Wall Street has found another scarce asset to overpay for: people who can turn AI promises into systems a bank can actually use.
The headline number is hard to ignore. According to Bloomberg Technology, some AI consultants and advisers are charging Wall Street banks as much as $25,000 a day to help deploy, test and govern artificial intelligence inside institutions that cannot afford to get the technology wrong.
That is not a normal consulting fee. It is a shortage price. Banks have money, data, regulatory pressure and impatient executives. What they do not have enough of is people who understand both modern AI systems and the awkward reality of financial services, where a model output can affect trading decisions, compliance reviews, credit risk, client advice or reputational exposure.
This is why the day rates matter. They are not just a story about expensive experts. They are a market signal that banks are moving faster than their own talent pipelines can handle. The same firms that spent years building data teams, cloud platforms and automation programs are now discovering that generative AI requires a different mix of skills: model evaluation, workflow redesign, prompt governance, data permissions, hallucination controls and old-fashioned judgment about where automation belongs.
Reuters reported last week that JPMorgan is rolling out AI tools across its investment banking business globally, while Goldman Sachs, Citigroup, Bank of America and Morgan Stanley have been testing or using similar systems. That is the real backdrop. AI in banking is no longer a side experiment run by innovation teams. It is being pushed into the machinery of work.
There are plenty of AI engineers in the market, and there are plenty of financial services consultants. The premium is going to the smaller group that can sit between the two and make decisions that survive contact with a bank risk committee.
That distinction matters because banking is not a clean software environment. A generic chatbot that improves office productivity is one thing. A tool that summarizes client calls, drafts research, screens transactions or assists a banker on a live deal has a much higher burden. It needs controls, audit trails, escalation paths and clarity about who is accountable when the machine gives a confident but wrong answer.
McKinsey has estimated that generative AI could add $200 billion to $340 billion in annual value to banking, largely through productivity gains. Those numbers explain why senior executives are willing to pay up. If a bank believes AI can shave hours from pitch books, speed up code development, improve fraud detection or make client service more responsive, a $25,000 day rate looks less absurd. It becomes the price of getting there before competitors do.
But the same number also reveals how unready many institutions still are. If internal teams were fully prepared, banks would not need to rent outside judgment at such a premium. The rush for advisers suggests that many firms have the budget and the mandate before they have the operating model.
A new opening for AI-native boutiques
For startups and independent experts, this creates a more interesting opportunity than another thin AI wrapper. Banks are showing they will pay enterprise prices for expertise, not only for software subscriptions. That opens the door for AI-native boutiques that can help with evaluation, integration, compliance workflows and model selection without pretending to be a full-stack platform company from day one.
This looks a little like the early fintech consulting cycle, but compressed. After the financial crisis, banks paid heavily for regulatory technology, digital transformation and cloud migration help. Those markets took years to mature. AI is moving on a shorter clock because executives can already see the tools working in everyday tasks, and because competitors are talking openly about productivity gains.
Jamie Dimon said in a recent Bloomberg interview that JPMorgan is likely to hire more AI specialists and fewer traditional bankers as adoption accelerates. That is the second signal in this story. Banks are not only buying advice. They are changing the profile of the people they want inside the building.
That shift will not be clean. Financial institutions still need bankers who understand clients, markets and risk. They also need technologists who can build and monitor systems. The difficult part is the middle layer, people who can translate business judgment into AI workflows and know when a process should remain human. That is where outside advisers are finding their leverage.
There is also a practical warning here. Expensive consultants can accelerate adoption, but they cannot substitute for internal capability forever. A bank that depends too heavily on outsiders may launch tools faster while failing to build the habits needed to govern them. In finance, that gap can become costly quickly.
The likely next phase is more disciplined spending. Banks will keep paying for scarce expertise, but they will also push vendors and advisers to prove outcomes: reduced processing time, better controls, fewer manual reviews, faster client response or measurable risk reduction. The winners will be the experts who can show that AI is not a demo, but a working part of the bank.
For the broader market, the takeaway is simple. Wall Street is not debating whether AI matters anymore. It is fighting over the people who can make it useful. When banks start paying $25,000 a day for that judgment, the opportunity is no longer theoretical, and neither is the talent shortage.
Also read: A disputed METR graph is testing AI's benchmark economy • ClickUp's layoffs show AI is rewriting productivity software. • Palantir turns a London police setback into a public test of AI procurement
