In a recent address to policymakers and market participants, the head of the Bank for International Settlements (BIS) sounded an alarm about the rapid escalation of capital expenditures in the artificial intelligence (AI) sector. While the surge in AI‑related spending reflects genuine technological breakthroughs and the promise of transformative applications, the BIS chief emphasized that much of this investment is being financed through opaque, often undisclosed debt structures. This hidden leverage, he argued, could sow the seeds of a systemic financial crisis if not addressed promptly. The warning was framed against a backdrop of historical episodes in which speculative fervor outpaced underlying fundamentals.
The BIS official, Pablo Hernandez, invoked two well‑known periods of exuberant investment: the 19th‑century railway boom and the late‑1990s dot‑com bubble. In both cases, investors poured massive sums into projects that were touted as the next great engine of growth, yet many of these ventures were underpinned by fragile financing arrangements that later unraveled. The railway era saw a proliferation of lines built on optimistic traffic forecasts, often financed by complex bond issuances that were difficult for regulators and investors to assess. When demand fell short, a wave of defaults rippled through the financial system.
Similarly, the dot‑com era was marked by a flood of venture capital and public market money into internet startups with lofty valuations but limited revenue streams, many of which relied on convertible notes and other forms of hidden debt that later forced painful write‑downs. Drawing on these analogies, Hernandez warned that the current AI capital‑intensive environment bears striking resemblances. Companies across sectors—from cloud service providers to autonomous‑vehicle manufacturers—are committing billions to develop proprietary models, acquire specialized hardware, and recruit top talent. However, a substantial portion of this outlay is being funded through private placements, off‑balance‑sheet vehicles, and other non‑transparent mechanisms that obscure the true debt burden.
In many instances, the terms of these financing arrangements are not disclosed to shareholders or regulators, making it challenging to gauge the level of leverage embedded in the AI ecosystem. The BIS chief highlighted several channels through which this hidden debt could amplify systemic risk.
First, the concentration of AI‑related borrowing among a relatively small group of high‑growth firms creates a network of interlinked exposures. If a leading AI player experiences a setback—such as a regulatory clampdown, a data‑privacy scandal, or a sudden technology shift—the resulting financial strain could cascade to lenders, suppliers, and downstream investors.
Second, the lack of transparency hampers the ability of central banks and supervisory authorities to monitor aggregate risk levels. Traditional stress‑testing frameworks rely on clear data about asset quality and liability structures; when debt is shrouded in complex contracts, these tools become less effective. Third, the rapid pace of AI adoption may outstrip the capacity of financial markets to price risk accurately, leading to misallocation of capital and inflated asset prices.
To mitigate these dangers, Hernandez called for a coordinated policy response. He urged regulators to enhance disclosure requirements for AI‑related financing, ensuring that both the amount and the terms of debt are visible to market participants. This could involve mandating that companies report off‑balance‑sheet obligations, convertible securities, and other hybrid instruments in a standardized format. Additionally, the BIS suggested that supervisory bodies develop sector‑specific macro‑prudential tools, such as counter‑cyclical capital buffers for banks heavily exposed to AI borrowers.
By adjusting capital requirements in response to rising leverage, regulators can dampen the incentives for excessive borrowing during boom periods. Hernandez also emphasized the role of market discipline.
Investors, he noted, should conduct deeper due diligence on the financing structures of AI firms, looking beyond headline revenue projections to assess the sustainability of debt loads. Credit rating agencies, meanwhile, need to refine their models to capture the unique risk profile of AI‑centric businesses, including the potential for rapid obsolescence and regulatory uncertainty.
Beyond regulatory measures, the BIS chief advocated for a broader dialogue among stakeholders—including technology firms, financial institutions, and academic researchers—to develop best practices for responsible AI investment. This collaborative approach could help establish industry standards for transparency, risk management, and ethical deployment of AI technologies. In concluding his remarks, Hernandez warned that while AI holds the promise of significant productivity gains and societal benefits, the path to realizing these gains must be paved with prudent financial stewardship.
Ignoring the hidden debt that fuels the current AI spending spree could lead to a correction that not only harms individual firms but also threatens the stability of the global financial system. By learning from the lessons of past bubbles and implementing robust oversight today, policymakers can help ensure that the AI revolution proceeds on a solid and sustainable foundation.