The effect of AI on entry-level jobs

The debate about AI and jobs in financial services has become background noise — abstract, speculative, and easy to ignore. But the data coming out of 2025 tells a more specific story. Analyst-level roles aren't disappearing overnight. They're being quietly hollowed out, and the firms that don't notice are building a succession crisis they'll feel in five years.

## The numbers are no longer hypothetical

A Stanford study analysing payroll records from millions of U.S. workers found that employment for 22–25 year olds in AI-exposed roles — including accounting and financial analysis — declined 13% between late 2022 and mid-2025. For software developers in the same age bracket, the drop was nearly 20%.

Meanwhile, employment for workers aged 35–49 in the same fields grew by over 9%. The squeeze is generational and specific.

This isn't a forecast. It's already visible in hiring pipelines. Revelio Labs reports that U.S. entry-level job postings have fallen roughly 35% since January 2023. SignalFire found a 50% decline in new role starts for people with less than one year of postgraduate experience.

## What the large firms are actually doing

The public statements from bank CEOs tend toward reassurance. The internal decisions tell a different story.

JPMorgan is proposing to cut its ratio of junior investment bankers to managers from 6:1 to 4:1, with half of the remaining junior positions moved to lower-cost locations in India and Argentina. Their chief analytics officer demonstrated an internal LLM producing a five-page deal presentation — complete with earnings data, peer comparisons, and recent news — in thirty seconds. That task used to occupy a team of analysts working through the night.

Goldman Sachs cut roughly 1,000 roles in 2025. Their own research division published a report suggesting AI-driven reductions would continue into 2026. Yet publicly, CEO David Solomon maintains the firm won't simply have fewer people.

PwC's UK practice cut 200 graduate roles — a 15% reduction — with its head explicitly citing AI as a factor. Internally, PwC US is reportedly planning to reduce entry-level hiring by almost a third over the next three years. EY trimmed graduate intake by 11%. KPMG by 29%.

These aren't layoffs dressed up as transformation. They're structural decisions about how many junior people a firm needs when AI handles the tasks those people used to learn on.

## The real problem isn't headcount — it's development

Here's what most commentary on this topic misses: the traditional analyst role wasn't just cheap labour. It was a training programme disguised as a job.

Building models, pulling comps, formatting pitchbooks at 2am — none of that was valuable in isolation. It was valuable because it taught people how deals work, how to spot errors in data, and how to develop commercial instinct through repetition. The grunt work was the curriculum.

When AI compresses or eliminates that work, the development pathway doesn't automatically replace itself. You end up with a cohort of analysts who are expected to operate at a higher level from day one but who haven't had the reps that build judgement.

Two Harvard economists studied 62 million LinkedIn profiles and 200 million job postings and found that firms adopting generative AI saw sharp declines in junior hiring while senior hiring remained flat. The implication is clear: firms are keeping experienced people and cutting the pipeline that creates them.

This is a succession problem. Not immediately — the people who learnt the old way are still in place. But in five to seven years, when firms need the next generation of directors and partners, the bench will be thinner than expected.

## What this means for hiring now

For firms recruiting at the analyst level, three things follow:

First, the profile has changed. Technical ability is table stakes. What differentiates candidates now is whether they can operate with judgement and commercial awareness before they've had years of repetitive exposure. That's a harder thing to screen for, and most hiring processes haven't caught up.

Second, development must become deliberate. If the work no longer teaches by osmosis, then firms need structured approaches to building the instincts that used to come free. The ones that treat AI adoption as a pure efficiency play without rethinking development will pay for it later.

Third, the window for action is narrow. The firms adjusting now — rethinking what analyst roles contain, how they assess candidates, and how they build capability — will have a structural advantage in talent quality within three to five years. The rest will be competing for an increasingly scarce pool of mid-level professionals who got their training somewhere else.

## The bottom line

AI isn't eliminating analyst roles. It's eliminating the version of those roles that doubled as informal training programmes. That distinction matters enormously, and most firms haven't grappled with it yet.

The question isn't whether you'll still hire analysts. It's whether the analysts you hire in 2026 will be ready to lead in 2032 – and whether you're doing anything to make that likely.

‍

Sources:

‍

What AI Is Actually Doing to Analyst-Level Roles Right Now

## The numbers are no longer hypothetical

## What the large firms are actually doing

## The real problem isn't headcount — it's development

## What this means for hiring now

## The bottom line

Cross-Border Hiring Between London and Paris: What's Changed in 2026

Hiring Hotspots in Financial Services: What’s Really Driving Demand in 2026

What High-Performing M&A Teams Do Differently And How to Replicate Their Success