Junior bankers in the City and on Wall Street have reason to worry. Is that the whistle of the wind they hear as they contemplate a potential fall through a trap door? The anxiety is palpable after CNBC reported that JPMorgan plans to overhaul its deal teams, cutting the ratio of junior bankers to senior managers from 6:1 to 4:1.
The main driver is artificial intelligence. When a large language model can churn out a credible pitch book, or at least a serviceable draft, in 30 seconds, it raises profound questions about the future of the industry’s newest recruits.
Investment banking has historically run on a straightforward, if pitiless, exchange. Top graduates chased high pay and prestige, and in return they endured long nights, heavy workloads, and relentless pressure.
The endless rounds of financial modeling, database runs, indicative pricings, term sheet production and pitch book preparation were not just drudgery; they were the crucible where future managing directors were forged and formed. Through exposure to live deals and client interactions, young bankers learned lessons no textbook could teach.
Now, artificial intelligence tears at the heart of this system. Goldman Sachs CEO David Solomon has said a large language model can draft 95% of an IPO prospectus in minutes (as someone who started as a prospectus-drafting securities lawyer in 1993, I find this alarming). When such a shift takes hold, the detailed analytical work that once built discipline and understanding largely disappears.
JP Morgan’s parallel plan to move many analyst roles to low-cost centres such as India only deepens this rupture. The old proximity between mentor and apprentice vanishes when an analyst working in Bangalore or Johannesburg reports to a director in London while relying on an algorithm for most of the grunt work and groundwork.
The immediate question becomes: how will judgement, subtlety and instinct be passed down from generation to generation when the osmotic process of learning-by-doing and learning-by-hanging-around has been replaced by prompting and reviewing?
This leads to a deeper strategic concern. Where will the next generation of senior bankers acquire their craft? Today’s managing directors learned through repetition, (a lot of) trial and (even more) error, and the slow accumulation of intuition that only comes from direct client exposure and market competition.
If tomorrow’s analysts spend their days reviewing machine output rather than building models or articulating a pitch, they may become adept at managing processes and yet lack the instinct or confidence to challenge assumptions or sense when something feels wrong.
Solomon’s observation that the human 5% of the job becomes more valuable in this scenario may be true, but that 5% rests on foundations built through years of mastering the soon-to-be-discarded 95%. It’s like perfecting the last mile of e-commerce fulfillment without the website or warehouses first.
Two immediate risks follow. The first is reliability. AI can process vast amounts of data but cannot read tone, gauge hesitation or spot the inconsistency that suggests a deeper story beneath the numbers. Banking decisions often hinge on precisely these nuances. Moreover, AI is prone to hallucinating and can get facts spectacularly wrong.
The second risk is client perception. A CEO handed an obviously AI-generated presentation might well wonder what value the bank is adding. Information is now ubiquitous; insight is what clients look for. If artificial intelligence democratises the former, banks must work far harder to justify the latter.
Both risks were illustrated in Australia, where Deloitte refunded part of its fee to the government after admitting a report it produced used AI that generated inaccurate footnotes and references.
Recent experiments also provide a cautionary tale. When Citi set up an analyst hub in Malaga, it was hailed as progressive way to assure work-life balance. The reality proved to be different. Most analysts requested transfers to London, and the project was eventually wound down. Ambitious young people wanted proximity to decision making, not isolation in back offices.
Removing the human apprenticeship may save money in the short term but risks eroding the motivation and ambition that drive the industry’s best and brightest.
Of course, there’s a more positive way to see AI’s role in investment banking. Instead of sidelining juniors, banks could use AI to handle repetitive tasks, freeing analysts to focus on higher value work like client strategy or deal structuring.
By pairing AI tools with mentorship, banks could help juniors develop critical thinking early, blending tech efficiency with human growth. That sounds promising in theory, but it raises more questions right now than it answers. Will AI move juniors up the value curve or just replace them?
The next financial crisis may not come from faulty algorithms or sloppy risk management, but from a generation of leaders who never had the chance to learn the skills their predecessors once took for granted
The commercial argument for automation is unimpeachable. Every bank faces pressure to improve efficiency and trim bloat. Yet this pursuit of short-term savings may carry a hidden long-term cost. Investment banking has always relied on deep institutional knowledge, built through repetition, mistakes and mentorship. If you replace those experiences with algorithmic shortcuts, then the institutional memory that underpins a lot of advisory and capital markets work will begin to wither.
AI thus risks hollowing out the profession from within. It may also hasten the commoditisation of a business that still likes to see itself as haute couture rather than prêt-à-porter. How would an investment bank distinguish what it can offer from its competitors if they’re all churning out the same AI-driven work?
The AI cat is now well out of the bag, but for the sake of sustainable success and client confidence, banks will need to balance the appeal of progress and technological adoption with the preservation of human learning and development.
Otherwise, the next financial crisis may not come from faulty algorithms or sloppy risk management, but from a generation of leaders who never had the chance to learn the skills their predecessors once took for granted.