AI Coding Agents Threaten Traditional Bank Profit Models
Cognition claims its AI agent can modernize legacy bank code in days, while Meta's Muse agent threatens bank deposit stability.
AI company Cognition claims its coding agent, Devin, can rewrite legacy COBOL bank code in days rather than months. This capability could significantly reduce technology expenditures for financial institutions, such as Chase Bank, which expects to spend $19.8 billion on technology in 2026 to modernize its data and application code.
While investor Bill Ackman argues these efficiencies will sharply cut costs, analyst Patrick Feeley suggests the savings may be offset by new AI-driven personal agents. Meta Platforms Incorporated launched Muse on September 8, a tool capable of reading accounts across 12,000 U.S. banks. This allows customers to easily move deposits from low-interest checking accounts to higher-yield alternatives, reducing deposit stickiness and increasing the cost of deposits for banks.
These competing agentic risks contributed to market volatility on September 22, when shares of major institutions including JPMorgan, Wells Fargo, and Schwab declined. The shift toward commoditized loans and more mobile deposits may prevent banks from retaining the financial benefits of AI-driven code modernization, potentially impacting the industry's current net interest margin of 3.32 percent reported by the Federal Deposit Insurance Corporation.
