The AI Gold Rush is facing an Uncomfortable Question: Will It Generate Enough Money to Justify the Enormous Sums Being Spent?
By LauraAboli
According to the Financial Times, bonds financing major data centers linked to Meta and Microsoft have fallen well below their original value, while some CoreWeave bonds are trading below 80 cents on the dollar, with yields approaching 15%. Investors are demanding a much higher potential return for taking the risk of lending to the company.
These are signs of growing concern about repayment, although bonds financing a facility leased to Microsoft should not be confused with debt issued by Microsoft itself.
Meanwhile, Gartner estimates that AI data center owners would need $8.5 trillion in cumulative revenue through 2030 to recover their investment and financing costs and deliver a 12% return, against projected revenue of just $3.6 trillion.
That is a forecast, not a loss that has already happened, but it raises serious questions about whether the scale of construction makes financial sense.
If financing becomes harder to obtain, projects could be delayed, chip orders reduced, and heavily indebted operators forced to sell or restructure. Gartner predicts that 80% of specialist AI cloud providers could lose their independence by 2030, through closures, mergers or acquisitions.
AI may continue transforming our lives while investors lose fortunes financing it, and if the weaker companies are swallowed by the strongest, we could end up with even more of this powerful technology concentrated in fewer hands.
