In a recent address, the head of the Bank for International Settlements (BIS) sounded a stark warning about the rapidly intensifying competition among corporations to pour capital into artificial‑intelligence projects. This "AI capex arms race," as it is being called, is increasingly being financed through layers of opaque borrowing that conceal the true extent of leverage across the financial system.
The chief’s message was clear: the combination of speculative spending and hidden debt structures could sow the seeds of a systemic crisis if left unchecked. The BIS chief began by contextualising the current surge in AI‑related spending within a broader historical framework.
He referenced two well‑known periods of over‑investment that ultimately ended in painful corrections: the 19th‑century railway boom and the late‑1990s dot‑com bubble. In both cases, investors and corporate managers were swept up by a fervent belief that new technology would generate unprecedented profits. This optimism drove massive capital outlays—building rail lines across continents or launching countless internet startups—often financed through debt that was not fully transparent to regulators or even to the companies themselves. Fast‑forward to today, and a similar pattern is emerging, albeit with far more sophisticated financial engineering.
Companies across sectors—from cloud providers and semiconductor manufacturers to traditional enterprises seeking to embed AI into legacy processes—are racing to acquire cutting‑edge hardware, hire top‑tier data scientists, and secure massive cloud‑computing contracts. The scale of these investments is staggering; analysts estimate that global AI‑related capital expenditures could exceed several hundred billion dollars within the next few years. Yet much of this spending is being funded through a mix of corporate bonds, revolving credit facilities, and private‑equity‑backed loans that are often bundled into complex securities.
The opacity of these financing arrangements poses a dual threat. First, it makes it difficult for regulators and market participants to gauge the true level of risk exposure. When debt is hidden behind special‑purpose vehicles or layered within multi‑tranche structures, the underlying leverage can be vastly underestimated.
Second, the reliance on future AI‑driven revenue streams—many of which are still speculative—means that companies may be counting on earnings that never materialise. If the anticipated returns fall short, firms could find themselves unable to service their debt, leading to defaults that ripple through the banking sector and beyond. To illustrate the potential fallout, the BIS chief highlighted several plausible scenarios. One involves a sudden slowdown in AI adoption due to regulatory push‑back, data‑privacy concerns, or a breakthrough in competing technologies.
In such a case, companies that have heavily leveraged themselves to fund AI projects could experience a sharp decline in cash flow, prompting a wave of loan restructurings. Another scenario envisions a broader macro‑economic downturn that reduces corporate profits across the board, making it harder for even well‑performing firms to meet debt obligations tied to AI investments. Both scenarios share a common denominator: the systemic risk that arises when a large swath of the corporate sector is simultaneously exposed to similar, poorly understood debt structures. The BIS chief warned that this could manifest as a credit crunch, where banks tighten lending standards, liquidity dries up, and asset prices—especially those linked to technology equities—plummet.
The resulting feedback loop could exacerbate the original shock, amplifying its impact on the real economy. In response to these concerns, the BIS chief called for a coordinated policy approach. He urged central banks, financial supervisors, and international standard‑setting bodies to enhance transparency requirements for AI‑related financing. This could include mandating detailed disclosures about the purpose of borrowed funds, the expected cash‑flow projections, and the underlying assumptions about AI‑driven profitability.
Additionally, he suggested that stress‑testing frameworks be updated to incorporate scenarios where AI projects underperform, thereby allowing banks to assess their resilience under more realistic conditions. The chief also emphasized the importance of prudent corporate governance. Board members and senior executives should critically evaluate AI investment proposals, ensuring that they are grounded in solid business cases rather than hype.
Independent auditors and rating agencies, he noted, need to develop expertise in assessing the unique risks associated with AI capital projects, including technology obsolescence, talent shortages, and ethical‑regulatory challenges. While the warning is sobering, the BIS chief also acknowledged the transformative potential of artificial intelligence. When deployed responsibly and funded with a clear understanding of the associated risks, AI can drive productivity gains, foster innovation, and create new markets. The key, he argued, is to balance enthusiasm with disciplined financial practices, thereby safeguarding both the stability of the financial system and the long‑term benefits of AI for society.
In summary, the BIS chief’s message serves as a timely reminder that the allure of AI should not eclipse the fundamentals of sound financial management. By learning from past bubbles, insisting on greater transparency, and embedding rigorous risk‑assessment into AI investment decisions, policymakers and business leaders can help ensure that the AI revolution proceeds without triggering a new wave of systemic instability.