Business · Companies · 1 hr ago
S&P warns hyperscalers may have more AI-related debt than disclosed
S&P Global says major tech firms may have more AI-related debt than disclosed.
The concern involves hyperscalers such as Amazon, Microsoft, Alphabet, Oracle and Meta, which are investing heavily in data centres and AI infrastructure.
Some financing arrangements can leave these firms backing data centre projects or smaller AI companies, making their total liabilities harder to see.
S&P expects six major US firms to spend more than $7 trillion on AI-related projects from 2025 to 2030.
Meanwhile, bonds linked to AI infrastructure have come under pressure, and several planned stock market listings in the sector have been delayed or withdrawn.
S&P says there is no current reason to change its ratings, though it previously lowered Oracle’s rating to the lowest investment-grade level.
That view could change if AI spending proves unsustainable or expected profits fail to materialise.
S&P Global says the largest technology companies may have more AI-related liabilities than are publicly known.
S&P digital infrastructure chief Mark Habib said hidden liabilities could arise from financing structures that support data centre projects and smaller AI companies.
S&P does not currently see a major gap between market concerns and its credit ratings, but says that could change if spending proves unsustainable.
Investors have sold some AI infrastructure debt, while several planned industry stock-market listings have been postponed or withdrawn.
S&P expects six major US hyperscalers to spend more than US$7 trillion on data centres and other AI-related investment from 2025 to 2030.
- Who
- S&P Global, major technology companies and investors are central to the story.
- What
- S&P warned that hyperscalers may have more AI-linked liabilities than are publicly known, as investors scrutinise debt tied to data centre construction.
- When
- Habib made his comments on October 19, 2026; the reports were published on October 19 and 20.
- Where
- Habib spoke at a roundtable in London.
- Why
- Data centre and AI investment requires substantial borrowing, and investors are concerned that future revenues may not be enough to repay debt.
Investors wary of AI infrastructure debt
S&P Global
Market risk
Investors wary of AI infrastructure debt
Investors have sold debt linked to AI infrastructure amid concerns over borrowing and whether future revenues will cover repayments.
S&P Global
S&P says it does not currently see a major disconnect between market concerns and its ratings.
Potential liabilities
Investors wary of AI infrastructure debt
Investors are wary of complex financing structures and the risks they may pose.
S&P Global
S&P says hyperscalers may have additional debt that is not visible, but has no estimate of its amount.
S&P downgraded Oracle to BBB-, citing its AI spending and an uncertain path to profitability.
Spain's Ignis postponed its planned stock-market debut until 2027, according to El País.
Australia's Firmus withdrew a planned US$5 billion share offering after failing to find demand at its requested price.
AirTrunk postponed its planned Singapore listing until the following year.
Mark Habib discussed possible hidden AI-related liabilities at a roundtable in London.
- Expected spending
- More than US$7 trillion by six hyperscalers from 2025 to 2030
- Hyperscalers named
- Amazon, Microsoft, Alphabet, Oracle, Meta and SpaceX
- S&P analyst
- Mark Habib, global head of digital infrastructure
- Oracle rating
- BBB-, lowered in July 2026
- Data centre debt
- Morgan Stanley estimates data centre-owning companies have issued about US$46 billion in bonds
4 sources
What people are saying
It is safe to assume that there is more (debt) out there that we are not seeing
For now, we don't have the view that there is a major disconnect that needs to force our ratings to change at this point
Further increases in the investments (by AI firms), or a view that the revenue rate of competition is getting too intense