Bank of America's $800 Million AI Bet May Not Survive Its Earnings Call

By Grace Kim

Bank of America CEO Brian Moynihan put a number on his bank's AI program on Sept. 14, 2026: 130 to 140 projects, costing $400 million, returning $800 million. He said it at a Barclays conference, not an earnings call.

On Oct. 14, when BofA reports third-quarter results, investors will find out whether the $800 million survives the leap to the bank analyst line.

How rare is Moynihan's number?

Milos Maricic, founder of SPEC Research, and his team read 919 earnings calls from 60 major U.S.-listed financial firms since January 2023. Among them, only one firm put a specific dollar figure on what AI saved or earned: S&P Global, about $9 million a year from its Kensho Scribe tool, named on a February 2023 call. Analysts asked for such a number 29 times. They never got one.

Maricic widened the lens. Across 100 major U.S. companies outside tech, communications and finance, only FedEx has named an AI dollar figure on a call. In December 2024, FedEx credited "a new AI image capture process" with a surcharge benefit of "over $180 million annualized." That makes two companies in 160.

Two companies in 160 have named an AI dollar figure on a U.S. earnings call. Bank CEOs have preferred ambiguity.

Why won't the CEOs say it?

The reluctance shows up in the quotes. Goldman Sachs CEO David Solomon told analyst Mike Mayo in January 2026: "We don't have the full confidence to put information out publicly." JPMorgan CEO Jamie Dimon came closest in June 2025 at a Morgan Stanley conference: "we do know we get benefits around $2 billion to pay for the $2 billion." JPMorgan has not repeated that figure on any of its 13 earnings calls since mid-2023.

On JPMorgan's October 2025 call, CFO Jeremy Barnum told analysts that proving AI savings "turns out to be a very hard thing to do, hard to prove." That phrasing has become the default answer when analysts ask.

CEOs will name an AI number at a conference but won't repeat it under analyst questioning. The conference crowd gets a figure. The shareholders get "hard to prove."

What does the calendar say?

Maricic flags three concerns. The first is timing. AI investment pays off on what economists call the productivity J-curve, and a bank that can't put a number on AI returns is sitting where the curve says it should be. The history rhymes: in 1987 Robert Solow wrote, "You can see the computer age everywhere but in the productivity statistics." A 1994 survey of 99 completed reengineering projects found 67% produced "mediocre, marginal, or failed results." Yet by the end of that decade, IT accounted for about two-thirds of the pickup in U.S. productivity growth.

Banks that can't price AI today may simply be early. That's how every general-purpose technology has worked.

Where does the edge go?

The second concern is the edge. China's long-only quant funds beat human stock pickers by 20 percentage points in 2025, a run Bloomberg tied to AI adoption. Their edge over the index halved by April 2026. The funds then lost an average 17% in July 2026.

Even real, materialized AI advantage fades fast. The first adopters win a narrow window, and the window closes.

What's the cost angle?

The third is cost. AI cost came up on 53% of the financial calls Maricic's team read last season, up from 3% in late 2022. Across 40 engagements Maricic has led with financial firms since 2025, optimization has typically removed about 40% of annual AI spend. The most intense users now cut token bills and shop suppliers. The age of burning tokens to show adoption, he says, is already over.

Users are already cost-sensitive. Value will arrive late. It will accrue to a narrow segment of adopters. Then it will erode.

What should investors listen for next week?

Next week is the test. JPMorgan, Goldman Sachs, Wells Fargo and Citigroup report on Tuesday, Oct. 13. Bank of America and Morgan Stanley follow on Wednesday, Oct. 14.

Maricic's instruction: listen for a specific dollar figure for what AI saved or earned, said on the call itself. If an analyst asks, listen for whether the answer is a percentage, a headcount, or "hard to prove." If Moynihan repeats his $800 million, Bank of America will be the first of the 21 banks Maricic tracks to put a number on a call.

If the $800 million makes it onto the Oct. 14 call, Moynihan will have moved U.S. bank disclosure past where his peers have been willing to go. If it doesn't, the "hard to prove" answer keeps winning — and the AI savings story stays where investors can't price it.