AI Is Already Reshaping M&A Recruiting — And Most Candidates Don't See It Yet

Class sizes are shrinking, the ideal candidate profile is shifting, and the summer internship class is the leading indicator. Here is what the evidence actually shows.

Artificial intelligence is quietly restructuring investment banking, and nowhere is that more visible — or more misunderstood — than in M&A recruiting. The changes are not dramatic headlines about mass layoffs. They are subtler: smaller class sizes, shifting interview criteria, and a slow repricing of what makes a candidate valuable. If you are recruiting into M&A right now, the evidence says this shift is already underway — and the candidates who understand it will have a meaningful edge.

The Work Is Changing, So the Hiring Is Changing

A significant portion of what junior bankers traditionally did — building comps, pulling precedent transactions, formatting CIMs, constructing first-pass financial models — is now being automated or heavily assisted by AI. Bloomberg Intelligence has projected that global banks will cut as many as 200,000 jobs over the next three to five years as AI takes over tasks currently performed by humans. A Citigroup report found that 54% of financial services jobs have high potential for automation — more than any other sector. Accenture found that 73% of working time spent by U.S. banking employees has high potential to be impacted by generative AI.

Perhaps the clearest signal of where this is headed: Bloomberg reported that OpenAI enlisted more than 100 former investment bankers from Goldman Sachs, JPMorgan, and Morgan Stanley to train its AI on how to build financial models — a project internally called "Mercury," specifically designed to automate the grunt work performed by junior banking staff.

The result at the hiring level is quieter than most people expect. Banks are not announcing layoffs — they are simply reducing the number of offers extended per class. Goldman Sachs posted record Q3 2025 revenues while simultaneously announcing another round of cuts, explicitly citing its new AI-powered operating model "OneGS 3.0" as the rationale. The headline at the time was blunt: revenues are booming; jobs are not.

Summer Classes Feel It First — And That's Already Happening

Banks make headcount decisions 12 to 18 months before someone actually shows up to work. So when senior leadership looks at AI productivity gains today and decides it needs fewer analysts, that decision flows directly backward into the summer analyst offer numbers being extended right now — for people who will not start for another year. The full-time hire feels it later. The intern class is the leading indicator.

The data from recent cycles reflects this. In a recent recruiting cycle, Goldman Sachs and Bank of America NYC both had fewer than 10% of superday candidates receive offers — well below historical norms, with practitioners directly attributing it to smaller class sizes. Citi London cut its IBD summer analyst cohort from 60 to 30 in a single year. Industry observers confirmed that some banks froze full-time analyst headcount entirely, while others trimmed internship cohorts by as much as 30%.

The associate pipeline compounds this further. Banks are also questioning whether the analyst-to-associate promote path needs to be as wide as it once was. If an AI-leveraged analyst can handle what two analysts previously managed, you need fewer associates to oversee them. The squeeze propagates up the chain with a delay, but it propagates.

A few practical implications for candidates:

• Return offer rates are a more meaningful signal than ever — earning one from a compressed class is genuinely harder than three years ago.
• The "safety school" bank strategy gets riskier when class sizes are shrinking simultaneously across the Street.
• Boutiques may behave differently — smaller shops could lean into lean teams plus AI leverage rather than cutting headcount, making each hire more substantive.

The MBA Pipeline Is Feeling It Too

Bloomberg's analysis of M7 employment data found that job placement outcomes declined at every single elite business school since 2021. At Harvard Business School, only 4% of MBA students received no job offer within three months of graduation in 2021 — a figure that grew to 15% by 2024. MIT saw nearly identical deterioration over the same period.

Only 5% of the HBS Class of 2024 went into investment banking, and 6% of the Class of 2025 — small numbers for what has historically been one of the school's flagship pipelines. For the first time in at least five years, tech overtook both consulting and private equity as the top hiring industry for HBS graduates, absorbing 22% of the class.

An HBS working paper studying job postings across nearly all U.S. vacancies from 2019 through early 2025 found that the largest reductions in AI-exposed postings were concentrated in finance and technology — the two sectors most relevant to candidates reading this.

What Banks Are Actually Looking for Now

The bar for what you need to know has shifted — and so has how banks test for it. The old model of technical interviews was largely about memorization and pattern recognition: define the three financial statements, walk me through an LBO, what happens to EPS if depreciation increases. That playbook is becoming obsolete, and not because the bar has dropped. It is because AI can now produce those answers instantly.

What has replaced it is harder to prepare for. Banks are increasingly using case study formats and open-ended scenario questions designed to test how you think under ambiguity — not whether you can recite a framework. You might be handed a CIM and asked to identify the three biggest risks to a deal thesis, or walked through a live negotiation scenario and asked how you'd advise a client on deal structure. The question isn't whether you know what a DCF is. It's whether you know when not to rely on one.

The broader profile banks are hiring for combines three things:

• Scenario-level technical fluency. Not definitions on demand, but the ability to reason through a complex, ambiguous situation — valuation trade-offs, deal structure optionality, how macro conditions change a model's assumptions — in real time.
• Practical AI fluency. Not as a resume line, but as something you can speak to concretely in the context of deal work — where you'd use it, where you wouldn't, and what it changes about how you allocate your time.
• Real domain expertise. A candidate who knows healthcare, energy, or technology deeply is increasingly more attractive than a pure finance generalist. Sector knowledge cannot be prompted.

Soft skills are also being weighted more heavily in final rounds. With AI handling more execution work, banks can afford to be more selective about communication, presence, and whether a candidate can credibly sit across from a CFO. If you walk into a superday and cannot articulate how you would use AI in your deal work, you look out of touch. But if you lean on AI fluency as a substitute for technical depth, you will get exposed — the scenarios are designed to find exactly that gap.

The Honest Counterargument

Not everyone agrees the sky is falling. Some experts argue that what looks like AI-driven job cuts is largely the hangover from pandemic-era overhiring combined with economic uncertainty — and that banks will need to rebuild headcount as deal volume recovers. There is also a credible case that AI will increase M&A deal volume over time, partially offsetting headcount compression at the junior level.

What to Do With This

If you are actively recruiting into M&A, the path is still navigable — but the margin for error is narrower than it was a few years ago.

• On technicals: Prepare for case studies and scenario questions, not just definitions. Practice thinking through messy, open-ended deal situations out loud — that is the format the room is moving toward.
• On AI: Build genuine fluency in a finance context. Know where these tools add value in deal work and where human judgment is still the product.
• On domain expertise: Develop real conviction around an industry before you walk into recruiting season. Sector knowledge is one of the few things AI cannot replicate on your behalf.
• On reading the market: Pay close attention to class size signals at your target banks. They tell you more than the official recruiting timelines do.
• On internships: Treat return offer rates as meaningful data. A compressed class with a high conversion rate is a better signal than a larger class that does not convert.

The candidates who will thrive are the ones who understand that AI has not made the analyst redundant — it has made the average analyst redundant. There is still room at the top. The question is whether you are positioning yourself to be there.