Could AI End Banks' Cheap Deposit Advantage?

The gap between what banks pay savers and what the Federal Reserve pays on reserves has long been one of the quietest profit engines in finance. For most of the past decade, the average U.S. savings account has yielded a small fraction of the benchmark federal funds rate, even as that rate climbed above 5% in 2023 and 2024. That spread exists because moving money is a hassle—and banks have built their funding models around the reasonable assumption that most customers simply won't bother.

That assumption is now under pressure. According to reporting from The Wall Street Journal, AI agents—software programs that can act on a customer's behalf without constant human supervision—could soon move cash automatically into accounts offering higher interest rates. If that happens, the cheap deposits that underpin bank profitability could start to evaporate. The question is not whether the technology works. It is whether enough customers will use it to matter.

How Banks Profit from Customer Inertia

Start with the math. The FDIC's national rate cap for savings accounts has hovered near 0.4% to 0.5% for much of the past two years, while the Fed's target range sat between 4.25% and 5.5% during the same stretch. That is a spread of roughly 4 percentage points—money that banks keep rather than pass to depositors. On a $10,000 balance, the difference between 0.4% and 4.4% is about $400 a year. Multiply that across hundreds of billions in deposits, and the scale of the subsidy becomes clear.

Not every bank pays rock-bottom rates. Online-only institutions and money-market funds routinely offer 4% or more. But the majority of Americans keep their primary checking and savings accounts at the same institution they have used for years, if not decades. Bankrate's annual savings surveys have consistently found that fewer than half of savers have switched banks in the past year, and a substantial share report earning less than 1% on their savings. NerdWallet's consumer research has reached similar conclusions: convenience, direct deposit habits, and automatic bill pay create switching costs that no interest-rate comparison can easily overcome.

Behavioral finance has a name for this. It is called status quo bias—the tendency to stick with a default even when a better option is plainly available. Banks did not invent it, but they have learned to harvest it. Low-cost core deposits, the industry's term for the sticky checking and savings balances that customers rarely move, are among the most valuable assets on a bank's balance sheet. They are cheaper than wholesale funding, less sensitive to market panic, and—crucially—insurable up to $250,000 per depositor.

What AI Agents Can Do That Customers Won't

A human being with a savings account and a laptop could move money to a higher-yielding account in about fifteen minutes. Most don't. An AI agent could do it in seconds, continuously, without being asked.

The WSJ report frames this as a genuine structural threat: AI agents capable of monitoring rates, comparing accounts, and executing transfers could strip away the friction that has protected bank margins for generations. Unlike a rate-shopping website, which still requires a human to act, an agent could be programmed with a simple rule—keep my cash in the highest-yielding FDIC-insured account—and then execute it indefinitely.

That changes the economics of deposit gathering. Today, a bank can count on a large share of its customers to leave money in place even when better rates exist elsewhere. Tomorrow, a customer with an AI agent might hold only the minimum required to pay bills, sweeping the rest into whichever institution pays the most that week. Deposits become a commodity, and the bank's cost of funds rises accordingly.

The Threat to Bank Funding Models

Analysts at Keefe, Bruyette & Woods and other research firms have long estimated that net interest income—the difference between what banks earn on loans and pay on deposits—accounts for roughly half to two-thirds of total revenue at a typical commercial bank. A meaningful share of that comes from the spread on low-cost core deposits. If those deposits reprice upward, the squeeze is direct.

Consider the mechanics. A bank funds a 6% auto loan partly with deposits paying 0.5%. That is a 5.5-point spread. If AI-driven competition forces the bank to pay 4% to keep the same deposit, the spread collapses to 2 points. The loan doesn't change. The borrower doesn't change. Only the funding cost moves—and it moves against the bank.

The systemic implication is more subtle but potentially more serious. Stable deposits are not just cheap; they are predictable. Banks rely on that predictability to manage liquidity and meet regulatory requirements. If deposits become rate-sensitive and prone to rapid movement—as they did during the 2023 regional banking crisis, when customers fled Silicon Valley Bank and First Republic in days—the industry's funding base becomes less stable. AI agents could accelerate that dynamic, turning what was once a slow drift into an instantaneous reallocation.

Which Banks Face the Most Exposure

Not all banks are equally vulnerable. The largest institutions—JPMorgan Chase, Bank of America, Wells Fargo—have enormous branch networks, deep customer relationships, and a mix of fee income that cushions them. They also tend to pay among the lowest savings rates, which means they have the most to lose if customers start moving.

Regional banks sit in a more precarious position. They typically lack the scale and diversification of the megabanks and rely more heavily on deposit spreads for profitability. Many are still recovering from the funding shocks of 2023. An AI-driven shift toward higher-yielding accounts would hit them first and hardest.

Online banks and money-market funds, by contrast, would be natural beneficiaries. They already compete on rate, and an AI agent searching for the best yield would find them quickly. That dynamic could accelerate a migration of deposits away from traditional branch-based banking—a trend that has been underway for years but could now speed up dramatically.

How Banks Might Respond to AI-Driven Deposit Flight

Banks have three broad options. The first is to pay more. That is the simplest and the most expensive. If institutions raise savings rates to retain deposits, net interest margins compress and earnings fall. Some banks can absorb that. Others cannot.

The second is to make switching harder—not through outright barriers, which regulators would frown upon, but through product design. Bundled accounts, tiered rates that reward loyalty, and perks tied to long-term relationships all raise the cost of leaving. Expect to see more of this.

The third is to fight fire with fire. If customers will use AI agents to manage their cash, banks can offer their own. A bank-sponsored agent that sweeps idle balances into a higher-yielding internal account keeps the money inside the institution—and preserves at least part of the spread. Whether customers trust their bank to act as their agent is an open question, but the logic is compelling.

What This Means for Everyday Savers

For savers, the arrival of AI-driven deposit management is unambiguously good news—in theory. A world where software automatically routes cash to the highest-yielding insured account is a world where the gap between what banks pay and what the Fed pays narrows. That is money in consumers' pockets.

In practice, the benefits will depend on adoption. If only a small fraction of customers use AI agents, banks will keep paying low rates to everyone else. If adoption becomes widespread, the economics shift. History suggests adoption curves are slower than technologists predict but faster than incumbents expect.

There is also a regulatory question. Deposit insurance exists to protect savers, not to guarantee bank funding costs. If AI agents make it easy to move large sums quickly, regulators may need to rethink liquidity rules—or accept that deposit funding is simply less stable than it used to be.

The deeper point is this: the bank deposit business has always been a bet on human inertia. AI does not eliminate that inertia, but it removes the friction that sustains it. Banks that treat customer laziness as a permanent asset are making a bet that may no longer pay off.

Source: WSJ.com: Markets