Logic would suggest that the companies that use AI the most will inevitably reduce their human workforces. But new data analyses find the opposite is true, at least for now: While a few departments, including back offices and call centers, are starting to see hiring freezes or cutbacks, by and large, the companies spending the most on advanced AI are hiring more humans than those that spend less.
In a research report released Tuesday, Ramp and Revelio Labs analyzed AI spending data from 21,559 U.S. companies' card and bill pay data linked to their workforce records to get a picture of AI's effect on jobs, regardless of survey results or executive interviews. In separate research released earlier this month, PwC found U.S. banks' AI job postings are up 77.4% over 2025.
For the Ramp paper, researchers examined how much companies paid their AI vendors from January 2021 through February 2026. Unsurprisingly, they found that AI spending is highest in the information, finance and insurance sectors, and lowest in construction, healthcare, arts and entertainment and food services.
Among companies that spend the most on AI per employee, headcount rose 10.2% in the two years after AI adoption. They had a 12% rise in entry-level workers, a 7.7% increase in non-entry-level headcount and a 6.7% growth in managers and higher-level officers.
Companies that spent little or nothing on AI during the same period had no increase in headcount.
"If you are a young person entering the labor market, and you are choosing between otherwise similar firms, choose the one that is using AI," the researchers advised. "If you are an engineer worried that AI will eliminate engineering jobs, the firms adopting AI and hiring engineers faster, not slower. And if you are reading headlines where CEOs blame layoffs on AI, be skeptical."
The AI jobs banks are desperate to fill
PwC's 2026 AI Jobs Barometer examined more than 449,060,164 U.S. job ads and tracked changes in employer demand over time, including the number and types of roles advertised, the skills requested by employers, AI wage premiums and demand for specialist AI skills.
The firm found banks are hiring expert users more than creators of AI models. In 2025, AI user roles accounted for 88% of AI-related job listings in financial services, versus 12% for AI developer roles, according to the PwC study.
"Think of that as using [Microsoft] Copilot or a custom GPT versus someone who's actually developing AI-related models," Peter Pollini, financial services industry leader at PwC US, told American Banker.
Job postings that include artificial intelligence in the job description have risen as a total percent of financial services job postings from 3.9% to 6.1% over the last couple of years.
This is unsurprising, Pollini said. "If you continue to have a desire to AI-enable your firm, you will likely continue to need talent that knows how to build, maintain and monitor AI or agentic automated businesses," he said.
In late June, Lloyds Banking Group said it will be recruiting for almost 300 agentic AI-related roles, including data and AI scientists, engineers, responsible AI specialists and AI product managers, as it expands the teams building and deploying agentic AI across the company.
More than 700 bank employees already do this work. The London bank recently launched AI-based fraud detection agents that analyze payments in real time to identify and prevent scams. The bank plans to expand customer-facing tools such as an AI financial assistant that's already in use by 500,000 Bank of Scotland customers.
"AI is becoming an increasingly important part of how we support customers and how we work across Lloyds Banking Group," said Sharon Doherty, chief people and places officer, in a statement. "As we scale its use, our focus is on making sure it delivers real benefit in day-to-day roles — helping colleagues make better decisions and enabling us to provide faster, more effective and more personalized support for customers."
One reason banks are hiring people with AI skills is that they're rolling out hundreds of AI use cases that all have to be maintained and monitored, said Jeff McMillan, CEO of training and consulting firm McMillanAI. Up until February, he was head of firm-wide AI at Morgan Stanley.
"This doesn't just happen with an AI button," McMillan told American Banker.
Why AI efficiencies have not led to massive layoffs yet
One reason AI is not massively displacing people in banks yet, according to McMillan, is that so far banks are mostly deploying copilots or virtual assistants that help a person do their work, but don't do the whole job.
"Most of the AI that's been rolled out is solving for task-based work — it's recording a meeting, taking notes, booking travel, booking meetings," he said. "You do 20 things in the day, and it goes after two of them, and it makes them 30% better." This means copilots have little impact on the organization, McMillan said.
Employees may self-report that an AI copilot saved 48% of their day, but this is generally untrue, McMillan said. Typically these bots provide efficiencies of around 5%, he said.
The bigger opportunities, according to McMillan, are using AI for entire workflows, such as account opening, creating S-1 reports for IPOs and anti-money-laundering case reviews. "It's the same task done 40 times a day," he said.
Even in coding, where some companies are reporting 30% or 40% efficiencies from AI, "they're only going after the development side of coding," McMillan said. "They're not going after the requirements documentation, they're not [using AI for] the user acceptance testing, the synthetic data creation, the validation, all the paperwork." Often firms don't understand their workflows well, he noted.
Another reason AI is not eating large numbers of banking jobs yet is because banks have been in a favorable, growth market recently. "There is no real incentive right now to actually start to take out heads," McMillan said.
When the next recession comes, McMillan predicts, "you are going to see the largest percentage of layoffs that we've ever had."
And those jobs may never come back. "These organizations are going to do without those resources for two years, and they're not going to have to hire, because they're going to be forced to use AI to drive efficiencies," McMillan said.
The types of people banks need
The PwC research also found there's a 54% wage premium for roles calling for AI skills in the U.S. financial services sector. "Every firm in every industry needs access to AI capable users and AI capable technical people, and that trend probably isn't going away," Pollini said.
In McMillan's view, existing staff who understand a bank's account opening process and other arcane systems and workflows in the company are more useful than outside AI experts.
"For generative AI, what you really need is people who understand the processes really well, and then can apply those technologies," he said. "You don't need a technology background to apply Claude or ChatGPT. They need people who can design and build prompts. They need people who can engineer better data quality solutions, better controls."
Some banks also need data engineers and people who know how to run data quality checks and establish controls, policies, procedures and escalation mechanisms, McMillan said.
"If you want to build something in investment banking or in commercial banking, you need a senior executive who cares, and you're going to put a team of people on that project in the business to design requirements, prototype a solution using vibe coding, and then test the heck out of it," McMillan said. "There is nobody at Google or OpenAI who could do that."
