The global data-centre buildout has raised about $1.8 trillion in financing since 2020, BloombergNEF says, as AI infrastructure spending adds to credit risks. At least $1 trillion of debt is due to mature over the next five years, putting refinancing on the agenda alongside the race to build.
Watch Desk analysis
What happened
In its report, Financing Compute, BloombergNEF says private markets are playing an increasingly prominent role in financing data-centre expansion. It links the surge in capital spending on AI infrastructure to more complex deal structures and heightened credit risks.
BNEF puts total financing raised since 2020 at approximately $1.8 trillion and says at least $1 trillion in debt is scheduled to mature during the next five years, creating refinancing pressure.
Why it matters
AI infrastructure is often discussed in terms of chips, power and construction. The financing behind it has its own timetable: debt coming due has to be refinanced, repaid or dealt with some other way, even if demand for computing shifts. The sums BNEF cites make that a substantial part of the buildout story, not a footnote for the finance team.
This is not a forecast that the market will fail. It is a report of the scale of financing and upcoming maturities, alongside BNEF’s assessment of growing credit risks. The crucial question is how comfortably those obligations can be managed if the pace or economics of data-centre investment change.
Our read
The $1.8 trillion figure is eye-catching; the maturities are the more useful number to keep in view. A boom can look persuasive while money is flowing in. The test is whether the projects can support their financing when repayment dates arrive. Infrastructure plans deserve scrutiny not just for how much computing they promise, but for who is financing them and when the bill falls due.
What to watch
- How much of the debt due over the next five years is refinanced, repaid or restructured.
- Whether private-market financing continues to grow in importance.
- How data-centre operators and lenders respond if AI infrastructure spending or demand changes.
Discussion spark: When AI infrastructure is financed on this scale, should investors focus more on the capacity being built or on whether the debt can be refinanced if demand disappoints?
Sources and evidence
- Financing Compute: The $1.8 Trillion Bet on AI Data Centers (19 November 2026, 22:17 UTC)
- Powering the AI Data Boom With Recycled EV Batteries – Business Insider (19 November 2026, 22:17 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.