AI Agents May Drain Banks’ Cheap Deposits and Raise Funding Costs, Slok Warns

AI agents that automatically optimize household cash could gradually pull deposits away from traditional banks, Apollo Chief Economist Torsten Slok warned in a Sunday research note.

Writing under the title “Is an Agentic bank run coming?”, Slok said AI-powered financial assistants could monitor household balances and automatically transfer money into accounts offering better returns. The process could create a gradual shift in deposits rather than a traditional sudden bank run.

Slok cited Meta’s personal agent Muse and other agentic AI assistants as examples of technology that could eventually handle these decisions. Such systems could move cash out of checking accounts paying the roughly 0.1% national average and into higher-interest alternatives.

Slok, a widely read Wall Street investment economist and partner at Apollo Global, said the difference in yields could make automated cash optimization attractive to consumers. Apollo manages about $1 trillion in assets.

He pointed to Revolut, SoFi, Varo, LendingClub and Wealthfront, which offer annual deposit rates ranging from 3.3% to 5%. A $10,000 balance earning those rates could generate approximately $330 to $500 in interest each year, compared with only about $10 in a checking account paying 0.1%.

The potential disruption stems from banks’ reliance on relatively inexpensive deposits to fund lending. If AI agents continuously search for better yields and move household cash accordingly, banks could lose part of their low-cost funding base and face increased financing costs.

“If every household used AI agents to optimize the return on their cash balances, banks could lose a large share of the cheap deposits they rely on to make loans,” Slok wrote, adding that the resulting pressure could extend across the financial system.

Automated Money Management

Agentic finance describes AI that can execute tasks on behalf of users rather than simply respond to instructions or provide recommendations.

A financial agent could monitor balances continuously, compare rates offered by different institutions, shift idle funds into higher-yield accounts and move money back when bills are due.

The potential market is difficult to size, with estimates varying significantly. Mordor Intelligence estimates agentic AI in financial services at $7.78 billion in 2026 and projects it will reach $43.52 billion by 2031. MarketsandMarkets estimates the narrower AI agents segment at about $845 million in 2025.

Stablecoins Offer a Payment Option for AI

Crypto infrastructure could also play a role as autonomous financial agents become more widespread.

Coinbase’s x402 protocol allows AI agents to make payments for online services using stablecoins, with transactions completed within seconds and without requiring a traditional account, card or individual human approval.

The x402 Protocol has reportedly processed approximately 188 million to more than 205 million transactions in total and has around 69,000 active agents.

Cloudflare, Google, Visa, Mastercard, AWS, Circle and Stripe are among the companies participating in the x402 Foundation, which is governed by the Linux Foundation.

Nate Geraci, co-founder of the ETF Institute, has previously argued that artificial intelligence and cryptocurrency could both challenge established parts of the traditional banking model.