Institutional
A Wall Street Journal analysis finds nine major technology companies carrying roughly $3 trillion in off-balance-sheet AI infrastructure commitments, with Morgan Stanley warning the structure makes true leverage hard for investors to see
About $1.2 trillion of the total comes from unstarted leases and $1.9 trillion from purchase commitments tied mostly to AI data-centre buildout, none of which shows up on a standard balance sheet. Morgan Stanley separately projects total AI infrastructure spending will reach $3.2 trillion by 2028, with roughly $1.75 trillion of that needing to be raised through credit markets.
This site has spent the past month tracking what happens when leverage tied to the AI trade sits somewhere investors cannot easily see it, first through Situational Awareness's forced unwind and the SEC subpoenas that followed, then through JPMorgan's warning about hidden leverage across AI-thematic funds. The WSJ's balance-sheet analysis extends that same concern to the companies actually building AI infrastructure rather than the funds betting on it, and the numbers are large enough to matter: roughly $3 trillion in commitments across nine major technology companies that a standard balance-sheet read would simply miss.
The mechanics matter here. Unstarted leases and purchase commitments are real obligations, a hyperscaler that has agreed to buy a decade of data-centre capacity or lock in GPU supply is on the hook for that spending whether or not the underlying AI products generate matching revenue, but accounting conventions let both sit outside the liabilities line investors check first. Morgan Stanley's own framing, that the real capex is now substantially off-balance-sheet, is a direct statement that headline balance-sheet leverage ratios for the largest AI spenders currently understate the actual funding pressure those companies face.
The scale of what still needs financing sharpens the point. Morgan Stanley's $3.2 trillion infrastructure forecast through 2028 implies roughly $1.75 trillion has to come from credit markets rather than existing cash flow, which means the bond and private-credit markets underwriting that debt are effectively underwriting the AI trade's physical build-out, often without full visibility into how much off-balance-sheet exposure a given borrower already carries.
For institutional readers already watching prime-broker exposure to AI-thematic hedge funds, this is the corporate-credit mirror of the same question: how much AI-linked leverage has accumulated in structures that don't show up in the numbers everyone checks first, and what happens to that leverage if AI monetisation timelines slip. Both threads point toward the same regulatory pressure point, better disclosure of commitments that function like debt but aren't booked as debt.
Read the original: TipRanks - A Wall Street Journal analysis finds nine major technology companies carrying roughly $3 trillion in off-balance-sheet AI infrastructure commitments, with Morgan Stanley warning the structure makes true leverage hard for investors to see. Commentary is the independent editorial view of Share Trading; the original article is credited to its publisher.