Why private equity’s most valuable skills may be the ones AI finds hardest to replicate.
There is a new colleague in private equity. It reads faster than the entire deal team, never complains about late-night data-room uploads, analyses thousands of customer contracts before breakfast and produces an investment committee paper at 2 a.m. It has no carried interest, wants no promotion and has never flown business class to a meeting that could have been an email.
Meet AI: the analyst every firm wants, and perhaps the colleague that forces the profession to ask what it actually gets paid for.
The promises are real: faster due diligence, automated screening, live portfolio monitoring, sharper operational advice. For an industry built on analysing businesses and pricing risk, AI goes straight to the core of the model.
The better question is what happens when everyone has the same intelligence. If every firm has a brilliant analyst, who has the better judgment? If everyone can produce an impeccable market map, who finds the deal that is not on it? If every portfolio company gets the same advice on pricing and procurement, who actually builds the better business?
AI will automate much of private equity’s analytical work and erode some traditional edges. Relationships, judgment, influence and execution are harder to copy. AI may get very good at telling us what to do. Getting people to do it is another matter.
Better analysis, more expensive deals?
Private equity is data-intensive. Professionals spend huge amounts of time reviewing contracts, benchmarking competitors, building models and turning it all into a thesis. AI speeds up most of this. It spots inconsistencies, compares terms, stress-tests assumptions and flags patterns a tired associate would miss. This is genuine progress, not another round of technology theatre.
But there is a catch: the more widely a capability is available, the less advantage it confers. Picture two funds bidding for the same company. Both use AI to benchmark margins and find gains in procurement, pricing and working capital. Both reach much the same conclusions. The auction is no less competitive.
It may be more so. Better analysis does not guarantee better returns; it can simply let everyone justify a higher price with more confidence. A DCF is still a set of assumptions about the future, however sophisticated the machine that wrote them.
When analysis becomes a commodity, advantage moves to whatever is hard to copy.
The founder who never read the algorithm
Take a founder-owned industrial business: €200 million in revenue, loyal customers, a promising succession story. AI can assess its history, line up comparables, size the margin upside and model returns under different financing structures. It can even draft a persuasive deck on why the company makes a good target.
What it cannot do is make the founder trust the buyer.
Maybe the founder has spent 35 years building the company and fears a financial sponsor will dismantle its culture. Maybe the real worry is the management team, not the price. Maybe the family has turned down several good offers and is waiting for the right partner. None of this shows up in a model. Much of it is never said aloud.
The investor who reads those motives, earns credibility over years and shows real commitment to the business may win before an auction begins. AI can find the company and help prepare for the meeting. But relationships are not databases with nicer interfaces. Access, reputation, timing and trust build slowly, and no enterprise subscription buys them.
There is a difference between spotting an opportunity and being the person the owner wants to call.
Knowing the answer is not making the decision
Investment committees rarely lack information. Their problem is deciding what to believe when the data is incomplete, contradictory or unhelpful. The target has strong growth but an unproven team. The asset is strategic but priced for perfection. The market looks resilient until a new technology or regulation changes its economics.
AI can model scenarios, estimate probabilities and challenge assumptions. But an investment decision is not a maths problem with one right answer. It is a judgment about trade-offs, under uncertainty, with someone accountable. Walking away is hardest when rivals are bidding hard and the team has spent six months on the chase. A model can recommend discipline. It cannot make people follow it.
And when a deal goes wrong, nobody is comforted to hear that the algorithm gave it a 73 per cent chance of success. The responsibility sits with whoever committed the capital.
The real work begins after closing
Suppose AI flags deteriorating margins at a portfolio company: unprofitable customers, heavy discounting, inefficient procurement, a sales force buried in low-value work. The recommendations are sensible and the savings substantial.
Then reality arrives. The head of sales says the analysis misreads the market. Procurement says suppliers cannot be changed that fast. Regional managers resist the new pricing. The CEO agrees in principle but balks at confronting two underperforming executives. The dashboard is green. The company is going nowhere.
Operational improvement means challenging habits, resolving conflicts, redesigning incentives and making unpopular calls. It takes credibility with management and the ability to tell a real constraint from plain resistance. AI may become a capable adviser, but value appears only when behaviour changes.
Buying a business is a transaction. Improving it is sustained intervention. That is where the thesis is proved or lost.
When every company has the same AI, who wins?
AI will also change the companies private equity buys. Portfolio businesses will use it to cut costs, improve service, speed up product development and find new revenue. So will their competitors. An AI-powered service platform lifts efficiency, but it is no moat if anyone can install it.
The real opportunities come when AI changes how a business operates, competes or creates value: a redesigned service model, a differentiated product, a new way to win customers, markets that were once uneconomic. That takes strategic choices and organisational change, not just procurement.
So the question for investors is not whether a portfolio company uses AI. It is whether its use creates an advantage rivals cannot easily match. Otherwise the technology improves the business without making it any different.
The future belongs to the AI-enabled investor
None of this means human judgment is safe from progress. AI will get better at research, planning, communication and decision support, and some systems will carry out multi-step tasks with little supervision. The answer is not to defend every task as uniquely human. Much of today’s work will become faster, cheaper or unnecessary.
The strongest professionals will know where AI is reliable, where it is fragile and where people add the most. They will spend less time producing what everyone can obtain, and more on proprietary access, unconventional theses, management relationships and operational reality. They will use AI to challenge their convictions, not to confirm them, and treat its output as hypotheses to test rather than answers to adopt. And they will not mistake polished analysis for a sound decision.
At firm level, advantage will come less from any one tool than from pairing technology with proprietary data, deep expertise, a distinctive process and superior execution.
The final investment thesis
Private equity has always blended science and judgment: financial rigour and contractual discipline, but also relationships, negotiation, leadership and the nerve to act when the future is unclear. AI shifts that balance. Analytical work will be more automated, and some skills will lose their scarcity value.
The core challenge stays the same: buy the right business at the right price, manage the risks and make it worth more before selling. AI can help with every part of that. It cannot yet guarantee the result. Perhaps the best way to see it is as a test of what the investor adds beyond the analysis.
If your edge is a better spreadsheet, the future may be uncomfortable. If it is finding deals others cannot reach, making calls others will not make and persuading people to deliver what others cannot, AI may prove an exceptionally useful colleague.
Just do not expect it to take the founder to dinner, read the room, talk the CEO into replacing the head of sales and then defend the deal to a sceptical committee. Not all in one evening, anyway.
The next era of private equity will not belong to the firms with the most artificial intelligence, but to those that pair it with the rarest forms of human intelligence, and know the difference.
