dblnews.

Clear, practical, independent coverage

A column by Sylvia Parrish

News

Why Wall Street Banks Are Still Struggling to Prove AI's Real Financial Value

Business Insider published a roundup this week titled "Everything we know about how biggest Wall Street banks are using AI," and I want to be honest with you: the fact that we still need an…

Sylvia Parrish, Chief Business Columnist·updated August 16, 2026

Why Wall Street Banks Are Still Struggling to Prove AI's Real Financial Value

Business Insider published a roundup this week titled "Everything we know about how biggest Wall Street banks are using AI," and I want to be honest with you: the fact that we still need an "everything we know" piece is itself the most interesting data point.

The bulge bracket has spent the better part of two years telling anyone with a press release that artificial intelligence is rewriting the plumbing of finance. Vendor deals, in-house builds, "thousands of use cases," the usual liturgy. Yet here we are in the second half of 2026, and the best a serious newsroom can offer is a compendium. No consolidated numbers. No audited productivity claims. Just fragments stitched together from earnings call color, vendor case studies, and the occasional LinkedIn post.

The shape of the haze

Let me translate this for you. When banks are genuinely winning on a technology bet, they don't hide the result in compliance theater. They trot out the operating leverage on an earnings call and let the analysts fight over the multiple. The fact that no major institution has produced a defensible, line-item P&L impact from generative AI tells you everything the roundup itself doesn't.

The pieces we do have are the easy ones: chatbots for retail customers, copilots for advisors, back-office summarization of compliance documents. None of that is the moat. None of that moves the needle on cost-to-income at a major universal bank. It's table stakes dressed up as transformation, and the firms know it. They just don't want you to know they know it.

What I'm actually watching

Here's what I'd want to see before I get excited. First, an honest disclosure on the operating expense line — not a "productivity" anecdote, but a real number that survives a 10-K review. Second, any signal that AI is changing the front office: actual pricing, actual deal selection, actual risk-taking rather than the back-office boilerplate. Third — and this is the one nobody on the industry PR circuit wants to touch — what happens to the junior banker class of 2027 and 2028 when the workflow they've been trained on no longer exists.

Until then, "everything we know" remains a confession, not a report. The roundup is the press holding a mirror up to an industry that would very much like to move on before anyone checks the receipts.