In a recent address to the financial community, the chief of the Bank for International Settlements (BIS) sounded an alarm about the rapidly accelerating flow of capital into artificial intelligence (AI) projects. According to the BIS chief, the surge in AI‑related capital expenditures is not being financed through transparent, sustainable sources but rather through a growing veil of opaque debt. This hidden borrowing, he argued, could sow the seeds of a systemic crisis if left unchecked. The warning was framed within a broader historical perspective.
The BIS official drew parallels with two of the most infamous speculative episodes of the past century: the railway mania of the 19th century and the dot‑com bubble of the late 1990s. In both cases, investors poured massive sums into new technologies that promised transformative change, often on the basis of overly optimistic forecasts rather than solid fundamentals.
When the anticipated returns failed to materialise, the resulting corrections were severe, leading to bankruptcies, job losses, and a loss of confidence that rippled through the wider economy. Applying this historical lens to today’s AI boom, the BIS chief highlighted several concerning dynamics.
First, the sheer scale of AI‑related spending has exploded. Corporations across sectors—from finance and healthcare to manufacturing and retail—are allocating unprecedented budgets to develop, test, and deploy AI models, especially large‑scale generative systems. While these investments have the potential to unlock productivity gains, they also represent a sizable portion of corporate capital budgets, often at the expense of other strategic initiatives.
Second, the financing of these projects is increasingly opaque. Traditional balance‑sheet reporting is being supplemented—or in some cases replaced—by off‑balance‑sheet arrangements, special purpose vehicles, and complex derivative structures that conceal the true level of indebtedness. This lack of transparency makes it difficult for regulators, auditors, and even company boards to gauge the real exposure to debt tied to AI initiatives.
Third, the cost structure of AI development is distinctive. Training cutting‑edge models can require billions of dollars in compute power, data acquisition, and talent. To meet these costs, firms are turning to short‑term borrowing, revolving credit facilities, and even issuing corporate bonds earmarked specifically for AI projects.
The rapid turnover of technology cycles means that many of these loans are short‑dated, creating a perpetual refinancing risk. The BIS chief warned that this confluence of high‑risk borrowing and speculative optimism could create a feedback loop. If AI projects fail to deliver the projected revenue streams—whether because of technical limitations, regulatory push‑back, or market saturation—companies may find themselves unable to service the hidden debt.
This scenario could trigger a wave of defaults that spreads beyond the AI sector, given the interconnected nature of modern financial markets. To illustrate the potential magnitude of the problem, the BIS head cited recent data showing that global AI‑related capital expenditures have risen by more than 150 percent in the last two years. At the same time, the proportion of corporate debt classified as “unsecured” or “high‑yield” has climbed, indicating that lenders are demanding higher risk premiums for financing AI ventures. The combination of rapid spending growth and rising risk premia suggests that the market is already pricing in a heightened probability of distress.
In response to these findings, the BIS chief called for a coordinated policy response. He urged central banks, supervisory authorities, and international standard‑setting bodies to develop clearer reporting guidelines for AI‑related debt, ensuring that companies disclose the purpose, maturity, and terms of any borrowing tied to AI initiatives. He also advocated for stress‑testing frameworks that incorporate AI‑specific risk factors, such as model failure, data breaches, and regulatory sanctions.
Furthermore, the BIS chief recommended that policymakers consider macro‑prudential tools—such as counter‑cyclical capital buffers or sector‑specific loan‑to‑value limits—to curb excessive leverage in the AI sector. By imposing limits on the amount of debt that can be used to finance AI projects, regulators could dampen the incentive to over‑invest based purely on hype.
The speech concluded with a reminder that innovation, while essential for long‑term growth, must be balanced with prudential oversight. The BIS chief emphasized that the goal is not to stifle AI development but to ensure that the financing of such projects is transparent, sustainable, and resilient to shocks. He called on industry leaders to adopt best practices in corporate governance, including rigorous project evaluation, clear risk‑adjusted return metrics, and robust contingency planning.
In summary, the BIS chief’s warning serves as a cautionary tale for both investors and policymakers. The AI investment boom, if financed through opaque and excessive debt, carries the risk of triggering a systemic correction reminiscent of past bubbles.
By learning from historical precedents and implementing stronger disclosure and supervisory measures, the financial system can better navigate the opportunities and challenges presented by the AI revolution, preserving stability while fostering genuine, value‑creating innovation.