Healthcare AI InsightsStudies in healthcare AI · by Chi-way Li

Study 1 · The commercialization of clinical AI

Cleared to Paid

Who pays for clinical AI, and why insurers pay for so little of it. A study in five phases; the first, on the money, is out now.

Cleared by FDA1,614AI devices, decisions to June 29th 2026 Ever paid per use4.8% of devices; 73 still have a route Medicare, 2024$10.3Mpaid on AI codes; 96% for coronary CT Payment cases36compared across the US, China, Germany and the UK

Phase 1 · The money

Cleared, but rarely reimbursed

The FDA has cleared 1,614 AI medical devices. Only 78 have ever had a way to be paid per use. Who pays for the rest?

The Food and Drug Administration has cleared 1,614 medical devices that use artificial intelligence, and it clears more each year. Only 78 of them, fewer than one in 20, have ever had a way to be paid for each use: a billing code, a Medicare laboratory code or an inpatient add-on payment. Today 73 still do.

Only 78 of 1,614 have ever had a per-use payment route

FDA AI-enabled devices by payment route, one square per device

73 devices still have a per-use payment route, 5 had one that has ended, and 1,536 have never had one.
  • Per-use route today (73)
  • Route has ended (5)
  • Never had one (1,536)

A device counts only when a CMS rule, a Medicare contractor or its maker names it for a code, so 78 is a floor. See who pays for every device in the full report.

Show the numbers
Payment routeDevicesShare
Per-use route in force734.5%
Per-use route ended50.3%
Never had a per-use route1,53695.2%
All devices1,614100%

Sources: FDA; CMS; Medicare contractors; makers' coding guidesHealthcare AI Insights

Where insurers do pay, the money is concentrated. Medicare's payments to doctors for AI codes reached $10.3m in 2024. Analysis of coronary CT scans took 96% of it.

Everything else is paid for some other way, mostly by hospitals and imaging groups out of their own budgets. Even AI built into scanners is usually sold separately: where makers' documents say how they sell such features, 154 of 195 are options, licences or upgrades rather than part of the machine's price.

Three in four are paid from a buyer's budget

FDA AI-enabled devices by who pays, % of total

  • 5.0%Paid per use (80 devices): the 78 named for a billing code, lab code or add-on, plus 2 maker-run services some insurers pay per use
  • 4.0%No extra payment (65): rides on a service insurers already pay for
  • 15.0%Inside a hardware price (242): scanner features, medtech systems, consumer devices
  • 76.0%From a buyer's budget (1,227): standalone software, scanner options and upgrades, per-case services

The three unpaid groups rest partly on inference from who filed each device. The hardware share is an upper bound: scanner features whose packaging could not be checked stay counted as hardware.

Sources: FDA; CMS; makers' documentsHealthcare AI Insights

The pattern holds abroad. Across 36 payment arrangements in America, China, Germany and Britain, per-use payment appears only for AI that produces a result of its own, such as a measurement or a risk score, and for AI that treats. Insurers, in short, pay for answers, not for help.

Why? One hypothesis is a verification tax. AI that promises efficiency is hard to price while a clinician still checks every case, because the time it saves is hard to see. That may explain why AI scribes have spread fastest: their mistakes are cheap to fix. A later article in the series tests the idea.

The series in five phases

Phase 1 · Live now

The money

Who pays for clinical AI, and for what?

  1. Phase 1 · Article 1 · Coming

    Paid for, but not by insurers

    Regulators approve AI devices by the hundred. Insurers pay per use for only a few; for the rest, hospitals pay, betting on time savings that are hard to bank.

  2. Phase 1 · Article 2 · Coming

    Pay for answers, not for help

    Two questions predict it: does the AI produce a result of its own, and is it inside a service that is already paid?

  3. Phase 1 · Article 3 · Coming

    Follow the money, not the model

    Who captures the value of medical AI? Mostly not the companies selling AI software to hospitals.

Full report

Phase 2 · Next

The buyer

Who actually uses paid AI, and does evidence decide?

Planned: Paid AI clusters in a few places · The verification tax · The payer becomes the buyer

Phase 3 · Later

The builders

Who wins, who gets bought, and where should AI companies sell?

Planned: The scanner eats the software · Europe's third way · From approval to payment, market by market

Phase 4 · Later

The verdict

What held up, and what did I get wrong?

Planned: What I got wrong · In their own words

Phase 5 · Later

The next wave

What changes when AI is not a regulated device?

Planned: Scribes that bill

Planned titles may change as the research goes on.

About the author

I run the research like a small newsroom of AI agents: each works to written evidence rules, a separate agent tries to break their answers, and I make the judgement calls. I'm a biomedical engineer (B.S., Penn State; M.Eng., Cornell) with an MBA from Yale, and I've worked on medical devices and medical AI in the US and China. This is an independent project; views are my own and not those of any employer.

Chi-way Li · Comments and corrections on LinkedIn