Concerns over an artificial intelligence investment bubble are shaking the U.S. stock market once again. Recently, the Nasdaq index plummeted by more than 2%, concentrating selling pressure on semiconductor and AI-related stocks. In particular, the funding burden is growing for companies building data centers relying heavily on debt.
In this climate, attention is turning to major Wall Street banks that are lending money and arranging financing for AI infrastructure investments. Summarizing the conference calls held last week by six major banksβJPMorgan Chase, Goldman Sachs, Citigroup, Bank of America (BofA), Wells Fargo, and Morgan Stanleyβa common perspective emerged regarding the next phase of the AI infrastructure investment cycle. In short, while AI investments will continue for the time being, funds cannot be unconditionally granted to every project.
Some banks predicted that AI infrastructure investments would persist based on current computing power and power shortages. Ted Pick, CEO of Morgan Stanley, stated that AI capital expenditures are still in the ‘early stages’ and suggested that annual investments of around $1 trillion by major cloud companies could continue over the next 5 to 10 years. He expressed a strong willingness to actively engage in fundraising and transaction advisory, calling AI a ‘once-in-a-generation technological shift.’
However, they did not conclude that the scale of investment would grow indefinitely. Periods of potential overinvestment and phases where momentum slows down due to power shortages could alternate. Jane Fraser, CEO of Citigroup, also forecasted that ‘high interest rates will not dampen the necessity for AI investment.’ She noted that there are no signs of a slowdown in investment across the AI ecosystem, and that computing power and electricity remain bottlenecks not only in the United States but globally.
Strict Lending Standards and Risk Management
Expectations for AI investment did not translate into unconditional funding support from Wall Street. Wells Fargo stated that it checks not only the permitting and power procurement of data centers but also who leases the facilities and what guarantees are in place. The bank explained that it assesses borrowers’ repayment capacities and places strict limits on the scope of financing it provides.
Charlie Scharf, CEO of Wells Fargo, evaluated that the AI investment cycle itself has become a complex macroeconomic variable. While demand is exploding across semiconductors, network equipment, power generation, cooling, and storage systems, he warned that massive debt and price inflation pressures in the supply chain must also be guarded against. However, he added that the investment benefits still outweigh the risks.
Interest Rates and Macroeconomic Pressures
As U.S. Treasury yields remained high, interest rate forecasts were also a major topic of discussion. Executives focused more on the causes and speed of the increases rather than the absolute rate level. David Solomon, CEO of Goldman Sachs, stated, ‘From the perspective of the operating environment, a 10-year Treasury yield of 5% is not frightening,’ but warned that ‘if rates rise rapidly, the speed of building data centers and financing conditions could change.’ He also pointed out that if funds concentrate heavily on massive AI investments, capital available for other investments could shrink, and borrowing costs could rise.
When asked about the possibility of the U.S. 10-year Treasury yield hitting 6%, Jamie Dimon, CEO of JPMorgan Chase, replied, ‘We are not ruling out that scenario.’ He also pointed out that the impact of rising interest rates on the economy depends on whether they are driven by investment demand and productivity improvements or by stubborn inflation.
