In a recent address, the chief of the Bank for International Settlements (BIS) sounded an alarm about the rapidly accelerating pace of capital expenditure in artificial intelligence (AI) technologies. While the surge in AI-related investment is being hailed as a catalyst for future growth and productivity, the BIS chief warned that much of this spending is being financed through opaque forms of debt that could conceal underlying vulnerabilities in the financial system. The warning is rooted in a historical perspective: past periods of exuberant investment—such as the railway boom of the 19th century and the dot‑com bubble of the late 1990s—demonstrated how speculative optimism, when decoupled from genuine earnings, can culminate in broad‑based economic corrections. The BIS chief’s remarks highlighted several key mechanisms through which AI‑driven capital outlays could pose systemic risk.
First, the sheer scale of AI projects, ranging from large‑scale data centres to sophisticated machine‑learning platforms, requires substantial upfront capital. Companies, eager to secure a foothold in what they perceive as a transformative technology, are increasingly turning to complex financing structures—often involving special purpose vehicles, mezzanine debt, or other less‑transparent instruments—to fund these initiatives. Because these financing arrangements are not always fully disclosed in public financial statements, regulators may struggle to gauge the true exposure of banks and other financial intermediaries. Second, the BIS chief pointed out that the return on AI investment is still highly uncertain.
While early adopters have reported gains in efficiency, cost savings, and new revenue streams, many projects remain experimental, with long development horizons and ambiguous pathways to profitability. When capital is allocated based more on hype than on rigorous business cases, the risk of default rises, especially if the anticipated cash flows fail to materialise.
In such a scenario, the hidden debt that underpins these projects could become a source of contagion, spreading losses across the banking sector and beyond. Third, the chief emphasized the potential for a feedback loop between AI‑driven credit growth and broader macroeconomic stability.
If banks extend credit to finance AI ventures without fully appreciating the risk profile, they may inadvertently amplify credit cycles. An abrupt slowdown in AI spending—perhaps triggered by a regulatory clampdown, a technological setback, or a shift in market sentiment—could lead to a rapid contraction in credit, exacerbating economic downturns. This dynamic mirrors the pattern observed during the dot‑com bust, when inflated valuations and aggressive financing gave way to sharp corrections that rippled through the financial system. To illustrate the historical parallels, the BIS chief referenced the railway mania of the 1840s, when speculative investment in railway construction was financed largely through private bonds and unregulated credit.
The optimism surrounding rail transport led to an over‑extension of capital, and when revenues fell short of expectations, many investors faced severe losses, prompting a broader financial panic. Similarly, the dot‑com era saw a flood of venture capital and public market funding directed toward internet startups, many of which lacked viable business models. When the bubble burst, the fallout was felt not only by the tech sector but also by banks and investors exposed to the hidden liabilities of those companies.
Drawing lessons from these episodes, the BIS chief urged policymakers and regulators to increase transparency around AI‑related financing. This includes requiring clearer disclosure of debt structures, enhancing stress‑testing frameworks to incorporate AI‑specific risk factors, and fostering cooperation between supervisory authorities across jurisdictions. By shining a light on the opaque debt that underpins much of the AI investment surge, regulators can better assess the resilience of the financial system and intervene before systemic cracks emerge. The speech also touched on the broader societal implications of an AI‑driven investment race.
While AI holds promise for improving productivity, healthcare, climate modelling, and many other domains, the pursuit of technological leadership should not come at the expense of financial stability. Sustainable investment practices—such as aligning AI projects with long‑term strategic goals, conducting thorough due‑diligence, and maintaining prudent capital buffers—are essential to ensure that the benefits of AI are realized without creating hidden fragilities. In practical terms, the BIS chief suggested several concrete steps for the financial sector. Banks should enhance their credit‑risk assessment models to capture the unique characteristics of AI projects, including the pace of technological change, the concentration of talent, and the potential for regulatory shifts.
Supervisors could introduce reporting requirements that break down AI‑related exposures by industry, geography, and financing type, enabling a more granular view of risk concentrations. Moreover, central banks might consider incorporating AI‑related credit growth into their macro‑prudential monitoring tools, allowing for early detection of overheating in specific segments of the economy. The message resonated with many market participants who have expressed concerns about the speed at which AI investments are unfolding.
Some industry leaders have already begun to adopt more disciplined capital allocation frameworks, emphasizing incremental deployment, pilot testing, and clear pathways to revenue generation. These approaches not only mitigate financial risk but also provide valuable learning loops that can refine AI strategies over time. In conclusion, the BIS chief’s warning serves as a timely reminder that the excitement surrounding AI should be balanced with a sober assessment of the financial underpinnings that support it.
By drawing on historical analogies, emphasizing the need for transparency, and outlining actionable policy measures, the speech seeks to prevent a repeat of past bubbles where hidden debt and speculative fervour led to systemic upheaval. The ultimate goal is to foster an environment where AI can thrive as a genuine engine of growth—one that is built on solid financial foundations rather than on the fragile scaffolding of opaque borrowing.