The Best Algorithm Won't Win Healthcare AI

Everyone is arguing about the wrong thing.

The healthcare AI conversation fixates on the technology. How accurate the model is. Which clinical problem it solves. Whether it beats the radiologist on a test set. Those are reasonable questions. They aren’t the ones that will decide who wins the market.

The force reshaping healthcare AI right now is a regulatory framework most people haven’t read. The FDA’s evolving approach to AI as a medical device is rewriting who gets to sell to hospitals and who takes the blame when an algorithm gets it wrong. If you build, buy, or invest in this space, that framework may matter more than any benchmark.

Regulation stopped treating AI like a hip implant

The old FDA model was built for products that never change. Approve a hip implant once, and it stays a hip implant. AI doesn’t sit still. It learns and drifts, and developers push updates after approval. So the FDA moved from one-time clearance toward lifecycle oversight, where you’re expected to manage the product for as long as it’s in use.

That sounds like a technical distinction. The business consequence isn’t. Compliance stops being a launch event and becomes a permanent operating cost, and permanent costs favor the players who can afford them.

Compliance became the thing hospitals buy

Procurement used to turn on performance, price, and whether the software talked to your existing systems. Regulatory readiness has moved from tiebreaker to threshold. By late 2025, most health systems had stood up AI governance committees, and vendor decisions now run through them. The checklist asks about your change-control plan, your monitoring, your bias testing.

For a small company, clearing that bar can run well into six figures before a single hospital says yes. That isn’t a line item. It’s a wall. Procurement has quietly become a filter that favors incumbents with deep compliance infrastructure, even when a smaller company has the better algorithm.

The market is concentrating, and the money knows

Capital isn’t only flowing into healthcare AI. It’s clustering at the top. AI-enabled companies captured most digital health funding in 2025, and much of that pooled in the largest firms.

We’ve seen this movie. Two vendors already hold most of the hospital electronic health record market, and both have built AI directly into their platforms. If your hospital already runs one of them, the path of least resistance is to switch on their AI rather than vet, contract, and govern a third party. Switching costs and compliance friction point the same direction: stay where you are.

The liability question nobody has answered

When an adaptive algorithm contributes to a bad diagnosis, who is responsible? The developer, the hospital, or the physician who trusted the output? There’s still no clear answer. Manufacturers appear largely insulated. Clinicians carry frontline exposure even when they rely on AI in good faith. Uncertainty like that doesn’t freeze a market. It tends to tilt it toward whoever can absorb the risk.

What this means

The FDA writes patient-safety rules. Those rules now also set the terms of competition, deciding which vendors can clear the bar and stay cleared.

Founders should treat compliance as part of the product rather than a tax bolted on at the end, because it’s becoming the moat. For a health system, the governance committee now works as a competitive gatekeeper, so it’s worth using deliberately instead of defaulting to whatever your platform vendor already sells. And if you invest, the compliance infrastructure may tell you more than the algorithm does.

The framework is still taking shape. That’s the reason to pay attention now, before the market locks in.

Douglas Grimm and I lay out the full case, with the data and the competition-policy implications, in the CPI TechREG Chronicle.

Read the full article here.

Douglas Grimm, FACHE, is a Partner and Health Care Practice Group Leader at ArentFox Schiff. I am Founder and Managing Partner of Acumen Partners, a Washington, D.C. healthcare investment and advisory firm working with medtech companies and clinical practices, including on health AI implementation.