Buyers sizing up a private company today ask a question that barely existed three years ago: how much of the work this company gets paid for could AI do instead? For a lot of businesses the honest answer is some of it, and that answer is already changing what companies sell for. Most owners hear about AI as a story about their operations, meaning tools to adopt and hours to save. The bigger change is happening to the price a buyer would put on the whole company, and because that repricing happens on the buyer’s side of the table, an owner who never asks the question about their own business will get the answer for the first time inside an offer, long after the chance to change it has passed.
To see why, it helps to remember what a buyer is actually paying for. The price of a business is mostly a bet that its profits will still be there in five years. For decades, what protected those profits was some mix of expertise (knowing how to do something complicated), labor leverage (doing it more cheaply with trained people), information advantage (knowing something the customer didn’t), and position (sitting between parties who needed each other). Those are exactly the advantages AI is eroding fastest. A bookkeeping practice, a staffing firm, a marketing agency, and a compliance consultancy look nothing alike on the surface, yet each gets paid for work that AI can now partly do. These are still valuable businesses. But buyers used to focus mostly on how fast a business could grow. Now they are looking just as hard at its moat: what is protecting its profits, and how much of those profits will still be there in five years.
None of this is theoretical anymore. When CVC sold the Greek e-commerce platform Skroutz to Blackstone this spring in a deal valued around $747 million, it reportedly ran the sale with AI rather than an investment bank, pointing buyers to a data portal where a chatbot answered their questions about the financials. And when OpenAI launched ChatGPT for Financial Services a few weeks ago, it built the product with Morgan Stanley and Evercore, and the live demonstration showed the platform analyzing a potential acquisition target. AI is entering dealmaking through the largest institutions first and working its way toward smaller deals, which means far more companies are going to get this kind of scrutiny, and far earlier, than human deal teams ever managed. The discount that follows rarely announces itself, showing up instead as a lower multiple or as deal terms that shift more of the risk onto the seller.
The same force is raising prices for other companies. AI strengthens certain advantages even as it erodes others, and buyers are paying up for proprietary data no model can find on the open web, for workflows a customer would find painful to leave, for work where a named person carries the accountability, and for positions protected by regulation or license. Those qualities are getting scarcer and more valuable, which means this is really a sorting rather than an across-the-board markdown, and owners have more influence than they assume over which side of it their company lands on.
The objection owners raise most often is that theirs is a relationship business and clients will never hand the work to a machine. The relationships are real, and they will outlast several generations of AI models. But a buyer is asking a narrower question than whether AI replaces you outright. The buyer wants to know how much of today’s revenue holds at today’s margins over the next five years, and no one builds a five-year forecast on loyalty alone. The second comfort, that adopting AI internally settles the matter, runs just as thin. Adopting the tools is necessary, but when every competitor has the same tools, the savings get passed through to customers at lower prices, and the question of what protects your profit is left sitting exactly where it was.
The useful response is to run the question yourself, early and honestly, the way a buyer eventually will. Go through your revenue line by line and ask what a capable model does to each piece of it over the next five years. Then use the time before any sale to build value around what AI makes stronger, not what it replaces. That might mean owning the data your business generates or moving your work toward the judgment clients will always pay a person for. None of that can be done in the ninety days of a sale process, which is exactly the point. AI is repricing private companies whether their owners participate or not, and the one part of the outcome an owner still controls is whether the answer a buyer finds is one they spent years shaping.
Scott Yenor is the chief executive of Dealade, a platform that gives business owners a private way to understand, improve, and track the value of their companies.
