The Death of Traditional Investment Banks
In five years, every traditional boutique investment bank will be dead. Or at least unrecognisable from what it is today.
In five years, every traditional boutique investment bank will be dead. Or at least unrecognisable from what it is today.
But before I dive deeper, let me be clear about what I mean by traditional banks.
When people hear investment banking, they think Goldman Sachs, JP Morgan, Morgan Stanley. But an estimated 95% of M&A advisory firms operate on deals well below $200M. And while investment banking covers many things, the overwhelming majority of activity at these firms is sell-side M&A advisory. Helping business owners sell their companies. This article is about that work. Not the billion-dollar relationship game at the top. The bulge brackets are safe, at least for now. That is a topic for another post. This post is about the other 95%.
---
The Structural Problem
At Eilla AI, we spent three years building AI automations for investment banks and private equity funds. We met thousands of them. The sales cycles were brutal. In most banks, adopting technology is painful. Especially in the larger ones. You need multiple people to approve even the simplest AI tool. And no, that is not safety or compliance. That is bureaucracy.
Traditional boutique investment banks have a business model that does not work for most of the market they claim to serve.
The math is simple. A proper sell-side M&A process requires hundreds of hours of work: building buyer universes, preparing teasers and CIMs, creating financial models, running outreach, managing NDAs, coordinating due diligence. For a small deal, the advisory fee might be $300K to $500K. That barely covers the cost of the team needed to run the process properly. So either the deal does not get done, or it gets done badly.
Corners get cut. Buyer outreach is limited to a handful of contacts the lead banker already knows. The process drags on for months. Competitive tension is weak because there are not enough buyers at the table. And the client pays for all of it regardless of outcome.
As Jump Capital put it: selling a small business is slow, manual, and structurally broken. Over the next decade, roughly three-quarters of business owners plan to exit, putting an estimated $10 trillion in wealth into motion. Over 60% of family-owned businesses have no documented succession plan. The infrastructure to serve them does not exist. AI changes all of this.
After three years of watching banks fail to adopt even basic AI tools, we asked the obvious question: why sell the tool when we can do the work instead?
The distinction between selling tools and doing the work is not just our opinion. It is now a consensus across the biggest firms in venture capital. Emergence Capital originally articulated it in their "Death of the Big 4" thesis. Sequoia recently expanded it into a full framework with "Services: The New Software." YC is calling for AI-native agencies with software margins. a16z, OpenOcean, Bessemer, and a dozen other firms are all backing the thesis with billions in committed capital.
---
The Quant Hedge Fund Parallel (And the VC Preview)
Here is a question I ask every banker I meet.
In five years, would you bet your life that an M&A advisory firm with 25 bankers and 5 engineers will outperform one with 25 engineers and 5 bankers?
Five years ago, most bankers would have laughed at the question. But then ChatGPT went public and forced the entire world to understand what LLMs can actually do.
This exact transition already happened in asset management. Quantitative hedge funds went from a niche curiosity to managing roughly $1 to $1.5 trillion, around 25 to 30% of total hedge fund AUM. Algorithmic strategies now account for 60 to 73% of all US equity trading volume. The shift took two decades. But it happened.
And it is happening right now in venture capital. If you are a banker and want to see your future, read Andre Retterath's DDVC newsletter. You will be ahead of 99% of your peers. Andre's thesis is clear: the investment firm of 2030 has dramatically fewer people making dramatically better decisions. Sourcing becomes infrastructure, not labour. Screening becomes algorithmic, not vibes. The junior layer shrinks, but the people who remain become what he calls "super-analysts" operating at 3 to 5x the productivity of traditional teams.
The data backs it up. 85% of VC funds now use AI to automate daily tasks. Two-person BD teams manage 500+ intros annually with AI. In my opinion, VC is roughly 2 to 3 years ahead of investment banking in this transition. And VC is arguably more relationship-driven than sell-side M&A. If AI is reshaping venture capital, the argument that it will not reshape investment banking is weaker, not stronger.
---
What AI-Native M&A Advisory Actually Means
Let me be specific about what it means and what it does not mean.
AI-native does not mean “no humans”.
Experienced senior bankers lead our processes. Junior bankers manage them. Then we 10x them with AI automations. The AI handles buyer universe creation, outreach, material preparation, data structuring, a large chunk of the communication, and much of due diligence support. The banker handles the calls, the dinners, the negotiations, the handshakes, the judgment, the relationship. That is all human. That will stay human. And because our deal team is freed from the manual work that consumes 80% of a traditional banker's time, they actually spend more time talking to clients, more time thinking strategically, and more time on the decisions that actually move a deal forward.
But not everyone calling themselves AI-native actually is.
Now that AI-native is becoming a buzzword, a lot of traditional advisors will start rushing to slap the label on themselves. It won’t make it true.
Do you have more engineers than bankers on your team, or are you on a clear path to get there? Are you building your technological edge in-house? Is your company run like a tech startup? If the answers are no, you are a traditional advisory firm with a few AI subscriptions. That is fine. But do not call it AI-native.
Having a Bloomberg Terminal does not make you a quantitative hedge fund. Having a Claude subscription and a Rogo license does not make you an AI-native advisory firm.
---
Common Misconceptions About AI-Native Advisory Firms
1. "AI-native advisory firms can't build relationships with buyers."
This one is the opposite of reality. We are working on 10x more deals than a traditional boutique. We contact significantly more high-fit buyers on average per process. The amount of deal flow we provide to buyers is orders of magnitude more than what a traditional boutique sends them. Why would buyers not prioritise us? They want early access to the best deals. Volume of quality deal flow builds relationships faster than occasional phone calls.
2. "Cold outreach to buyers doesn't work."
This is the biggest myth in the industry. Bankers have convinced themselves and their clients that only warm introductions work. It is simply not true. Imagine you are reaching out to the Head of Corporate Development at a strategic buyer whose job literally depends on finding the right acquisition target. You present them with a perfect fit on a platter, backed by clear research on why the synergy makes sense. Do you really think they need a warm intro to respond? Our experience shows a completely different picture.
3. "Outreaching to a broad list of buyers will destroy the process."
This one always makes me smile. Imagine telling a founder that talking to more than 5 VCs will destroy their fundraise. Imagine telling that to a YC founder. The level of delusion shows you how behind the industry is. Obviously outreach is anonymised and there are ways to mitigate most risks of broader distribution. When talking about SMEs, the benefits of competitive tension from a wide process far outweigh the small negatives.
As an M&A advisor, you simply cannot know the strategic direction of every potential high-fit buyer on every deal. That is especially true on the smaller deals where the options are so many. Broad outreach to a targeted long list is how you make sure no serious option goes unexplored.
And yes, some niche processes require narrower outreach. But that is the exception, not the rule.
4. "You need retainers to ensure the client is serious."
Maybe traditional advisory firms do. Not AI-native ones. We can test how serious a client is by working on the process. We do not invest as much human capital upfront, so we can easily drop unserious clients. But the opposite keeps happening. Our momentum is so quick that clients become more and more engaged with every day, especially when conversations with real buyers start within 2 weeks instead of months. I have had this debate with bankers multiple times. I get the arguments. Retainers can be a powerful tool. But they can also be the only source of revenue for an incompetent advisory firm that barely closes deals. Our thesis is simple: if we cannot close deals, we die. So we better be closing deals.
5. "AI-native advisory firms only fight for the long tail of bad deals."
Could not be further from the truth. The moment we show a client what we can do in a sales meeting is the moment they are sold. Especially when the person on the call is a senior banker with real reputation, real experience, and real deal track record, backed by technology that traditional firms simply cannot match. This is in the infancy stages. Imagine what it looks like in two years. Traditional boutiques will not be winning competitive pitches against this. And that is completely disregarding the horror stories we hear from clients about their previous M&A processes.
6. "AI-native means lower quality execution."
The opposite. Small deals have always been economically unfeasible for banks running full processes. So they either ignore them or cut corners. AI-native advisory firms do not need to cut corners. The AI handles the volume. The senior banker handles the quality. The client gets the full process regardless of deal size. And on top of that, we do it at lightning speed. Time kills deals. You need momentum. Our process can go from first call to buyer outreach in days. Teaser created, 300+ high-fit buyers identified and contacted. CIM and financial model built within a week. NDAs signed and management meetings scheduled within two weeks. First non-binding offers in less than a month. That is not lower quality. That is what institutional-grade execution looks like when you remove the inefficiency.
7. "You might be fast, but everyone else will slow the deal down."
This is the best objection on this list. And it is partially true, for now. But the ecosystem is catching up faster than people think. AI-native law firms like Lawhive are already compressing legal timelines. Traditional law firms are adopting tools like Harvey. Buyers are deploying AI for due diligence. Accountants are automating audit prep. The entire chain is accelerating. And the firms that move first will naturally gravitate toward each other, creating deal ecosystems that are fast end to end. On top of that, we create real competitive tension on our deals, so buyers have an additional incentive to move quickly.
---
What Happens Next in Investment Banking
Stage 1: Denial. Already happening. Bankers dismissing posts like this one, saying AI will never understand the nuances of deal-making, saying you cannot automate trust. They are right that AI cannot replace judgment or trust. They are wrong that this means nothing changes.
Stage 2: Tool adoption. Banks will start buying every AI sourcing tool, data provider, and automation platform available. Claude will be installed by default instead of junior bankers having to fight to use it (we use Claude too. One thing is certain: every bank will be using a frontier AI model, AI-native or not). Goldman Sachs has already deployed its AI assistant to 46,500 employees. Morgan Stanley has Debrief. JP Morgan has LLMSuite. The big banks are moving. The boutiques are still watching.
Stage 3: Build vs buy. More advisory firms will realise that buying off-the-shelf AI tools gives them no edge because their competitors have the same tools. Tool adoption is just scratching the surface of what is possible. The whole process needs to be rebuilt with AI in mind. The ones that realize that will start building proprietary technology. This is where most banks will fail because they do not have engineering DNA. And the AI tools built for M&A by external software providers will always be inferior to what AI-native advisory firms build internally. It is one thing to build software for a problem you observe from the outside. It is another to have all the skin in the game. If our tech does not perform, the advisory firm dies. That feedback loop produces better technology than any SaaS vendor can match.
But what about the banks that skip the vendors and build themselves? Even when traditional banks do hire engineers, having engineers is not the same as having engineering DNA. We have met banks with large engineering teams that are still years behind. Their entire business model was built around human leverage, not technology. To truly compete, they would need to fundamentally restructure how they operate. Some will, and when they do, they will effectively become AI-native firms themselves. But that transformation takes years, and the firms that started from scratch are compounding their advantages with every deal.
Stage 4: AI-native advisory firms multiply. Today, the main examples are Eilla (Europe), OffDeal (US), and M&A Research Institute (Japan, built by Shunsaku Sagami, Japan's youngest billionaire at 33, who became one specifically from AI-powered M&A). But the wave has started. I predict that by the end of 2026, there will be 100+ firms calling themselves AI-native M&A advisory firms. Most will be empty claims. But enough will be real to permanently shift the industry. Especially now that VCs are committing billions to back AI-native service providers. A category that was considered an uninvestable services model just two years ago is now being called "the next trillion-dollar company" by Sequoia.
Stage 5: The holdouts. A select few very traditional, very relationship-driven bankers will forever refuse to change. And that is fine. It will be the equivalent of a manual gearbox car. You just want "the real thing." Which will come with fewer buyers contacted, less competitive tension, and lower valuations. But at least it was done only by overworked humans. And no, that is not the equivalent of a handmade leather bag. It is the equivalent of a dentist pulling your teeth with pliers like in the 18th century.
---
Why the Timing Has Never Been Better
Take M&A Research Institute in Japan. Shunsaku Sagami built it to capitalise on Japan's succession crisis: over 1.25 million small-business owners turning 70 with no succession plan, 6.5 million jobs at risk, over $100 billion in potential GDP loss. He used AI to compress deal timelines from 12+ months to 6.2 months on average. The result: a publicly listed company and a $1.9 billion personal net worth. TIME profiled him as a next-generation leader.
Guess where the next biggest succession crisis is. Europe. The European Commission estimates that a third of EU entrepreneurs will withdraw from their businesses in the next decade, putting roughly 7 million businesses and 30 million jobs at risk. In Germany alone, 626,000 SMEs plan ownership transfers by 2027, and 231,000 are considering closure because they cannot find successors.
But it is not just succession. The AI boom itself is creating an unprecedented wave of M&A activity. Vibe coding and tools like Cursor and Lovable gave birth to a whole new generation of entrepreneurs building startups at record pace. These startups are also being acquired at record pace. Google closed its $32 billion acquisition of Wiz, just five years after Wiz was founded. General Catalyst committed $1.5 billion specifically for AI rollup strategies: acquiring traditional service businesses and transforming them with AI.
Record deal volume. Record startup creation. Record succession pressure. Record capital committed to AI rollups. All at once. The firms best positioned to capture this wave are not the ones with the most bankers. They are the ones with the best technology.
---
This Is Not Just About Investment Banking
Most of what is in this post applies to law, insurance, accounting, tax, compliance, and consulting. Lawhive raised $60M for an AI-native law firm growing revenue 7x year over year. Harper raised $47M to build an AI-native insurance brokerage backed by Emergence Capital. Rillet raised $108M from Sequoia and a16z for AI-native accounting. McKinsey cut 11% of its own headcount while deploying 12,000 AI agents.
The pattern is the same everywhere. Regulated, relationship-driven, slow to adopt technology, built on human capital leverage. These are not reasons AI will skip your industry. They are reasons it will hit harder.
And guess what. AI-native M&A advisory firms will prefer to work with AI-native law firms on their deals, for the exact same reasons explained above. The entire professional services ecosystem is shifting together.
---
One Last Thing
Before you instantly push back on any of this, consider something. The automations I am talking about go far beyond creating teasers, CIMs, and buyer lists. Just because you have been in M&A for 20 years does not mean you have all the answers about where M&A in the world of AI can go. Frankly, no one knows exactly where any industry will end up. That is why it is so important to stay open-minded in this fast-moving world (Henry Ford’s faster horses).
And I actually believe the technology itself is not the long-term moat. All banks will eventually have engineers and AI. The Red Queen Effect will take care of that. But the first movers will have data flywheels, proprietary workflows, and network effects that compound with every deal. Those are much stickier moats than a legacy brand built on a model the market is leaving behind.
The future M&A advisory firm will run deal teams of one super-human senior banker, one super-human junior banker, and AI. A single deal team like that will consistently outperform much larger traditional teams. Not in ten years. Not in five.
The future has already started.
---
A Note to Bankers
If you are a junior banker: the layoffs are already here. Investment banking layoffs increased 28% in 2025. Goldman, Morgan Stanley, Citi, Barclays, all cutting thousands of roles. Mid-level associates and VPs face the highest layoff risk. If you are not constantly using AI every single day (which frankly will be a stunning fact on its own), start now. Build decks with Claude. Automate your research workflows. Either position yourself as the AI-native junior who can do the work of three, or think seriously about alternatives: financial lead at a startup, start your own company, or join an AI-native advisory firm. There will be many soon.
If you are a senior banker: AI-native advisory firms need you. Your judgment, your relationships, your ability to read a room and close a deal. That is what AI cannot replace. But the juniors doing your slide decks and buyer lists at 2am? AI is already better at that. The question is whether you want to keep managing a team of 10 to do what a team of 3 with AI can do faster.
