Goldman Sachs Puts Big Tech’s Off-Balance-Sheet AI Commitments at $3.2 Trillion

Goldman Sachs estimates Alphabet, Amazon, Meta, Microsoft and Oracle have about $3.2 trillion in future commitments tied to the artificial intelligence buildout. The estimate, in a Nov. 3 credit strategy note by Goldman Sachs credit strategist Amanda Lynam, concerns obligations that are not cash already spent or funded debt on the companies’ balance sheets.

The estimate combines roughly $1.3 trillion in lease commitments for leases that have not yet commenced with $1.9 trillion in other purchase commitments. The figures are rounded, and the reported total is consistent with adding those two amounts.

The five companies have about $1.7 trillion in total lease commitments. More than 75% of that amount — roughly $1.3 trillion — is tied to leases that have not yet started. The full $1.7 trillion in lease commitments is not part of the $3.2 trillion total; the estimate’s lease component is the portion for leases not yet commenced.

The companies also have $1.9 trillion in other purchase commitments, including $685 billion due by Dec. 31, 2027. Those obligations are future payments, not a measure of cash already spent.

Alphabet’s filing for the quarter ended Sept. 30 listed $924.4 billion in future material purchase commitments and $130.5 billion in payments for leases not yet commenced. The filing said those future cash commitments were not reflected on its balance sheet and were subject to counterparties performing on their obligations.

Lynam wrote that Goldman believes some of the liabilities are captured in consensus capital expenditure forecasts for hyperscalers, while the category still represents an additional “use of cash.” She wrote, “This leaves us comfortable with our view that debt financing — across syndicated and private markets — will play a key role in the AI buildout.”