AI agents expose a bank funding risk in idle cash flows

AI assistants could steer cash from 0.1% checking accounts to 3.3%-5.0% alternatives, testing banks' deposit base. Apollo flagged the risk.

Apollo Global Management Inc. Chief Economist Torsten Slok warned that agentic software could make household cash more mobile than banks expect. In a Sunday note, he said broad use of AI agents to optimize cash returns could strip banks of a large part of their low-cost funding and strain lending across the financial system.

Muse sharpens deposit debate

Meta Platforms Inc.'s Muse has become the clearest example in the debate over how AI assistants might change consumer finance. Slok cited the product as part of a wider shift toward tools that can search, compare and act for users rather than merely answer questions.

The concern is not that a chatbot tells a depositor a better rate exists. Ebrahim Poonawala, an analyst at Bank of America Corp., said in a note last week that an agent can identify spare cash, compare available yields and move funds, weakening the inertia that has helped banks retain cheap deposits.

Higher yields test checking inertia

Slok said AI agents may direct cash toward products yielding between 3.3% and 5.0%, compared with the 0.1% national average for checking accounts. That spread matters because deposits that pay little interest are a core funding source for banks that lend to households and companies.

He also pointed to fintech firms offering rates above those available at many traditional banks. SoFi Technologies Inc. offers 4.5% on deposits, according to Slok, while Happen Inc., previously LendingClub, was included among companies competing for cash with higher-rate products.

The pressure is familiar in one form and new in another. Online-only banks such as Goldman Sachs Group Inc.'s Marcus and Ally Financial Inc. have used lower branch costs to offer some of the industry's higher deposit rates, but AI agents could reduce the effort consumers need to move money.

Bank shares absorb AI concern

Bank stocks moved lower last week as investors weighed whether AI tools could make deposits more rate-sensitive. JPMorgan Chase & Co. and Wells Fargo & Co. each fell more than 3% on Tuesday, after analysts circulated notes on the risk to consumer-inertia business models.

The mechanism is straightforward if the adoption case holds. If consumers let agents scan cash balances and execute transfers, banks with large low-rate deposit bases would face higher funding costs, while fintechs and online banks offering higher rates could receive a larger share of idle cash.

If usage remains narrow, the effect may stay closer to the earlier competition from online savings products. In that case, the macro effect would likely be limited to deposit pricing at the margin, while large banks would retain the customer relationships and scale that have historically slowed cash movement.

If agentic AI reaches mass adoption, the strain would run through three channels: less stable bank funding, higher competition for deposits and tighter margins for lenders that depend on cheap checking balances. The main open question is whether consumers will trust AI assistants to move cash without reviewing each transaction themselves.