- Global AI-related debt issuance fell from $113B in June to $23B in September 2026.
- At least 47 U.S. data center projects faced setbacks; only 35% of plans look credible.
- Gartner warns over $4 trillion in AI assets could face write-downs through 2032.
The U.S. AI data center boom is facing mounting debt risks as rising construction costs and project delays threaten financing for a projected $10 trillion infrastructure expansion. Developers are borrowing billions, while some contracts allow AI customers to withdraw if financing or delivery deadlines are missed.
Debt Markets Tighten as CoreWeave Shares Slide
Global AI-related debt issuance fell to $23 billion in September 2026 from $113 billion in June, according to figures reported by the Financial Times. JPMorgan CEO Jamie Dimon confirmed that his bank was strengthening its credit to reduce concentration risk.
CoreWeave shares lost nearly 25% in one week as investors questioned its debt-funded expansion. Some of its bonds traded at yields above 15%, while bonds funding data centers leased to the company also weakened. Nvidia CEO Jensen Huang later publicly expressed support for CoreWeave’s business.
Elsewhere, approximately $18 billion in bank loans tied to an Oracle data center campus entered challenging territory. Nvidia-backed Firmus shelved an initial public offering that had sought a valuation exceeding $30 billion.
Other postponed or shelved public listings include Blackstone-backed AirTrunk REIT and SoftBank-backed SB Energy. OpenAI has also pushed back its IPO plans.
Project Delays and Local Restrictions Mount
Construction setbacks are adding to the financing pressure. RegLayer counted at least 47 major U.S. data center projects delayed, withdrawn or blocked in 2026, according to the Financial Times.
Bernstein analysts estimated that only about 35% of the planned U.S. data center pipeline was credible. Separately, FT research found that more than 1,000 American communities had sought moratoriums or other restrictions on AI infrastructure this year.
Power access presents another obstacle. Utilities are increasingly refusing to connect projects that cannot supply their own electricity. Meanwhile, contractual exit rights can leave developers carrying debt on unfinished facilities if customers withdraw after missed deadlines.
Spending Forecasts Outpace Expected AI Revenue
These borrowing problems come as AI providers face a gap between anticipated revenue and planned expenditure on data centers and chips. Gartner forecasts that AI providers will spend roughly $7 trillion on infrastructure through 2030, while related enterprise and consumer revenue will total no more than $3.6 trillion over that period.
If investment continues at its current pace, more than $4 trillion in asset value could be at risk of write-downs through 2032. Gartner also predicts that over 80% of specialized AI cloud providers, known as neoclouds, will cease operating independently by 2030.
Related: Tech Firms Pledge Power Supply to Support Growing AI Data Centers
Disclaimer: The information presented in this article is for informational and educational purposes only. The article does not constitute financial advice or advice of any kind. Coin Edition is not responsible for any losses incurred as a result of the utilization of content, products, or services mentioned. Readers are advised to exercise caution before taking any action related to the company.
