic.
Sources: t3n.