As astronomical sums pour into the AI industry, a warning has emerged that weaknesses could spread across financial markets if the sector fails to secure revenue commensurate with the scale of investment. A key variable is whether money poured into AI data centres can generate sufficient returns.

On Oct. 26, American Banker pointed to an imbalance between revenue and the recovery of investment capital as the biggest risk in the AI investment boom. Global companies are investing trillions of dollars in AI data centres, semiconductors and power infrastructure, but it is unclear whether service revenue will rise enough to cover that spending, it said.

Gartner said AI data centre operators would need cumulative revenue of $8.5 trillion by 2030 to recoup investment and financing costs and achieve a 12 percent return on investment. It forecast that corporate and consumer AI revenue that could be attributed to data centres over the same period would total at most $3.6 trillion.

AI companies' performance was also cited as a gauge of investment risk. Anthropic posted revenue of about $16.3 billion in the first half of 2026, but its net loss reached about $80.7 billion. Most of the net loss, however, was a non-cash accounting loss driven by valuation gains in convertible bonds and similar instruments. The actual operating loss was about $693 million, and the full net loss should not be interpreted as a cash outflow.

Attention is also focusing on the financial sector's exposure to risk. The National Association of Insurance Commissioners said the insurance industry's private credit investments totalled about $1.2 trillion at the end of 2025. The share tied to AI-related investments was not clearly identified.

Private credit is a form of finance in which companies borrow directly from lenders outside the banking system. Because investment details and asset values are not sufficiently disclosed, critics say it is difficult to predict the scale of losses if problems emerge.

Experts worry that if AI revenue falls short of expectations, it could lead to declines in the value of data centre assets, bad loans and corporate restructuring.

The sustainability of the AI industry ultimately depends less on expanding investment itself than on its ability to generate real profits capable of shouldering massive debt and equipment costs, the analysis said.