[Financial Planning for the 100-Year Life] AI Rally: Innovation or Bubble?

Artificial Intelligence (AI) now stands at the center of the global economy and financial markets. As investments in AI infrastructure—including semiconductors, data centers, power grids, and network equipment—expand to record levels, the corporate value of U.S. stock markets as well as Korean semiconductor companies has soared. However, debates over the sustainability of this trend have recently intensified.

The Bank for International Settlements (BIS) recently warned that the ongoing AI investment boom could eventually lead to a "CAPEX bust." The National Bureau of Economic Research (NBER) pointed to another risk of AI investments: the vulnerability in funding structures. This suggests that not only the technological success but also the resilience of the supporting financial system must be considered.

BIS’s concerns focus not on AI technology itself, but on the speed of investment. The world’s five largest hyperscalers are expected to invest more than 1 trillion dollars in building AI infrastructure in 2025 and 2026—a scale without historical precedent. However, if the anticipated returns fail to materialize, companies will be forced to cut back on capital expenditures. In that process, the entire AI supply chain—including semiconductors, servers, power equipment, and network devices—could contract simultaneously.

The NBER’s recent research highlights risks from a financial perspective. Private credit funds with quasi-liquidity, which have grown rapidly and now manage assets exceeding 300 billion dollars, have become important sources of funding for AI data centers and infrastructure investments. These funds hold long-term, illiquid loans but allow investors to redeem their shares on a quarterly basis. While this appears unproblematic under normal conditions, if markets become volatile, funds may need to make fire sales of loan assets or increase borrowing to meet redemption requests. As a result, asset prices may fall and leverage may rise at the same time, creating greater losses for remaining investors and thereby incentivizing further redemptions. The NBER analyzed that this structure exposes vulnerabilities similar to a bank run.

This is closely tied to AI investment, as a significant portion of AI infrastructure funding is raised through the private credit market. If the profitability of AI projects falls short of expectations, both investment cutbacks and fund withdrawals may occur simultaneously, setting off a vicious cycle that amplifies financial market instability. On the other hand, Wall Street emphasizes the winner-takes-all nature of the AI market. Leading big tech companies such as OpenAI, Microsoft, and Google believe that slowing investment now is equivalent to forfeiting future market leadership.

However, the focus of financial markets is gradually shifting. Investors are now placing greater importance on whether AI can generate cash flows and profits commensurate with massive investments, rather than simply asking if AI will change the world. Going forward, it will be necessary to monitor not only the growth rate of hyperscaler capital expenditures, but also AI service revenues, GPU utilization rates, HBM prices, data center utilization, free cash flow, and the health of the private credit market that underpins AI investments.

Although the AI revolution is likely to continue, the success of a technological revolution does not automatically guarantee success for investors. BIS has warned of a potential collapse in capital expenditures triggered by overinvestment, while NBER has shown that the private credit market could become a new source of financial instability in this process. Innovation will continue, but the market will ultimately favor performance over vision, and cash flow over growth. The success or failure of the AI supercycle will depend not on technology alone, but on the ability to secure both profitability and financial stability. Now is the time to be wary of excessive optimism about AI.

Kim Youngik, Adjunct Professor at Hanyang University Future Talent Education Institute

This content was produced with the assistance of AI translation services.