BIS warns that surging AI investments pose significant financial risks, potentially leading to global economic repercussions if returns fall short.
The Bank for International Settlements (BIS) has raised alarms regarding the financial stability risks linked to the rapid increase in artificial intelligence (AI) investments. The organization cautions that a downturn in the AI sector could have widespread effects on the global economy.
Pablo Hernández de Cos, General Manager of BIS, expressed concerns during the Global Fintech Fest in India on September 10, noting that AI is profoundly influencing global macroeconomic dynamics. This technological surge is reshaping demand, supply, and financial markets, complicating central banks' roles.
Projected AI Investments Reach $4 Trillion
BIS forecasts that the top five tech companies will invest over $1 trillion in AI by 2025-2026. Experts predict that worldwide AI investments could soar from the current $500 billion to between $3 and $4 trillion by 2030.
Hernández de Cos highlighted that a significant portion of these investments is being financed through debt rather than corporate profits, raising concerns about the financial system's vulnerability if AI returns do not meet expectations.
Concerns Over Potential Investment Bubbles
The BIS chief warned of inflated valuations in the AI sector, which are concentrated among a few major players. Companies are increasing capital expenditures while increasingly relying on debt. There are also worries about circular financing, where chipmakers and cloud providers invest in AI firms that, in turn, commit to purchasing their services, creating intricate financial ties that can be hard to assess.
Hernández de Cos cautioned that if AI investment returns fall short, it could trigger a significant decline in investment activity, shifting the current investment boom into a downturn. The competitive race among AI firms also poses risks of excessive spending.
The BIS compared the current landscape to previous tech investment bubbles, such as the railway mania of the 1800s, the electrification boom of the 1920s, and the late 1990s dot-com bubble, where real technologies attracted more capital than future earnings could support.
AI's Potential for Growth
Despite these risks, BIS recognizes the potential of AI to enhance productivity significantly. Research cited by Hernández de Cos indicates that generative AI could improve productivity for certain tasks by 10% to 65%, with time savings of 20% to 50%. Long-term projections suggest a rise in aggregate productivity by about 0.5 percentage points annually.
However, AI may also reduce the demand for workers in routine cognitive roles, with early signs of displacement already emerging in fields like customer support and programming. Approximately 80% of companies are reportedly considering automation in their operations.
Challenges for Central Banks
AI's influence extends to monetary policy, complicating central banks' ability to forecast economic performance. The technology affects productivity, demand, supply, and financial markets, making it difficult to gauge potential output and interest rates.
BIS is analyzing various future scenarios: one where AI drives sustainable productivity growth, another where it creates self-reinforcing growth, and a third scenario involving a "demand trap" where automation shifts income away from labor, weakening consumer demand.
Additionally, BIS issued a warning about cyber risks associated with AI, noting that while AI can aid both attackers and defenders, attackers may have an advantage due to the need to exploit only one vulnerability.
Despite the caution, Hernández de Cos did not claim that the current investment boom would inevitably lead to a crisis. He emphasized that the rapid scale of AI investments calls for careful consideration, with the technology's impact dependent on its capabilities, regulatory frameworks, competition, and institutional quality.
Ray Dalio, founder of Bridgewater Associates, also commented in August, likening the current AI enthusiasm to bubbles seen in 1929 and 2000, highlighting mechanisms that could contribute to a potential collapse.
