Fields / City
Finance & Investment Banking
First stop
What this field is
Finance is the field that decides what money does — where capital goes, what it costs, how risk is priced, and which ventures, companies, and governments get funded. It is not a single job but a family of roles united by a common structural pull: the act of distinguishing what is worth investing in from what is not, what is fairly priced from what is mispriced, and what risk is acceptable from what will destroy value.
That is why the primary gradient is Judgement. The field exists because capital is finite, information is imperfect, and somebody has to make the call — to lend or not, to buy or sell, to advise a merger or walk away, to price a bond or structure a deal. The quality of that judgement is what separates a good financier from a bad one, and the consequences of getting it wrong are measured in billions.
Resolution is the daily texture for much of the advisory and structuring work. Investment bankers, in particular, spend their working lives taking transactions from messy to done — an acquisition that needs structuring, a company that needs capital, a balance sheet that needs restructuring — and the process is a sustained, document-heavy, multi-party push toward a signed deal. The work is incremental, adversarial in a procedural sense (counterparties, regulators, lawyers), and dense with spreadsheets, models, and legal documents. Organization runs alongside Resolution because finance is fundamentally about ordering complexity: building a financial model is an act of turning a chaotic business reality into a structured, comparable, decidable framework.
Connection is the relational substrate. Deals happen because people trust each other — bankers trust analysts, clients trust advisors, investors trust fund managers, and the network of relationships is the infrastructure on which transactions run. The relational dimension of finance is consistently underplayed in how the field presents itself; technical skill gets the interview, but the ability to build and maintain trust across a deal table is what makes a career.
What nobody tells you is how strongly the lifestyle divides the field from inside. Investment banking at the junior level is among the most demanding working environments in any profession — eighty-to-hundred-hour weeks during live deals are not exceptional but structural, and the pace is set by the deal timeline and the client, not by the analyst's preferences. Graduate analysts at bulge-bracket firms in London currently earn total compensation of roughly £85,000–£140,000 (base salary around £60,000–£70,000 plus bonus), rising to approximately £155,000–£255,000 at the associate level, with senior bankers earning substantially more [from salary and job-posting data, Glassdoor/Canary Wharfian/PrepLounge 2025-26]. The compensation is not separate from the lifestyle — it is what the lifestyle costs, and many junior bankers leave for private equity, hedge funds, corporate finance, or industry roles after two to four years, taking the technical training with them. The exit options are part of the value proposition, and the industry knows it.
The field is less credentialled than medicine or law — there is no single mandatory professional gate — but the practical barriers to entry are steep. Most front-office roles at major banks require a strong undergraduate degree (often in a numerate discipline), a summer internship that converts to a graduate offer, and increasingly a master's in finance or an MBA for lateral entry. The CFA (Chartered Financial Analyst) qualification is the closest thing to a professional standard for investment management and is recognised by the FCA as an appropriate qualification for several regulated activities [CFA Institute / FCA 2025]. In the UK, the Financial Conduct Authority (FCA) regulates firms and individuals performing regulated activities, and certain roles require FCA approval as "approved persons" or are subject to the Senior Managers and Certification Regime [FCA 2025]. In Portugal, the CMVM (Comissão do Mercado de Valores Mobiliários) regulates securities markets, and the Banco de Portugal supervises banking, with relevant EU frameworks (MiFID II, DORA, MiCA) applied through national transposition [CMVM/Banco de Portugal 2025].
The pull
Judgement
Undifferentiated → Distinguished
The pull to discern what's good from what isn't
Resolution is the daily texture for much of the advisory and structuring work.
The line forks
7 ways in
Service update
What AI just changed
What is newly possible in Finance & Investment Banking, what stays yours, and how AI is changing the way in. In that order, and only if you want it.
Barriers that dropped
2 barriers have dropped
The analyst-army barrier to running a deal processInference
Executing an M&A or capital-markets process required a pyramid of junior labour to build comps, populate data rooms, draft memoranda, maintain models and chase process. Only large institutions could field that bench, which concentrated advisory in firms with capacity and confined independent advisors to origination.
Generative and agentic tooling assembles standard model scaffolds, drafts market slides and comps tables, and reviews thousands of pages of filings in hours. Reported claim of a first-year analyst supervising AI to produce work that once required three analysts (reported by people in the field, unattributed trade press). If the bench can be small, the minimum viable advisory firm gets smaller.
The full deal-execution stack run at boutique or independent scale by a team that could not previously have staffed it.
Genuinely independent advisory practice, where the senior relationship-holder no longer needs an institution to supply capacity.
- independent and boutique advisory as a structural category — named verified exemplars of AI-enabled boutiques winning mandates they could not previously staff were NOT located, which is the honest weak point of this collapse
Very likely to keep going this way. Whether it changes which firms win the work is uncertain: if client relationships matter most, it may change the daily tasks without changing who gets hired.
Small advisory firms were already growing before AI, for other reasons, so that growth should not be credited to AI. Winning clients has not got any easier, and that is the real barrier to working independently. A tool that builds the comparison tables does not bring in the deal.
The quant PhD gate is moving, not fallingInference
Quantitative finance required a doctorate. Mode 1: 'Most quant roles at top firms require a PhD or master's in mathematics, physics, statistics, computer science.' The gate was mathematical research ability, certified by a PhD.
Accessible ML libraries, LLM-assisted coding and open-source backtesting lower the cost of building and testing systematic strategies. Simultaneously, at the frontier the reported competitive advantage is shifting from research to 'how cheaply and how quickly a desk can iterate — an engineering and data problem before it is a research one' (reported by people in the field, Quantt 2026).
Systematic strategy development as an engineering discipline rather than a research one, without a doctorate.
Signal research at independent or small-fund scale.
- open-source quantitative stacks and independent systematic trading as a category
- counter-evidence recorded rather than hidden: Citadel and Jane Street compensation remains structured to attract exactly the research profile this collapse supposedly displaces (reported by people in the field)
Could go either way. How far it goes is also less certain. It could reverse if leading firms go back to valuing new research over fast engineering, which would make a doctorate matter more again.
At top firms the bar has not dropped and may be rising: quantitative analysts (quants) now need machine learning skills on top of the maths, and the competition is global. It is easier to get in at the edges, not at the centre, and it does not mean a doctorate is no longer needed for a top quant job.
Jobs that did not exist five years ago
These are real jobs that exist now and did not exist before the current wave of AI. Tap one to see what it is and how well we know it.
Executive-level ownership of AI strategy, deployment and risk across the institution. Listed as an institutional-commitment signal rather than a career target — this is not an entry point.
HSBC appointed David Rice as first chief AI officer effective 1 April 2026; Commonwealth Bank of Australia appointed Ranil Boteju in the same window (named, verifiable appointments via trade reporting 2026). A reported rise in organisations with a dedicated CAIO from 26% (2025) to 76% (2026) is reported by people in the field with an unstated sampling frame and must NOT be read as a population statistic. · current fact
Owns the ethical, legal and regulatory boundaries around automated decision-making in financial systems; ensures machine-driven decisions survive supervisory scrutiny. Typically reports to the Chief Risk Officer or CIO. The most decision-relevant new role in the field for a young person: it is an actual career track and its demand driver is statutory rather than fashionable. Timing caveat: the Digital Omnibus deferral of Annex III obligations to 2 December 2027 means some 2026 hiring urgency was priced against a deadline that no longer exists — expect a softer 2026-27 than recruiter content implies and a sharper run-up through 2027. The role is real; the timeline moved.
Reported as a fast-growing hire at banking institutions and wealth platforms with a reported US band of roughly $125,000-$200,000 — job-board aggregation via GRC Careers / Murray Resources 2026. Recruiter-marketing provenance; the salary band is advertising, not data. Statutory driver: EU AI Act Annex III high-risk human-oversight obligations, now applying from 2 December 2027. · seen in the wild
Validation and governance of machine-learning models under model risk management regimes, extended to cover GenAI-era methods. The least hype-contaminated growth area in the field: the regulation is real and creates mandatory human-oversight headcount. Timing caveat: the EU statutory driver now bites from 2 December 2027 rather than 2 August 2026, so near-term demand rests on US model risk management practice and on firms preparing early rather than on an imminent EU deadline. Direction unchanged; urgency ~16 months softer than it looked six months ago.
SR 11-7 model risk management guidance (US Fed/OCC) long predates GenAI; EU AI Act Annex III 5(b) human-oversight requirements extend the mandate from 2 December 2027 (deferred from 2 August 2026 by the AI Digital Omnibus). · current fact
Same title, materially different work: validating machine-produced comps, models and drafts rather than producing them. Currently more an industry aspiration than an observed settled configuration.
Banks reported to source roughly 62% of AI talent from junior cohorts; reported claim that a first-year analyst can supervise AI to produce work that once required three analysts. Both figures are unattributed trade press and should be treated with real scepticism. · seen in the wild
This section needs its caveat more than any other in the file. Nearly all 'new AI role' evidence in finance originates from recruiter marketing, job-board SEO content and vendor commentary — sources commercially motivated to describe emerging roles as more numerous, better paid and more established than they are. Nothing here is graded above reported by people in the field except the two named CAIO appointments, which are verifiable events. A student should weight the AI-governance and model-risk track heavily and the rest lightly, because those are the only ones whose demand driver is a dated statutory obligation rather than a press release.
What stays yours · where AI reaches
Bars above the line are the parts of this work that still need a person. Bars below it are what AI can already do. Tap any column to see the actual work behind it.
The honest read. A strong triad — relational judgement, accountability, ambiguity-navigation — and every element of it is a property of SENIOR work. The profile describes a managing director almost perfectly and a 22-year-old graduate not at all. Juniors build the model that the accountable person uses; they do not negotiate, do not carry SM&CR accountability, do not decide whether the thesis is wrong. The quality flag passes comfortably, and passing it conceals the actual risk: the field is protected, the entrance is not. Read this section against the seniority stratification in §2, never on its own.
Two core domains that are not in tension — they are the two halves of the same job. Finance is a documents field and a numbers field simultaneously, and AI is now materially capable at both. The structurally important fact is what is NOT here: no robotics, no scientific AI, no physical substrate of any kind. Finance is the purest knowledge-work field in this corpus, which means it has the least structural protection from capability improvement. Nothing about the work requires a body to be anywhere.
How AI is changing the way in
5 ways into this field, and AI is not doing the same thing to each of them.
- Investment banking analyst / associate (front-office deal execution)Much harder to enter
Reported junior intake reductions up to two-thirds at some banks (reported by people in the field, unattributed trade press); Randstad reports global financial services 0-2 year hiring down 24% from early-2024 levels (reported by people in the field — a recruiter's placement book, not the market); Oliver Wyman CEO survey >40% plan to cut junior roles over 1-2 years vs 17% expanding (reported by people in the field). No single measure is strong. The convergence of three independently-sourced measures in the same direction at similar magnitude supports 'severe' as a direction with rough magnitude, NOT as a precise figure.
- Quantitative analyst / financial engineerHarder to enter
The gate is reported to be shifting from mathematical research toward engineering and data infrastructure (reported by people in the field). ML fluency is additive to the mathematics rather than a substitute for it — the bar moved sideways and arguably up, not down.
- Financial / risk analystSlightly harder to enter
EU AI Act Annex III 5(b) high-risk obligations mandate human oversight, documentation and post-market monitoring on creditworthiness systems — now from 2 December 2027 following the AI Digital Omnibus deferral from 2 August 2026 (happening now). Regulatory expansion remains a hiring tailwind and this is the one segment where AI deployment increases demand for the role; the deferral delays the statutory driver ~16 months without reversing it.
- Corporate treasurer / treasury analystSlightly harder to enter
Small teams, heavy external counterparty relationships, accountability for company solvency. AI's principal appearance in this segment is deepfaked executive-impersonation payment fraud — a threat to defend against rather than a substitution mechanism (inference).
- Fund / portfolio managerLargely unchanged
Minimal because there is no entry level. Mode 1: 'Direct entry from university into portfolio management is rare' — the industry wants analytical competence demonstrated before handing over capital-allocation authority. The AI risk to this archetype is entirely indirect, arriving via degradation of the analyst roles that feed it.
That is everything we currently know about AI in Finance & Investment Banking. It shows where things are moving so you can choose which way in suits you.
The job market
Getting work in this field
- People and jobs
- Changing who it hires
- The main limit on getting in
- AI doing entry-level tasks
- Entry-level jobs
- Shrinking
- Cost of training
- Worth weighing up
People and jobs. Not surplus and emphatically not shortage. Finance remains the leading career choice for university students and recent graduates globally — CFA Institute 2026 Graduate Outlook Survey, fielded 18 March-6 April 2026, 9,000 respondents aged 18-25 across 11 major markets (happening now; a well-specified survey with a stated frame, and the strongest single piece of evidence in this file). Demand for finance skills has not collapsed. Composition is what is shifting: only 3% of investment professionals report proficiency in AI/ML concepts (reported by people in the field, CFA Institute) while institutions are reported to source the majority of AI talent from junior cohorts. Fewer traditional analysts, growing demand for an AI-fluent configuration.
The main limit on getting in. AI displacement is newly binding, but it compounds a pre-existing credential gate rather than replacing it. Mode 1 is explicit that the target-university → summer-internship → graduate-offer pipeline is the real gate and that 'direct graduate applications without prior internship experience are significantly harder.' That gate has not gone away; the number of seats behind it is shrinking. A student facing both constraints simultaneously faces a materially harder entry than the same student five years ago, and the two interact multiplicatively rather than additively — fewer internships means fewer conversion offers means fewer analysts means fewer future MDs.
Entry-level jobs. Randstad: global financial services hiring for 0-2 year junior positions down 24% from early-2024 levels (reported by people in the field). Oliver Wyman: >40% of CEOs plan to cut junior roles over 1-2 years vs 17% expanding (reported by people in the field). Reported junior intake cuts of up to two-thirds at some banks (reported by people in the field). Directionally convergent across three independent sources; magnitudes individually unreliable.
Cost of training. Moderate rather than high because finance compensation is genuinely high — Mode 1 reports London bulge-bracket graduate analyst total compensation of roughly £85,000-£140,000, which services debt comfortably. The reason it is not 'low' is that the master's-in-finance and MBA lateral-entry routes are expensive and are precisely the routes being compressed. Someone taking on significant debt to buy entry into a tier that is shrinking carries a real and under-discussed risk, and the exit-options value proposition Mode 1 describes ('a training programme with a salary') depends on the training tier continuing to exist.
Finance is the field where AI's entry-tier effect is least cushioned, and the absence of a shortage is the reason. Vet science, cybersecurity, K-12, medicine, nursing and midwifery all have structural shortages that cushion their AI stories — automation lands where more hands were needed anyway and relieves pressure rather than removing jobs. Finance has no such cushion: the CFA Institute's 2026 survey confirms it is THE most-wanted destination for graduates globally, and Mode 1 describes an oversubscribed pipeline. Automation lands in a field with an excess of willing entrants and no unmet demand for their labour. Combined with an economic model that makes juniors cost rather than revenue, and a regulatory floor built around consumer decisions rather than analytical production, this is why the same deployment level produces relief in nursing and compression in banking. That contrast is the single most useful thing a student can take from this assessment.
Interchange
Change here for
People drawn to Finance & Investment Banking are often drawn to these.
End of the line
That is a conversation about this field, not a test. Or keep going on your own. Both doors stay open.