Here is the direct answer: no, AI will not replace investment bankers. The answer to “Will AI replace investment banking as an industry?” is also no. The junior analyst job is the most exposed role in banking, though, because so much of it is exactly what AI does well. Pitch-book formatting, comps pulls, model population from filings, drafting profiles, and diligence summaries are all compressing.
Obviously, AI cannot win a mandate, read a room in a negotiation, take responsibility for a valuation in front of a board, or carry the relationship that produces the deal. Banking has always paid for judgment and relationships, and it used junior hours as the training ground. The open question is how you build judgment when the grunt work shrinks.
This article explains which investment banking tasks AI is compressing and how analysts and aspiring bankers can build relevant experience and adapt their skills in 2026.
Note: AI affects investment banking differently from other finance functions. For role-specific coverage, see AI in FP&A, AI and financial advising, and AI and accounting.
How exposed each banking task is right now
Job-level questions get vague answers, so start at the task level. The table below scores the core banking tasks by how much of each AI can already produce and how much still requires a person who signs their name to the output.
The ratings in the last column are an editorial judgment, not a published index. High means AI can produce most of the deliverable today and a person edits it; medium means AI produces useful inputs while the judgment inside the task stays human; low means AI only supports preparation.
The calls come from what bankers describe using AI for in the practitioner discussions, and from where firms still require a named person to sign off.
We rate the task, not the person doing it: a high-exposure task performed by someone who verifies the output carries different career risk than the same task performed by someone who only produces it.
| Banking task | What AI does well today | What still needs the human | Task automation exposure |
|---|---|---|---|
| Pitch-book production | First drafts, formatting, standard market pages | Storyline, positioning, what the client actually needs to hear | High |
| Comparable company analysis | Pulling comps, screening, first-pass spreads | Choosing the right peer set, adjusting for one-off items | High |
| Model population | Filling models from filings and transcripts | Assumptions, scenario logic, catching errors that look plausible | High |
| Diligence review | Summarizing data-room documents, flagging keywords | Judging what a flag means for price and terms | Medium |
| Client origination | Meeting prep, account research | The relationship itself; winning the mandate | Low |
| Negotiation | Background briefs, precedent terms | Reading the counterparty, making trade-offs in the room | Low |
| Valuation sign-off | Supporting analysis | Accountability for the number a board relies on | Low |
| Deal execution | Checklists, status tracking, document summaries | Coordinating parties under pressure, decisions when things slip | Medium |
AI already produces the first draft of most banking deliverables
In January 2025, Goldman Sachs CEO David Solomon said AI can draft 95% of an S-1 IPO prospectus in minutes, work that previously took a six-banker team about two weeks. He added that “the last 5% now matters because the rest is now a commodity.”
Major banks now provide internal AI assistants to large parts of their workforce. Goldman rolled out its GS AI Assistant firm-wide in June 2025, after more than 10,000 employees, about a quarter of the firm, used it in a year-long trial for document summarization, drafting, and coding. JPMorgan rolled out its LLM Suite in summer 2024, and within eight months it was in the hands of 200,000 employees, mostly for generating ideas and drafting content.
In a May 2026 Wall Street Oasis discussion, participants described AI helping with research and analysis: slide drafts, data pulls, first-pass comps. One commenter said the time saved “just turns into more work or higher expectations.”
Other participants described AI as a first-pass quality-control tool. One commenter tagged as working in investment banking said Claude helped check confidential information memoranda (CIMs) and models for specific inconsistencies and mistakes. Another participant used Claude to flag font, color, logo, and data inconsistencies in slides. These are individual experiences rather than evidence of standard practice across banks, but they show how AI can review deliverables as well as draft them.
So the capability claims are real, but they cover a specific slice of the job.
Bankers remain responsible for judgment, verification, and client context
AI can produce more of the first draft, but a banker still decides which details matter, whether the assumptions hold, and how the analysis should shape client advice.
In that same May 2026 Wall Street Oasis discussion, professionals described how their own value shifted once AI entered the workflow. One said time saved on data pulls and initial research often went into correcting errors in AI-edited models; another described checking and verifying AI output as a core skill for analysts. Other contributors addressed the judgment side of the job. They pointed to deciding what matters for the client, working under pressure, and challenging implausible model assumptions.
What public data can and cannot show about investment banking job security
Available employment data is too broad to measure investment banking job security by seniority. The following evidence supports a narrow conclusion: AI is changing work at analyst and associate levels and may reduce how many additional people banks hire as the business grows. It does not yet show measurable industry-wide job losses for junior or mid-level investment bankers.
The U.S. Bureau of Labor Statistics counted 453,500 seasonally adjusted jobs in investment banking and securities intermediation in June 2026, barely moved from 454,800 in July 2025. But the category covers everyone these firms employ, bankers along with operations and technology staff, so a flat total says nothing either way about cuts to junior and mid-level roles.
Company statements and industry reporting fill in a little. In October 2025, Goldman Sachs CEO David Solomon told Axios he expected the firm’s headcount to keep growing, even as AI reshaped the work of analysts, associates, and investment bankers. Reuters Breakingviews noted in June 2026 that banks were still hiring new analyst classes and argued that removing juniors would deliver limited margin gains while weakening the future talent pipeline. Neither source provides headcount data by level.
Johnson Associates’ Q2 2026 compensation outlook offers one signal that AI is affecting banking workforce structures. Under its combined investment and commercial banking category, the consultancy cited headcount trimming linked to AI efficiencies and said firms were reconsidering career progression as workforce structures changed. The report also projected higher incentives in advisory and underwriting, which suggests that stronger business activity can coexist with leaner staffing. However, it provides no job counts by level and does not isolate investment banking from commercial banking, so it cannot show whether AI is reducing analyst or associate roles.
The junior analyst problem: where judgment comes from now
Banking used grunt work as its training system: you learned what a good comp set looks like by building fifty bad ones under supervision. If AI absorbs that work, how will juniors enter the profession and gain the experience that entry-level production work once provided?
José Parra Moyano, professor of digital strategy at IMD, calls this the judgment gap. He defines judgment as “knowing when you need to question an answer a machine gives you” and argues that it develops only through years of practice that entry-level work used to provide.
His proposed fix reshapes the junior role: less time on draft production, and more time to question AI output, spot what is missing, and hear seniors explain decisions earlier than before.
Applied to investment banking, this could mean using AI for first-pass comps, market pages, and model population while junior analysts check the numbers, flag questionable assumptions, and explain how the analysis should be tailored to the client. Senior bankers would bring them into valuation reviews and client-preparation meetings earlier so they can see which assumptions are challenged and why.
Build the AI skills investment bankers now need
No current study ranks a definitive set of AI skills for junior or more senior investment bankers. Three 2026 sources point to the same capability cluster, though.
- Use AI inside a complete banking workflow. Learn which tools and data your firm approves, then connect the stages of a task: retrieve sources, run the analysis, create the model or presentation, and prepare it for review. First step: repeat one low-risk task, such as public-company meeting preparation, with an approved assistant. Save the sources, prompts, draft, and corrections so you can see where the workflow breaks.
- Direct multi-step AI work. Hand the tool your required deliverables, source boundaries, assumptions, date cutoff, and formatting rules up front. Then ask it to name what’s missing and propose a work plan before it produces anything final. First step: take one completed analyst task and write the instructions an AI agent would need to reproduce its stages. Compare that plan with the process your team used.
- Verify and audit AI output. Check the method, formulas, key figures, sources, assumptions, and caveats, not just the final number. First step: next time AI supports a model or comps spread, trace three figures back to the filing and record every correction. That review log shows which errors recur and which checks you should run first.
- Reconcile the deliverables and extract the implication. A model, presentation, and memo should use the same names, numbers, assumptions, and conclusions. Once they match, explain what the analysis means for the deal or client and what remains uncertain. First step: compare one completed model with its presentation, note every inconsistency, and draft a three-sentence takeaway for a senior banker to challenge.
Use senior review as the learning loop. After a VP or associate comments on your work, record whether each correction concerned the data, method, assumption, consistency, risk, or client implication. Apply that checklist to the next AI-assisted task.
If you’re targeting banking as a career now
Should a student still aim for banking? Yes, if you wanted the job for its actual core: deals, clients, and responsibility under pressure. AI is unlikely to replace investment banking as a career path, because that core is the part AI strengthens rather than threatens, and seniors spend more of their time on it as production compresses.
Plan for a different first year than the one older analysts describe. Expect fewer hours of pure formatting and more expectation that you can question AI output from day one.
Recruiting differentiation shifts toward demonstrated AI fluency plus the fundamentals that let you check a machine’s work: accounting, valuation logic, and clear writing.
AI fluency can also blur job-search terminology. The phrase “AI analyst jobs in banking” can refer to technical and governance roles that build or oversee AI systems, as well as investment banking analyst roles that use approved AI tools in deal work. If you want the transaction path described here, check whether the role centers on valuation, client materials, and deal execution or on data infrastructure and model development.
For deal-focused roles, the practitioner pipeline argument still matters: banks need juniors who can grow into VPs, which gives firms a structural reason to maintain an entry-level talent pipeline.
No one can reliably forecast future analyst-class sizes. In practitioner discussions, bankers disagree about whether AI is reducing hiring or whether smaller classes reflect deal-market uncertainty and cost discipline. If that uncertainty makes you want to compare structurally insulated alternatives before you commit, the AI-proof careers guide explains what makes work resistant to automation.
If you decide to look beyond banking, AI fluency can also support careers beyond banking when paired with finance expertise. The highest-paying AI jobs guide explains how domain knowledge, communication, regulatory literacy, and model-evaluation skills apply to AI governance, risk, product, and other regulated-industry roles.
Either way, control what you can: the skill profile above, applied before your first interview.
FAQ
Will AI replace investment bankers?
Are analyst jobs in investment banking disappearing?
What parts of investment banking can AI actually do?
Is investment banking a good career with AI in 2026?
Will AI replace financial analysts too?
Can bankers use ChatGPT with deal data?
What AI skills matter most in banking?
Learn fundamental AI skills, even on a banking schedule
Start with the fundamentals: give an AI tool clear instructions, assess whether its answer is useful, refine weak output, and recognize when the work needs human verification. You need this base before you apply AI to research, models, or client materials. Banking hours make long evening courses unrealistic, so the practice needs to be brief and regular.
Coursiv is built around that constraint: bite-sized lessons that average about six minutes, daily challenges that turn AI practice into a habit, and built-in AI tools that let you apply prompts and workflows immediately, with a certificate of completion when you finish a course. As the next step after the skill plan above, it gives you structured, guided practice that fits between a morning call and a data-room session.
If you want to explore other options, the AI for finance course comparison explains how to choose a course and compares three programs by depth, cost, and time commitment.
The first action is small: choose one fundamental skill, such as writing clearer instructions or checking sources, and set a daily practice slot you can defend. Review in three months whether you can use AI more confidently and catch weak output before it reaches a senior banker.