FT - Revenge against the banking nerds is near

The writer is a former global head of equity capital markets at Bank of America and is now a managing director at Seda Experts.

Soon after moving into investment banking in 1997, I was sent with other new recruits to a hotel in Sussex for a week-long financial modelling course. It rained nonstop, and our main diversions were the hotel bar and the Snake game on our Nokias.

Our instructor, an American business school adjunct, seemed incredulous that his career had been reduced to teaching such a gormless bunch.

I disliked the training. Although I grasped the ideas, my rudimentary Excel skills and clunky Gateway computer made the exercises painfully tedious. As a new “associate director” crossing over from law, I set myself a career goal to never build a model myself. I missed plenty of targets over the next 25 years, but that one I achieved.

Now, the grunt work I sought to avoid appears on the brink of mass automation. According to Bloomberg, OpenAI has hired more than 100 former bankers from firms such as Goldman Sachs and JPMorgan, paying them $150 an hour to help train its artificial intelligence to build the models that define junior banking.

What’s surreal is they’re earning more per hour teaching their replacements than they did doing the job. A full-time junior banker on $200,000 annual compensation makes roughly $50 an hour, assuming an 80-hour working week.

For the past couple of decades, investment banking has become a story of “revenge of the nerds”. Hiring has shifted away from the verbally-fluent generalist or scratch golfer towards the technically proficient workhorse. The ideal junior is someone who can calculate discounted cash flow in every conceivable way and handle last-minute pitchbook edits from a mercurially insecure managing director.

In my own London team, we found ourselves taking on more graduates from continental business schools than liberal arts types from Oxbridge. It wasn’t intentional; they simply scored higher on technical assessments. The job rewarded precision, resilience and database mastery. The emerging archetype became the multilingual PowerPoint wizard with several internships under their belt.

AI now threatens to upend that logic. Once machines can model thousands of scenarios in seconds, the competitive differentiator will be less about accuracy or stamina, but more about the qualities that have always prevailed in the higher ranks of investment banking — strong judgment, credibility and the ability to tell a story that makes numbers mean something. The showhorses may yet eclipse the workhorses.

This transition won’t happen overnight. Automating complex financial work will require years of refinement. The entry-level grind will erode gradually, and then perhaps suddenly, affecting not only banks but also the private equity groups and hedge funds that rely on them to train their future talent.

The profession’s structure has always rested on apprenticeships: endure years of toil and eventually graduate to advising chief executives. Remove the lower rungs of that ladder and the structure becomes rickety and unstable. You can’t prompt a chatbot to learn client management or intuition. AI can model every scenario, but it can’t read a room (at least, not yet).

It’s ironic that the traits most prized in recruits — attention to detail, 24/7 responsiveness and Stakhanovite work ethic — are precisely those easiest for AI to replicate. Machines don’t misplace decimals, call in sick, or fly to Italy for a friend’s weekend wedding. They have no friends (at least, not yet).

In short, the “nerds” who displaced the smooth-talkers are now, perhaps unwittingly, training the machines that may take away some of their roles.

This need not spell extinction for the junior banker, but it does demand reinvention. A new hybrid may emerge: the managing analyst, who oversees the AI output, tests its assumptions, and interprets its conclusions. Modelling skills will still be required to do this. But the emphasis would shift from execution to oversight and communication.

Even senior rainmakers aren’t immune. The traditional pyramid relies on legions of juniors (spoon) feeding analysis to a handful of managing directors. If a smaller, AI-augmented team produces the same output, why sustain the old hierarchy? Clients might also balk more at fees supporting bloated overheads. Smaller boutiques or new entrants may stand to benefit.

In future, machines will take on more of the analysis and number-crunching in investment banking advisory, leaving humans to focus more on persuasion. Or perhaps, one day, the machines will learn that too. Until then, it’s striking to watch a profession built on stamina and sacrifice now outsource endurance itself.