In a recent address, the head of the Bank for International Settlements (BIS) sounded a stark warning about the burgeoning competition among corporations and governments to pour capital into artificial intelligence (AI) technologies. While the promise of AI is undeniably compelling—offering the potential to reshape industries, boost productivity, and create new revenue streams—the BIS chief cautioned that the current surge in AI‑related capital expenditures (capex) is being financed in ways that could jeopardize the stability of the broader financial system. At the heart of the concern is the reliance on opaque, often poorly documented debt structures to fund AI projects.
Unlike traditional financing arrangements, many of these new funding deals are shrouded in secrecy, lacking clear disclosure about the terms, collateral, and risk‑sharing mechanisms involved. This opacity makes it difficult for regulators, investors, and even the borrowing firms themselves to fully assess the true level of leverage being taken on. When large swaths of capital are funneled into AI initiatives under such uncertain conditions, the risk of a systemic shock—should any of these projects fail to deliver expected returns—rises sharply.
The BIS chief drew historical analogies to illustrate the pattern. He referenced the railway mania of the 19th century, when speculative investment in rail infrastructure surged, often financed through dubious debt that later contributed to financial panic when the anticipated traffic and profits failed to materialize. Similarly, he pointed to the dot‑com bubble of the late 1990s, where exuberant spending on internet‑related ventures, many of which were underpinned by speculative financing, culminated in a dramatic market correction that erased trillions of dollars in value. Both episodes share a common thread: a fervent belief in a transformative technology, combined with a financing environment that prioritized speed and scale over rigorous risk assessment.
The BIS warning suggests that the AI investment wave may be following a comparable trajectory. Companies are racing to adopt generative AI, large‑language models, and advanced analytics, often committing billions of dollars before fully understanding the technology’s limitations, integration challenges, or realistic return on investment. One of the most troubling aspects of the current AI capex race is the prevalence of “shadow” financing. This term describes funding that bypasses traditional banking channels and instead flows through private equity, venture capital, or even corporate balance‑sheet borrowing that is not transparently reported.
Such financing can mask the true leverage ratios of firms, making it harder for regulators to gauge systemic exposure. Moreover, the rapid pace of AI development means that many firms are signing multi‑year contracts with technology providers, locking in large, fixed‑price commitments that may become untenable if market conditions shift or if the technology fails to meet performance expectations. The BIS chief emphasized that the systemic risk is not limited to the borrowers alone. Lenders, including banks and non‑bank financial institutions, could find themselves exposed to a cascade of defaults if a wave of AI projects underperform simultaneously.
This scenario could amplify stress across credit markets, potentially leading to tighter lending standards, reduced liquidity, and heightened volatility in asset prices. To mitigate these risks, the BIS chief called for greater transparency in AI‑related financing. He urged firms to disclose the terms of their AI debt, including interest rates, maturity dates, covenants, and the specific projects the funds are earmarked for. Such disclosure would enable investors and regulators to better assess the concentration of AI‑related leverage within the financial system.
Additionally, he advocated for the development of standardized reporting frameworks that capture the unique risk characteristics of AI investments, such as technology obsolescence, regulatory uncertainty, and data privacy concerns. Regulators, according to the BIS chief, should also consider implementing macro‑prudential tools tailored to the AI sector. These could include caps on the proportion of a firm’s balance sheet that can be allocated to AI projects, stress‑testing scenarios that model adverse outcomes for AI deployments, and heightened capital requirements for lenders heavily exposed to AI‑financed borrowers.
Beyond regulatory measures, the BIS chief highlighted the importance of corporate governance in managing AI spending. Boards of directors need to scrutinize AI investment proposals with the same rigor applied to other strategic initiatives, ensuring that projected benefits are realistic, that implementation timelines are feasible, and that there are contingency plans for potential setbacks. By embedding disciplined decision‑making processes, firms can avoid the trap of “hype‑driven” spending that prioritizes being first to market over sustainable value creation.
In summary, while the AI revolution holds extraordinary promise, the BIS chief’s warning serves as a reminder that the path to harnessing this technology must be paved with prudent financing practices, clear disclosure, and vigilant oversight. Ignoring these safeguards could repeat the mistakes of past technological booms, where unchecked optimism and opaque debt structures paved the way for severe economic corrections.
By fostering transparency, strengthening regulatory frameworks, and encouraging responsible corporate governance, the financial system can better navigate the AI investment surge without compromising its stability.