Why AI will make investment bankers more efficient — not redundant — and what that means for how they get paid
Introduction: A Philosophy Built in the MIT AI Lab
Dave Blundin is a serial entrepreneur, MIT-trained artificial intelligence pioneer, and the Founder and General Partner of Link Ventures, a Cambridge-based venture capital firm. Having researched neural networks at the MIT AI Lab and built AI-driven companies since the early 1990s, his perspective on the technology carries the weight of someone who has watched it evolve from academic theory to enterprise infrastructure over three decades.
Blundin’s philosophy is deceptively simple: artificial intelligence should be built to augment human capabilities — making people more efficient, not redundant. It is a philosophy grounded not in optimism, but in evidence. He has watched AI move from academic curiosity to enterprise engine, and his consistent view is that the humans who learn to direct AI will be dramatically more powerful than those who either fear it or ignore it.
That philosophy is now playing out in real time on Wall Street — and its implications for investment banking careers, compensation structures, and the long-standing “pay your dues” culture of the industry are more profound than most people in the sector have yet reckoned with.
Part One: What AI Is Actually Doing to the Work
Investment banking has always run on a particular form of human capital: brilliant young graduates willing to trade their twenties for the promise of extraordinary pay, elite exit opportunities, and the credential of having survived one of the most demanding professional environments in the world. The junior analyst role — 80 to 100 hours a week, building financial models, assembling pitch books, scrubbing data, formatting slides — was the foundational layer of that system.
It is precisely that layer that AI is now disrupting.
From JPMorgan’s AI pitchbook tools to Deutsche Bank’s internal copilots, banks are already reducing repetitive tasks by 40 to 60%. JPMorgan’s co-head of investment banking has stated publicly that “AI will enable us to do tasks that take 10 hours in 10 seconds.” This is not hypothetical — one major bank has reported a 60% reduction in time to first draft for comparable company analyses and a 35% drop in associate-level errors after deploying automated modules. The practical picture is concrete: before, a comparable company screen required manual extraction of revenue and margin data from ten separate filings, reconciliation, and formatting — taking the better part of a full working day. After AI deployment, an ingestion pipeline populates the same table in under an hour, while the analyst focuses on identifying outliers, testing assumptions, and constructing the pitch narrative.
Goldman Sachs, arguably the most watched institution in global banking, has gone furthest in articulating what this transformation looks like architecturally. The firm has built its strategy around what it calls a “hybrid workforce” — deploying the GS AI Assistant across approximately 46,500 employees for routine document drafting, building a proprietary networking platform called Louisa, and becoming the first major financial institution to deploy Devin, an autonomous software engineering agent, across its 12,000-strong developer workforce. While earlier AI tools provided roughly a 20% boost in coding efficiency, Goldman reports that Devin has driven a 3–4x productivity gain by autonomously managing entire software development lifecycles. The firm is already experimenting with autonomous deal sourcing — AI models that identify potential M&A targets by analyzing global supply chain shifts, regulatory filings, and macroeconomic trends before a human banker has even picked up the phone.
The scale of the productivity opportunity is striking. Deloitte’s research shows that the top 14 global investment banks could boost front-office productivity by 27–35% using generative AI, potentially generating an additional $3.5 million in revenue per front-office employee. That is not revenue generated by replacing people — it is revenue unlocked by making existing people dramatically more capable.
Part Two: The Augmentation Case — Straight from the C-Suite
The most important endorsement of the augmentation thesis does not come from venture capitalists or academic researchers. It comes from the CEO of Goldman Sachs.
When asked directly about AI’s impact on his workforce, David Solomon pushed back sharply on the replacement narrative, calling it a “very simple media narrative.” His position: “There is no question that when you put these tools in the hands of smart people, it increases their productivity. You’re going to see changes in the way analysts, associates and investment bankers work. But if you’re looking at it and assuming an organization like Goldman Sachs is just going to have less people, I don’t think it works that way.”
Wells Fargo echoed this, saying AI enables more work without reducing headcount. JPMorgan’s head of consumer banking stated the bank had already doubled productivity in some areas. Bank of America announced plans to spend billions on AI technologies specifically described as tools to bolster banker productivity and drive revenue growth — not to shrink its workforce.
This executive consensus is not altruism. It reflects a structural reality of the investment banking business: the bottleneck was never computational power, it was judgment. The tasks that AI is best at — scraping data, populating templates, generating first drafts, running standard models — were never the tasks that generated fees. Fees come from trust, relationships, strategic advice, and deal execution. A Goldman Sachs MD who can close a $10 billion M&A mandate does not become less valuable because a junior analyst can now produce a 40-page pitchbook in 90 minutes instead of three days. If anything, they become more valuable — because the bank can now afford to pursue more mandates with the same headcount.
A first-year analyst can now supervise AI to produce work that once required three analysts. That is not replacement — it is radical productivity increase. The role is shifting from execution to orchestration.
Part Three: The One Thing AI Cannot Touch
There is a ceiling to AI’s penetration in investment banking that is not technological — it is human. It is trust.
As Marc Cooper, CEO of Solomon Partners, has written after decades of advising on major transactions: “I have a hard time seeing any technology disintermediating the essential relationship between investment banker and client. What clients buy above all is the ability to sleep soundly as their bankers manage the complexities of their financial affairs. It takes high-caliber insight from experienced professionals to generate the kind of personalized services our clients seek. That takes time, exceptional service — and trust.”
The data supports this. Only 6% of clients seeking investment advice would rely on an AI platform alone. The other 94% want a human — ideally a human who is well-equipped with AI tools, but a human nonetheless. When hundreds of millions of dollars, or a company’s entire future, are on the table, clients want someone whose reputation is on the line alongside theirs. They want accountability. AI, however capable, cannot provide that.
This dynamic is clear across industry analysis: senior bankers — relationship-led, judgment-heavy, reputation-driven — are the most protected category of banking professional. It is the purely transactional junior roles, the ones defined by execution rather than judgment, that are being hollowed out. The career ladder is not being eliminated; it is being restructured. Fewer people at the base, doing higher-value work earlier, rising faster to the level where human irreplaceability begins.
Part Four: The Bonus Problem — A Compensation Model Under Quiet Pressure
Before exploring what AI does to hours, it is worth understanding in precise terms what the bonus actually is in investment banking — and why it is structurally more vulnerable to AI disruption than almost anyone in the industry has yet admitted publicly.
The investment banking bonus is three things at once: a performance reward, a retention mechanism, and a compensation for suffering. Of these three functions, AI threatens the third most directly — and the industry has not yet reckoned fully with what that means.
The Numbers as They Stand
The bonus structure in 2026 is as follows. At the analyst level, bonuses run 65–100% of base salary — meaning a first-year analyst on a $115,000 base might take home an $80,000 to $130,000 year-end payment, for total compensation of roughly $190,000 to $245,000. Critically, analyst bonuses are paid entirely in cash — no deferral, no vesting schedule. As you move up the ladder, that changes significantly. Associate and VP bonuses carry 20–30% deferred into stock or restricted cash. MD bonuses can be 30–50% deferred, vesting over three to five years. The structure is designed to keep the people who matter most tethered to the firm.
This deferral architecture reveals something important: banks already know that junior banker retention is not the problem. First-year analysts are not leaving because their bonuses are unvested. They leave because the work is unsustainable. The absence of deferral at the analyst level is not generosity — it is an acknowledgment that the primary compensation for junior suffering is immediate cash, not long-term equity. You are being paid now for what you are enduring now.
The Divergence Already Happening
The most significant data point in the current compensation landscape is the growing gap between what AI productivity is generating for the bank and what is flowing down to the junior bankers generating it. In 2024 and 2025, M&A deal volumes rose 20–50% across regions. Investment banking fees rose 10–30% in each of those years. And yet total compensation for analysts and associates rose only approximately 5% — while managing directors saw gains of 25% or more, VPs and directors 10–15%.
The productivity surplus from AI is being captured almost entirely at the senior level. This is not irrational from the bank’s perspective: MDs are the ones originating the deals, and in a world where junior execution tasks are increasingly automated, the economic logic of concentrating rewards at the origination layer is sound. But it is a profound shift in the implicit bargain that has governed junior banking careers for decades.
Analysts are doing more — working on more deals, producing higher-quality outputs faster — while their share of the resulting fee revenue quietly shrinks. Analyst bonuses are averaging approximately 65% of base pay in 2025, up marginally from 62% in 2023, but substantially below the 80% highs during the 2021 hiring boom. Banks are managing variable costs at the junior level tightly, even as senior compensation expands.
The $200,000 Question
The most uncomfortable question circulating in talent management circles on Wall Street is this: if AI is handling 95% of the analytical work, why should banks continue paying junior bankers $200,000 or more in total compensation?
The traditional justification for junior banker pay rested on several interlocking pillars. First, scarcity: there were only so many hours in the day, and a bank needed multiple analysts working around the clock to meet deal demands. Second, suffering compensation: the 100-hour week extracted a real human cost, and the bonus was partly reimbursement for that cost. Third, talent competition: banks were competing for the same Ivy League graduates as consulting firms, tech companies, and law firms, and the pay premium was the price of winning that competition.
AI is quietly dismantling the first two pillars. One AI-equipped analyst can now do the work of three. The scarcity justification weakens. And if hours compress — which the productivity data suggests is a near-term likelihood — the suffering compensation justification weakens too. What remains is talent competition. And even there, the calculus is shifting, as the next generation of elite graduates increasingly weighs culture, meaning, and work-life balance alongside headline pay.
JPMorgan has already signalled the direction of travel. The bank has proposed cutting junior banker ratios from 6-to-1 to 4-to-1, with part of the shift to offshore roles. Goldman Sachs constrained headcount growth even during a blockbuster revenue year — its headcount grew by only 1,800 when deal volumes would historically have demanded far more. Wells Fargo’s CEO has explicitly stated the bank expects headcount to decline as AI efficiency gains compound. These are not coincidences. They are the early expressions of a structural repricing of junior labour.
One analysis from a leading banking careers platform put the scenario starkly: banks face a binary choice. Option A — keep paying junior analysts $200,000 because they are now doing the work of three people with AI assistance. Option B — pay them $120,000 because the AI is doing the heavy lifting and the human is supervising. Option B is more economically rational, and if Goldman moves first, every competitor faces pressure to follow. This could reshape compensation expectations for junior investment bankers not gradually, but within the next recruiting cycle or two.
The Bonus as “Suffering Premium” — and What Happens When Suffering Declines
There is a human dimension to this that the compensation data alone does not capture. The IB bonus has historically functioned partly as a psychic reward — validation for having endured something that most people could not. “You worked 100 hours last week. Here is $100,000 to confirm that was worth it.” This is not cynicism; it is how the culture has worked. The bonus was not just about economics. It was about identity, status, and the sense that the sacrifice meant something.
If AI compresses junior hours from 90 per week to 60 — a plausible near-term scenario given the productivity data — the psychic justification for the bonus premium shrinks alongside the economic one. You are no longer the exhausted soldier who crossed the desert. You are the reasonably-pressured professional who managed an AI system through a deal. Both roles may be equally valuable, but only one commands the cultural premium that has historically justified a $100,000+ year-end payment to a 23-year-old.
This cultural shift has no clear resolution yet. Banks are caught between competing incentives: they want to capture AI efficiency gains (which points toward lower junior compensation), but they also need to recruit talented graduates who currently have more options than ever (which points toward maintaining the premium). The likely outcome, in the medium term, is a compression and restructuring rather than a collapse — junior bonuses remain strong in absolute terms but grow more slowly than senior compensation, the hours differential between banking and consulting narrows, and the financial case for choosing IB over alternatives weakens incrementally.
The Deferred Reward Shift: Where the Money Actually Goes
The structural consequence of this dynamic is a gradual migration of investment banking toward a deferred-reward model more analogous to law, consulting, or private equity — industries where the outsized economics come not from the grind of junior years, but from reaching the level where your human judgment, relationships, and deal origination are genuinely irreplaceable.
In private equity, associates earn comparable junior pay to bankers but access dramatically higher long-term upside through carried interest — a structure explicitly designed to reward the judgment that compounds over a career, not the execution that can be automated. In law, the partnership track operates similarly: pay your dues as an associate, and the real financial rewards come when you have built a client book that belongs to you. In consulting, the partner track rewards those who develop the strategic insight and client relationships that no junior analyst — human or AI-assisted — can replicate.
Investment banking is moving, whether it acknowledges it or not, toward this model. The analyst of 2026 who embraces AI, develops genuine commercial judgment early, and positions themselves for the client-facing, origination-driven roles of the senior ladder will access extraordinary compensation — potentially more than their predecessors, because the efficiency gains from AI make the senior banker more economically powerful than ever. The analyst who treats the job as a two-year grind for a bonus and an exit to private equity, as previous generations have done, may find the maths less compelling as the “suffering premium” quietly deflates.
The message is not that banking will pay less. It is that banking will pay differently — and later.
Part Five: The Hours Question — and the “Not Yet” Caveat
Here is where the augmentation thesis requires an honest qualification. The productivity data is real. The efficiency gains are real. And yet, on the ground, junior banker hours have not yet come down.
Practitioners are noting that despite widespread AI tool deployment across pitchbook creation, financial modeling, and research, analyst hours remain stubbornly close to pre-AI levels. The reason is not that the tools do not work — it is that the efficiency gains are being captured as deal volume and headcount reduction rather than lifestyle improvement. Banks are using AI to pursue more mandates with fewer people, not to give existing people their evenings back. JPMorgan explicitly instructed managers to avoid hiring new people, using AI gains to absorb growth in workload rather than reduce hours. Goldman constrained headcount growth even during a record revenue year.
This is the first wave of AI in banking: banks benefit, bankers do not — yet.
The second wave, when it comes, will look different. As competition for talented graduates intensifies again, and as the generational shift toward valuing work-life balance continues, banks will face pressure to convert productivity gains into culture as well as margin. The firms that crack the human-AI collaboration model — genuinely better hours, genuinely more interesting work, genuinely higher per-deal upside — will win the talent competition of the 2030s. The ones that use AI purely as a cost lever will find themselves recruiting from a shrinking pool of candidates willing to accept the old bargain.
When you adjust for hours rather than total pay, the junior banking premium already looks fragile. A first-year analyst earning $200,000 at 80 hours a week achieves an effective hourly rate of approximately $48. A management consultant earning $120,000 at 55 hours a week earns roughly $42 per hour. The gap — the entire financial justification for the brutal junior banking experience — is $6 per hour. If AI compresses banking hours toward consulting levels while compensation structures remain unchanged, that gap disappears. And if banks respond by quietly compressing junior bonuses as well — which the current compensation data suggests is already beginning — the calculus of choosing banking over alternatives becomes harder to make.
Part Six: The Skills Gap — The Overlooked Risk
There is a genuine tension embedded in the augmentation story that any honest treatment of this thesis must confront.
The brutal junior analyst experience was never only about execution. It was about formation. Manually building financial models under pressure, stress-testing assumptions at 2am, iterating endlessly on client materials — these experiences developed something that cannot be easily replicated: commercial intuition, attention to detail, and the mental resilience required to succeed at the senior level. As one industry analysis bluntly states, removing too much foundational hands-on work too soon could create a dangerous skills gap.
The concern is real. If a junior analyst’s primary job becomes prompting AI and validating its outputs rather than building models from scratch, they may arrive at the VP level technically efficient but strategically underdeveloped. They will know how to supervise AI. They may not know why the model works, what its assumptions imply, or how to recognize when the output is subtly wrong.
This is perhaps the most important unresolved question in the AI-and-banking debate: who teaches judgment to a generation that learned the outputs without learning the process?
One answer is that banks will need to invest more deliberately in structured training — more like management consulting firms, which have long relied on formal programs rather than apprenticeship-through-suffering to develop junior talent. In that sense, the AI disruption may accelerate a cultural shift that was already overdue: from a model where learning happens implicitly through grueling execution, to one where it happens explicitly through mentorship, case-based training, and deliberate exposure to senior-level thinking.
The analogy to law is instructive. Junior lawyers spend years on tasks that are increasingly being automated — document review, legal research, drafting standard agreements — and firms are grappling with exactly this question: how do you develop legal judgment in someone who has never had to do the work that builds it? Investment banking is arriving at the same inflection point, and the answer will shape what the profession looks like for a generation.
Part Seven: The Deferred Reward Model — Where Banking Is Heading
The most consequential long-term implication of the AI augmentation thesis is a structural convergence of investment banking compensation with other elite professional service careers.
In law, consulting, and technology, the model is well-established: earn respectably in junior roles, develop genuine expertise, and access the outsized rewards later — through partnership tracks, equity, or carried interest — at the level where your human judgment and relationships are genuinely irreplaceable. The deferred reward is not a consolation prize; it is the design of the system. You pay dues, build expertise, and the economics reflect your value once that value is clear.
Investment banking has historically front-loaded the pain and the pay. The $200,000 first-year analyst package was the carrot that justified the 100-hour weeks and the cancelled weekends. That package made sense when the junior banker was genuinely the bottleneck — when the only way to get forty comparable companies analyzed was to have two analysts work through the night.
AI removes that bottleneck. The work gets done faster and with fewer people. The remaining junior bankers do higher-value work earlier. And the compensation structure, over time, is likely to reflect that shift: lower variable pay at the junior level (because the “suffering premium” is structurally smaller), higher emphasis on the skills that AI cannot replicate (judgment, relationships, deal intuition), and significantly larger rewards for the senior professionals who turn those AI-augmented capabilities into closed mandates.
This is not bad news for the next generation of bankers — provided they understand the new contract. A career in investment banking will increasingly look like a career in consulting or law at the junior level: demanding but more sustainable, oriented toward skill development rather than endurance, with the outsized financial rewards concentrated further up the ladder. The promise is still there. The timeline changes. The path changes. And the skills required to walk that path — technical fluency with AI tools, strategic judgment, client empathy — are ones that any genuinely excellent banker should want to develop anyway.
Conclusion: The Blundin Principle, Applied
Dave Blundin’s argument is not that AI will make jobs easier. It is that AI will make capable people more capable — and that the firms and individuals who understand this will generate significantly more value than those who don’t.
In investment banking, that means the analyst who can direct AI to produce a first-rate pitchbook in two hours is not a less skilled analyst than the one who spent three days doing it manually. They are a more valuable analyst — one who has two days available for the work that actually differentiates deals: understanding a client’s strategic position, seeing the angle that the model missed, building the relationship that wins the mandate.
The transition will not be seamless. Hours may not compress immediately. Compensation structures will lag the productivity reality. The skills gap is a genuine risk that banks will need to manage deliberately. And the implicit contract of junior banking — suffer now, earn later — will need to be renegotiated for a generation that has better options and less tolerance for the old model.
But the direction of travel is clear. The banker of the future is not replaced by AI. They are augmented by it. And the ones who thrive will be those who understand, as Blundin has understood since 1992, that the most powerful thing you can do with a machine is point it in the right direction.
Research sources include Link Ventures, Goldman Sachs, JPMorgan Chase, Deloitte, Fortune, Axios, Disruption Banking, Boston Institute of Analytics, Finalis, eFinancialCareers, and Prospect Rock Partners.