Banks Face AI Payoff Test as Deloitte Sees Growth in 2027

U.S. banks head into 2027 on solid footing, but the payoff from their heavy spending on artificial intelligence is coming under closer scrutiny. Deloitte’s 2027 Banking and Capital Markets Outlook says AI and digital assets are reshaping how banks operate and where they find growth, even as interest rates stay high and consumers feel the strain.

Deloitte expects the U.S. economy to grow 2.1% in 2027, a favorable backdrop for lending. It forecasts the effective federal funds rate will remain near 3.63% through 2027 and the 10-year Treasury yield will stay above 4.5%.

That environment points to modest gains in net interest income, which Deloitte projects will rise to about $503 billion in 2027 from an estimated $472 billion this year. Loan growth should stay positive but uneven. Commercial and industrial lending is expected to hold up as companies keep spending on AI, while consumer lending slows under high rates and soft wage growth. Fee income from dealmaking, equity trading and wealth management could be a bigger growth driver.

Deloitte said banks will face pressure to show results from AI investments as computing costs rise. In a Deloitte survey, 74% of banking technology leaders said they expect to run private AI infrastructure by 2028. Deloitte estimates that a self-hosted system could cut costs by more than 50% over three years compared with paying per use through outside providers. It projects banks’ efficiency ratio, which measures expenses as a share of revenue, will improve to 59.3% in 2027.

AI also creates new lending opportunities. The external financing gap for AI infrastructure could reach $3 trillion to $5 trillion by 2028, according to estimates cited from the Federal Reserve Bank of Dallas. Deloitte cautioned that risks such as GPU depreciation and early termination of computing contracts don’t fit neatly into traditional credit models.

The report also expects tokenized deposits, bank stablecoins and tokenized securities to gain momentum next year, though stablecoins could draw deposits away from banks. A slowdown in AI spending remains a key risk, potentially curbing loan growth and weakening credit quality.