Global borrowing linked to artificial intelligence (AI) fell sharply in September 2026. Investors are becoming more cautious about the rising cost of building AI infrastructure.
AI-related debt issuance dropped to $23 billion in September. That was less than half the amount raised in August. The decline followed months of heavy borrowing by technology companies to fund data centres, AI chips and other infrastructure.
According to the Financial Times, which cited data from Morgan Stanley, AI-related financing reached a record $113 billion in June. Since then, the amount of new debt raised each month has declined.
The slowdown comes as investors question whether massive AI investments will generate enough returns. Companies are spending billions of dollars on infrastructure. However, the financial benefits of these investments remain uncertain.
AI Debt Issuance Drops to $23 Billion in September
Morgan Stanley data show that global AI-related debt issuance fell to $23 billion in September. This was a sharp decline from the record $113 billion raised in June. This includes public bond sales and private debt placements. Both are used by companies to raise money for large projects.
The slowdown was also visible in the US investment-grade bond market. This market allows companies with relatively strong credit ratings to borrow money from investors. AI-related bond issuance in this market stopped completely in September. Technology companies had raised around $306 billion in debt between January and August.
However, the decline does not mean that companies have stopped investing in AI. Much of the money needed for planned projects may have already been raised earlier in the year.
Why Are Investors Becoming Cautious About AI Debt?
Technology companies are spending heavily to expand their AI operations. They need large data centres, powerful chips and reliable electricity supplies to support AI models and services.
Many companies are turning to debt to finance these projects. Borrowing allows them to fund expansion without relying entirely on their existing cash reserves. However, investors are now paying closer attention to the risks.
First, building AI infrastructure requires large amounts of money. Companies must spend on buildings, computing equipment, cooling systems and power supplies before many projects can generate revenue.
Second, the returns remain uncertain. AI services are attracting customers, but it is still unclear how quickly some infrastructure investments will become profitable.
Third, borrowing costs can put pressure on companies. Higher interest rates make debt more expensive to repay. They can also reduce the potential returns from long-term projects.
Investors are therefore looking more closely at companies’ financial strength, expected revenue and ability to repay their loans.
ALSO READ: AI Safety and Bubble Risks Take Center Stage at Singapore Conferences
AI Companies Face Growing Pressure to Prove Returns
The decline in borrowing comes after a major increase in AI-related financing during 2026. In June, Reuters reported that Morgan Stanley expected global AI-related debt issuance to reach nearly $570 billion for the full year.
The forecast reflected the growing need for capital as major technology companies expanded their AI infrastructure. However, the latest figures suggest that the pace of new borrowing has slowed.
This shift does not necessarily mean that investors have lost confidence in the entire AI sector. Some companies have stronger balance sheets and more established sources of revenue than others. Instead, investors may be becoming more selective about where they put their money.
Companies with clear business plans and reliable cash flows may find it easier to secure financing. Firms that depend on uncertain future earnings could face tougher borrowing conditions.
Data Centre Expansion Adds to Financial Risks
Data centres are a major part of the AI investment boom. They provide the computing power needed to train and run AI models. But these facilities are expensive to build and operate. They also require access to large amounts of electricity and suitable land.
Delays can increase costs and push back the date when a project starts generating revenue. Power shortages, construction problems and local opposition can create further challenges. These risks matter because companies often make large financial commitments before a data centre becomes operational.
If demand for AI services grows more slowly than expected, some projects could take longer to recover their costs. Companies with high debt levels could face greater pressure in that situation.
Investors are therefore looking beyond AI demand forecasts. They also want to know whether companies can complete projects on time and turn their investments into sustainable income.
ALSO READ: Nvidia-Backed Firmus Abandons $31 Billion IPO After Investor Concerns
What the AI Debt Slowdown Means for the Industry
The fall in AI-related debt issuance could lead to closer scrutiny of future funding deals. Investors may demand better borrowing terms or stronger evidence that projects can generate returns.
Some companies could also explore other funding options. These may include using their own cash, bringing in new investors or working with infrastructure funds. The slowdown could affect smaller AI companies more than established technology firms. Businesses with limited revenue and high funding needs may find it harder to attract lenders.
However, the September figures alone do not prove that an AI investment bubble is about to burst. The decline partly reflects the large amount of borrowing completed earlier in the year. Now it’s a question whether AI demand and revenue will grow enough to support the industry’s enormous spending plans.
For now, investors appear to be taking a more careful approach to AI-related debt. The next few months will help show whether the slowdown is temporary or a sign of tighter financing conditions for the sector.