Every coffee chat now ends with the same question, asked slightly nervously: is AI going to take the analyst job? It is the wrong question. The analyst job has never been a fixed bundle of tasks; it is whatever juniors must do so that seniors can advise. AI is changing the bundle. The right question is what the seat becomes, and whether you still want it.
I watched this from an unusual angle: fifteen years in banking, the last stretch as an Executive Director in TMT at Nomura in Hong Kong, covering the companies that build this technology while the banks around us began deploying it on our own desks. So I have seen both what the tools can do and what the sales decks claim they can do. The gap is instructive.
What is already automated, honestly
By 2026, most major banks run internal AI assistants, walled off from the public tools for confidentiality, and juniors use them daily. The work that has genuinely moved: first drafts of company profiles and industry overviews, summarising a data room or a 300-page annual report, meeting notes and call summaries, first-cut public comps pulls, translation, and the endless reformatting that used to eat analyst nights. A task that took an evening now takes an hour of prompting and checking.
Notice the pattern: everything on that list is a first draft of something standard. Nothing on it is a final answer to something specific. That distinction is the whole story.
What does not change, and why
- The error tolerance on live numbers is zero. A model that is right 95% of the time is 100% unusable unchecked, because nobody knows which 5% is wrong until a client does. Banking is a business of being accountable for numbers. The machine drafts; a person signs. That person is you.
- Judgement is the product. Which buyer is actually credible, what the founder really wants, whether a board will accept a number: none of this lives in a document a model can read. It lives in conversations, and juniors earn their way into those conversations by being reliable on everything else first.
- Trust is the business model. Companies do not hire banks for information. They hire them to be accountable, discreet and present at 2am when the deal wobbles. That is not a workflow; it is a relationship, and it still gets built the slow way.
What actually changes for juniors
The honest version has two halves. The comfortable half: less time formatting, more time on analysis, verification and client-ready synthesis. The seat becomes more interesting earlier. The uncomfortable half: if a first draft costs an hour instead of an evening, a team may need fewer drafters. Banks are openly rethinking junior class sizes and structures, and while nobody should trust any specific prediction, the direction is clear: fewer purely mechanical seats, and a bar that shifts towards judgement, scepticism and communication earlier in the career. The floor rises. Mediocre plus diligent used to be a viable analyst profile. It is becoming less viable each year.
There is a second-order effect nobody warns you about: when drafting is cheap, volume expands. Ten buyer profiles become thirty because they can. The hours, as I wrote in the honest day-in-the-life, are driven by client deadlines and layered review, not by typing speed, so do not expect the tools to hand you your evenings back.
AI as a sector, not just a tool
The other half of this story is that AI is now the dominant theme in dealmaking itself. Data centre platforms, the power and infrastructure to feed them, semiconductor supply chains, software consolidation: this is where fees are concentrating, and it runs through TMT desks. A candidate who can hold a ten-minute conversation about how a data centre gets financed, or what an AI capex cycle does to a chip maker's revenue, is differentiated in a way no prompt-engineering claim will ever be. If that is your target seat, start with what TMT interviews actually ask.
How to position as a candidate
- Do not sell tool fluency as your value. Saying you are efficient with AI is like saying you are efficient with Excel: assumed, and slightly worrying if led with. Everyone in your class has the same tools.
- Sell verification discipline. The scarce junior skill is now catching the confident wrong answer: the hallucinated figure, the outdated comp, the source that does not exist. In interviews, talk about checking work, reconciling sources, knowing when a number smells wrong. That is the analyst the 2026 desk needs.
- Learn the sector story. Read the AI infrastructure deals in your market until you can explain one end to end. It signals commercial curiosity, which was always the point.
- Keep the fundamentals anyway. Accounting, valuation and the model mechanics are more necessary, not less, because verification requires knowing what the right answer looks like. You cannot catch a wrong number you could not have produced yourself.
FAQ
Will analyst classes shrink?
Some already have, and structures are being rethought at several large firms. But banking headcount has always swung with the deal cycle far more than with any technology. Treat class size as cyclical noise around a slow structural drift, not a cliff.
Should I mention AI tools in interviews?
Reactively, yes; proactively, carefully. If asked, describe using them with verification discipline. Do not volunteer them as your edge, and never suggest work product goes out unchecked.
Is TMT the best seat for the AI wave?
It is the most direct one, since the theme runs through technology coverage. But the capex wave touches power, infrastructure, real estate and credit too. A strong seat covering any of those is exposure to the same story.
Should I learn to code?
Useful, not required, and never a substitute for accounting and valuation. SQL or Python help in data-heavy groups and signal curiosity, but no IBD interview will test them, and no desk will forgive weak fundamentals because you can script.
If you want to pressure-test how your profile reads against a desk that is changing shape, that is exactly what the IBD Recruiting Review is built for: one hour, one to one, current to how banks are actually hiring.
Book the 60-min IBD ReviewView Full Cycle
Former Executive Director in TMT Investment Banking at Nomura and M&A banker at BNP Paribas. Top-rated Head Mentor on Wall Street Oasis with 300+ sessions. About
