Companies that adopted AI grew their payrolls, and the growth landed almost entirely among senior staff. Senior headcount at those firms climbed 6.7 percent over five years while the junior share of the workforce slipped about two points12. Bharat Chandar of Stanford and Bouke Klein Teeselink of King's College London released the finding on Monday, drawn from 1.25 billion job postings and 154 million work records across 41 countries3. Read the headline alone and AI reads as a job creator. The researchers read the composition underneath it.
Their verdict is blunt. "AI is labor saving for junior workers and labor expanding for seniors in exposed occupations," Chandar and Klein Teeselink write3. Gains arrived among senior staff, and the thinning arrived at the entry level. Adopting firms ended up with roughly 3.3 percent more employees than their peers, so the total grew while its composition changed underneath2.
Chandar put the geography in one line: "The junior share declines in about three-quarters of countries, including significant declines in places as diverse as the US, Brazil, and Saudi Arabia"2. Breadth on that scale points at the tool itself. One country's hiring cycle would read as local.
Banking heard the cheerful half. Commercial and wholesale banking tops every corner of finance in optimism about AI and headcount, with 44 percent of industry respondents expecting job gains there, the Cambridge Centre for Alternative Finance found in its 2026 survey4. That optimism rests on a machine whose gift runs to senior work.
Here sits the problem. Banking runs on apprenticeship, and apprenticeship runs on volume: a first-year analyst learns credit by grinding through files until the patterns arrive unbidden. Software reads those files in seconds now.
Consider the work itself. A junior used to pull comparables, draft the credit memo, chase the missing statement, and then sit in the room while a senior took the analysis apart line by line. Machines pull, draft and chase. Sitting in the room is what remains, and seats in that room belong to people the bank already hired.
Debasish Patnaik runs QuantumBlack, the AI arm of McKinsey & Co., and he told Fortune in June exactly what the trade stands to lose: "Banking is an apprenticeship business. Today's junior analysts become tomorrow's managing directors. Senior judgment cannot be manufactured laterally"5. Weigh that last sentence. Manufactured is a production word, and Patnaik reaches for it to describe the one input a bank has to grow for itself.
Wall Street has already moved. Banks have trimmed junior analyst classes by as much as two-thirds, Fortune reported, while roughly 62 percent of their AI talent comes from those same junior ranks5. Firms are drawing down the pool they refill from.
Banks draw about 62 percent of their AI talent from junior ranks while cutting junior classes by up to two-thirds
Fortune's reporting on bank staffing, percent: AI talent drawn from junior ranks, and the deepest cut to junior analyst classes
Each bar is one share from Fortune's reporting: magenta for the AI talent banks draw from junior ranks, and blue for the deepest cut to a junior analyst class.
Source: Fortune [5]. Chart by Ryan Elliott Dennis.
The numbers behind this chart
| Item | Value | Note |
|---|---|---|
| AI talent drawn from junior ranks | 62% | Roughly 62 percent, per Fortune |
| Junior analyst class cut, at most | 67% | Cuts of as much as two-thirds |
Bill Winters, who runs Standard Chartered, described the substitution in his own words: "It's not cost cutting; it's replacing in some cases lower-value human capital with the financial capital and the investment capital we're putting in"5. Read the phrase he chose: lower-value human capital. It prices a person by the work they perform this quarter, and in a trade where worth accrues across a decade, a quarterly valuation is the wrong instrument.
David Solomon, chief executive of Goldman Sachs, made the opposing case at The Economic Club of Washington, D.C. on Oct. 31, 2025, when a moderator asked what artificial intelligence would do to the firm's headcount. "The firm is much more productive on a revenue-per-employee basis today than it's ever been, and it will be much more productive as we go forward, but that doesn't mean we will have less people. It means we have an opportunity to have more valuable people doing more valuable things to serve our clients and grow our franchise," Solomon said6. Value is the word he reached for too, the same word Winters used five months earlier to describe which employees a bank keeps and which employees it lets go, and the repetition says more than either man's own sentiment does.
Defenders of the trend hold a real case. A junior armed with a model does the work of three, so a desk needs fewer of them to clear the same queue, and that arithmetic holds beautifully for this year. It breaks in the decade when the bank needs a managing director who has read ten thousand credit files and can price a bad one on sight.
Judgment of that kind comes from volume. Volume is what got automated.
Here is the arrangement, stated flatly. Banking is the segment of finance most confident that AI will add jobs, and it is also the trade whose entire succession plan runs through the junior work AI performs first. Both facts come from the same industry, in the same year.
Ask the question the machine age keeps posing. Who owns the system, who feeds it, and who collects the gain?
Banks own the software. Senior bankers feed it judgment and collect a raise, since scarce experience prices well. Juniors supplied the material it learned from, in the files that used to teach them.
That 44 percent may prove right for a while4. Headcount can climb for years while the composition of a workforce inverts underneath, and a bank heavy with seniors and light on juniors reads strong on a payroll report today. Ten years out it reads thin.
The remedy belongs to the people signing the offers. Keep the analyst class and aim it at the machine's output: Bank of America booked 2,000 summer interns and 2,000 full-time recruits this year5. Pay for the years it takes. A bank that trains beginners through an era of cheap answers will hold the scarcest asset of 2031, which is people who can explain why an answer is wrong.
Watch the analyst classes this fall. They will tell you more about a bank's next decade than any AI budget does. The Signal tracks those hires every morning, one line each, with a source on every line. Subscribe by email below, and follow the next generation of bankers as they get hired or skipped.
Sources
- Bloomberg News. "AI Use at Companies Is Adding Jobs, But Mostly in Senior Roles." Bloomberg, Sept. 21, 2026. https://www.bloomberg.com/news/articles/2026-09-21/ai-use-at-companies-is-adding-jobs-but-mostly-in-senior-roles
- Chandar, Bharat. "AI's impacts on jobs around the world." Sept. 21, 2026. https://bharatchandar.substack.com/p/ais-impacts-on-jobs-around-the-world
- Insurance Journal Staff. "AI Adoption Is Driving Hiring, but Mostly for Senior Roles." Insurance Journal, Sept. 21, 2026. https://www.insurancejournal.com/news/international/2026/09/21/886207.htm
- Cambridge Judge Business School. "2026 Global AI in Financial Services Report: Adoption, Impact and Risks." Cambridge Centre for Alternative Finance, 2026. https://www.jbs.cam.ac.uk/faculty-research/centres/alternative-finance/publications/2026-global-ai-in-financial-services-report/
- Short, Meg. "Banks lay groundwork for mass workforce cuts as AI takes hold." Fortune, June 7, 2026. https://fortune.com/2026/06/07/banks-mass-workforce-cuts-ai-entry-level-jobs-junior-analysts/
- Mullen, Caitlin. "7 hot takes: Goldman's David Solomon talks AI, RTO and more." Banking Dive, Oct. 31, 2025. https://www.bankingdive.com/news/goldman-ceo-solomon-ai-banking-rto-fed-digital-assets/804363/
