Every CIM Says AI Now

On a recent Private Equity Funcast, ParkerGale partners Devin Mathews and Jim Milbery spent an episode on a problem every middle-market buyer now shares. Every CIM that crosses the desk claims AI.

The pitch is on every page. The hard part is what comes next: figuring out what the claim is worth, and telling the real thing from the mechanical Turk before you price the deal.

The gap is wider than it sounds, because the buyers often can't read the claim either. Kyle Roemer runs data and AI at Accordion and has worked across 350 private equity sponsors. His read on the last year: "There were a lot of sponsors who could technically spell AI, but still didn't really know how this stuff worked. There's a lot of hype. There's always been a lot of hype."

So, bottom line, diligence has to get concrete. Kirby Montgomery, who leads North America for the tech diligence firm Code & Co, has been on the seller's side of the oldest claim in the book. "We would tell people we have proprietary data," he says, "but I knew we couldn't do anything with it, because the architecture wouldn't allow us to get there." The proprietary-data slide is easy to write. Whether the architecture can actually feed that data into a model is the question, and the answer is often no, at least not easily.

The unit economics changed too. "It was cloud compute, now it's cloud plus AI," Montgomery says. "And if you get more users, do you understand whether that's actually going to erode your margins?" Cloud costs shrank with scale. AI inference doesn't. Growth can now make the margin profile worse, and that belongs in the model, with the challenges on pricing future inference.

The claim often moves the price whether or not it holds up. Roemer watched AI promises prop up valuations that should have come down. "The valuations didn't get adjusted," he says, "so the companies were saying, don't worry about that, because AI initiatives are going to drive growth back up to 30, 40%. That wasn't materializing." This is where the new processes and work need to materialize from the diligence teams and operations teams to make sure these can be real.

So how do you price a claim like that? You get concrete. Can the architecture actually feed the data into a model without a rebuild. Do more users help the margins or quietly erode them. Is the AI in the product doing work, or decorating the slide.

With all new technology the work is separating the hype from the value. This is the new work of deal teams, it will take new skills, new tooling and new process. Most firms will need to bring on specialized talent to work through this change until the AI fluency builds across the entire deal team.

#in first comment: See my post on the rise of the AI operating partner.