Rising yields squeeze AI data‑center financing: SoftBank pays up, lenders get choosier

Higher base rates and tighter credit screens are lifting the cost of building AI capacity—even as demand stays robust.

Benchmark yields at multi‑year highs are raising the price of money for the AI buildout. JPMorgan Chase estimated in June that roughly $4.1 trillion in AI‑related debt could be issued through 2030 as data‑center operators and linked firms race to add capacity. If average coupons clear even 100 basis points higher than previously assumed, that simple math implies about $41 billion more in annual interest expense once the full stock of new debt is outstanding.

That higher cost is already visible at the riskier end of the market. SoftBank raised $11.1 billion in a high‑yield sale this week, with the 7‑year tranche reportedly priced as high as 9.75%. Mark Malek, chief investment officer at Siebert Financial, said in an interview that many issuers are becoming “price takers” to secure the capital needed to compete.

The financing filter is tightening. Riley Thompson, a vice president at Mitsubishi HC Capital America, said lenders are getting pickier, estimating the market is now truly interested in roughly 20 neocloud providers rather than a far wider roster. A senior private‑credit investor told CNBC that such deals will be harder to finance as costs rise and cushions thin.

The rate channel matters in two ways. First, higher Treasury yields lift the base on which coupons are set. Second, for borrowers with floating‑rate exposure, the hit is immediate. CoreWeave’s latest quarterly filing shows that every 100‑basis‑point increase in rates would add about $30 million to its annual interest expense based on its outstanding floating‑rate debt.

There are early signs of project‑level stress alongside continued demand. Bloomberg reported that Oracle sent a “force majeure” notice tied to its New Mexico data‑center project to protect against higher expenses and potential delays; Oracle said the campus remains on its planned schedule. At the same time, lawyers and advisors active in AI infrastructure say activity is not abating. “Somebody will have to absorb it,” said Haim Zaltzman, vice chair at Latham & Watkins’ emerging companies and growth practice, adding that strong demand makes that easier. Andrew Giudici, who leads corporate, project and infrastructure finance at KBRA, expects issuance to remain relatively large even as rates bite.

Where pressure concentrates

Issuers with pre‑sold capacity and investment‑grade backers should retain market access, albeit at higher coupons. The squeeze is sharper for leveraged neocloud builders without anchor contracts. As costs rise, lenders’ shorter lists and covenants will triage marginal projects, pushing some timelines out or forcing repricing.

What to watch next

• Coupons and spreads on upcoming AI data‑center and GPU‑financing deals, especially in high yield.
• Issuer‑level exposure to floating versus fixed rates, and sensitivity disclosures like CoreWeave’s.
• Project milestones and any further delay or “force majeure” notices from large campuses.

Demand for AI services remains strong, but the cost of funding the next wave is moving higher. The market’s message is simple: capacity still clears — just at a steeper price.

This content is for informational purposes only and does not constitute financial advice.

Related Stories

Quantum Solutions and Hyperscale Data Monetize Crypto Treasuries to Fund AI Data Centers Bond yields hit multi-decade highs as tech rallies and cyclicals lag ahead of key U.S. da… Empery Digital Sells Bitcoin Treasury, Commits $20M to Cardinal Data Power’s AI Campus