A new analysis from Morgan Stanley finds that seven major technology companies — including Google, Microsoft and Nvidia — have roughly $3 trillion in off-balance-sheet commitments tied to AI infrastructure. That amount sits on top of an estimated $770 billion in debt and lease obligations that already appear on their balance sheets.
What are off-balance-sheet commitments?
Off-balance-sheet commitments are promises to make future payments that do not appear as official debt on a company's standard balance sheet. As an analogy, a household balance sheet might list a mortgage and car loan as liabilities, but a multi-year rental commitment for a vacation home would not affect the mortgage evaluation in the same way.
Who and what did Morgan Stanley analyze?
The research examined public filings from the hyperscalers Google, Meta, Microsoft, Oracle and Amazon, along with chipmakers Nvidia and Broadcom. Key findings include:
- The hyperscalers have committed $1.1 trillion in payments for data-center leases that have not yet begun.
- All seven companies have agreed to purchase $1.7 trillion worth of items such as chips, memory and networking gear.
- Purchase agreements accelerated this year: Morgan Stanley notes Google’s commitments reached $707 billion in the most recent quarter, up from $72.5 billion in all of 2025, according to the research.
Why this matters
Those future payment guarantees become valuable collateral for suppliers and data-center developers, who can borrow against long-dated leases, guarantees or purchase commitments from investment-grade hyperscalers. This allows capacity to be built before the hyperscalers make payments or recognize liabilities on their books.
Morgan Stanley researchers summarize the dynamic: suppliers and data-center builders can obtain financing by pledging the long-term commitments from highly rated hyperscalers, enabling construction and capacity expansion in advance of actual payments.
Risks and open questions
Each company’s spending program differs in type and risk level. It is not yet clear over what time frame the roughly $3 trillion would actually be spent — or whether all of it will be spent at all. These are contractual commitments and, in theory, could be renegotiated.
Timing is the central concern for investors, says Todd Castagno, head of global valuation, accounting and tax at Morgan Stanley and a coauthor of the analysis. He compares the situation to building a market for cars without fully understanding what the cars will be capable of or what they will be worth in five years, raising questions about how to finance and value those investments.
Bottom line
Even as large numbers become commonplace, a small group of technology companies is driving unprecedented levels of spending and borrowing tied to AI infrastructure. Those off-balance commitments can generate substantial lending and capacity expansion, but whether the investments will ultimately pay off — and when their effects will appear on official balance sheets — remains uncertain.
