The AI trade started with Nvidia Corp. (NASDAQ:NVDA) alone. It then spread broadly to chipmakers, memory producers, and the cooling and power gear that keeps data centers running. Energy — nuclear, wind, solar — is the leg most investors are watching next.
Almost no one is watching banks. A sector that seems unrelated to artificial intelligence is quietly becoming one of its clearest expressions.
Federal Reserve data published Friday shows commercial and industrial loans, the money banks lend businesses for equipment, expansion and day-to-day working capital, stood at $2.933 trillion in the week ending Aug. 19.
That is 10% higher than a year earlier and the fastest annual pace since March 2023.
Business Lending Is The Only Line Still Accelerating
Bank of America Corp. (NYSE:BAC) analyst Ebrahim Poonawala said the acceleration is concentrated in exactly that category.
Total bank loans grew 7.5% year over year in the latest H.8 report, the Fed’s weekly snapshot of what commercial banks own and owe, up from 7.1% at the start of the third quarter.
Commercial and industrial lending rose from 8.6% to 10% and was the only major category still speeding up on its six-week trend measure.
Residential mortgages, commercial real estate, credit cards and auto loans all decelerated.
Lending to non-depository financial institutions, meaning private credit funds and other non-bank lenders, remained the fastest-growing line at 19.4%, though it slowed as well.
The funding side is holding. Deposits rose 6.7% year over year while borrowings fell 2.4%, and the mix tilted slightly toward cheaper core accounts. Poonawala said pressure on funding costs is still unlikely to ease.
“AI buildout is going to be pretty extraordinary, and banks want a piece of it,” Poonawala said.
Wall Street Gets One Version, Main Street Gets Another
Nvidia set the scale of the opportunity last week. The company said revenue of $96 billion more than doubled from a year earlier and guided to roughly 70% growth in fiscal 2028 even with supply still constrained.
Someone has to write the checks.
Poonawala said the largest banks capture that directly, by arranging debt financing for data center projects, underwriting initial public offerings and managing the wealth the cycle creates.
Regional banks capture it indirectly.
Most are nowhere near a hyperscaler credit line, but the construction crews, electrical contractors, equipment dealers and landlords around a new data center all borrow locally.
Bank of America’s economists estimate roughly a third of US growth in the first half of 2026 came from artificial intelligence capital spending, 60 basis points out of 1.8% annualized.
The Market Is Starting To Notice
Bank stocks spent most of this year behind the market. Financial stocks — as tracked by the Financial Select Sector SPDR Fund (NYSE:XLF) — are up 6.1% in 2026, against 12.8% for the S&P 500, as tracked by the SPDR S&P 500 ETF Trust (NYSE:SPY).
The last three months say something different.
XLF has gained nearly 12% since the end of May.
Regional banks, as tracked by the SPDR S&P Regional Banking ETF (NYSE:KRE) are up 5.8%.
Both are beating the S&P 500 and tech, as tracked by the Invesco QQQ Trust (NASDAQ:QQQ)
The Part Nobody Has Priced Yet
There is a reason business borrowing is growing this fast, and it is not entirely comfortable.
The Bank for International Settlements, which serves as a bank for the world’s central banks, published research this year on how the AI buildout is being funded.
Borrowing that does not go through banks or the bond market comes from private credit, meaning investment funds that negotiate a loan directly with a company and hold it until maturity.
Loans outstanding from those funds to artificial intelligence companies have risen from near zero to more than $200 billion, accounting for close to 8% of all direct lending.
The BIS estimates the figure could reach $300 billion to $600 billion by 2030.
Image: Shutterstock
Login to comment