Alphabet, Microsoft, Amazon, Meta and Oracle have built up $1.65 trillion in hidden obligations tied to AI infrastructure expansion, BlockBeats reported on July 24. That total is about eight times the level from four years ago and is larger than the group’s combined on-balance-sheet debt of $1.35 trillion over the same period.
AI buildout is driving large off-balance-sheet commitments
As competition in AI heats up, the five U.S. tech companies have been spending heavily on data centers, GPUs and servers, using long-term lease agreements and equipment procurement contracts to lock in computing capacity.
Under U.S. GAAP accounting rules, some equipment purchases that have not yet been delivered, along with data center leases that have not yet started, do not need to be recorded on the balance sheet right away. Those future payment obligations are usually disclosed in financial statement footnotes.
Capital spending continues to rise across the group
Data cited in the report shows AI-related capital expenditure continuing to expand across the five companies:
- Alphabet reported $85 billion in capital expenditures for fiscal 2025.
- Microsoft posted $34.9 billion in capital expenditures in the first quarter of fiscal 2026.
- Amazon plans to invest about $200 billion in 2026, mainly for AWS data centers and AI chips.
- Meta expects 2025 capital expenditures of $60 billion to $65 billion.
- Oracle, supported by OpenAI cloud service orders, expects fiscal 2027 capital expenditures to rise to $90 billion to $95 billion.
Moody’s puts off-balance-sheet lease commitments at $662 billion
Moody’s data shows that Amazon, Meta, Alphabet, Microsoft and Oracle have signed about $662 billion in data center lease commitments that have not yet been recognized on their balance sheets. That amount equals 113% of the five companies’ adjusted debt.
As those lease agreements gradually take effect, more than $500 billion of related obligations could move into formal financial statements.
Warnings focus on cash flow pressure and circular financing risk
Analysts warned that if revenue growth from AI businesses fails to keep pace with rapid capital expansion, the scale of these long-term commitments could weaken corporate cash flow flexibility.
The report also said the AI industry’s expansion model is raising concerns about “shadow lending” and circular investment risk. Some technology companies are supporting AI infrastructure construction through bond financing, equity financing and off-balance-sheet contracts. At the same time, capital flows linking Nvidia, cloud service providers and major technology companies have drawn questions in the market over whether they may amplify expectations for the sector’s boom while masking the risk of insufficient real demand.