Every founder in the diagnostic AI space understands the FDA clearance milestone. It is the first major regulatory gate, and for good reason — demonstrating safety and effectiveness to a rigorous federal standard is a meaningful achievement. But a pattern has emerged in the past three years that deserves more attention from early-stage investors and founders alike: FDA clearance does not predict commercial traction. In fact, there are now dozens of FDA-cleared diagnostic AI tools that are generating almost no revenue. The bottleneck is not regulatory. It is reimbursement — and more specifically, the gap between clearance and a reimbursable Category I CPT code.
The path from FDA clearance to CMS coverage is long, uncertain, and often underestimated. When a diagnostic AI product earns 510(k) clearance, it is legal to sell in the US. But insurance reimbursement — which determines whether hospitals and health systems will actually adopt it — typically requires a separate process through the American Medical Association's CPT Editorial Panel, then coverage determinations from CMS and commercial payers. This process routinely takes three to five years after FDA clearance, and it is not guaranteed to conclude favorably.
The consequence is that companies with cleared products must either find ways to generate revenue through hospital capitation arrangements, direct-pay models, or niche self-pay populations — or they must bridge the reimbursement gap with capital. For well-funded Series B and C companies, this is a solvable problem. For early-stage companies that raised on the assumption that clearance would unlock a reimbursable market, it is a fundamental business-model stress.
At Basil Health, we have started to weight clinical evidence strategy much more heavily in our diligence on diagnostic AI companies. Specifically, we look for founders who have a coherent plan for generating the health-economic evidence that CMS and commercial payers require — not just the clinical-utility evidence that FDA requires. These are different evidence standards. FDA asks whether the tool is safe and effective at the endpoint level. CMS asks whether it improves outcomes in ways that reduce downstream costs or expand access. Building clinical utility evidence that satisfies both standards simultaneously requires intentional study design from the earliest clinical programs.
Founders who have thought carefully about payer strategy before their Series A — who have modeled the CPT pathway, engaged with specialty society coding committees, and understood the difference between a Category III (tracking) code and a Category I (reimbursable) code — are substantially better positioned than those who treat reimbursement as a post-clearance sales problem. We now ask about this directly in our first meeting. The quality of the answer is one of the stronger signals we have for whether a founding team has the clinical-operational depth to take a diagnostic tool to meaningful scale.
The companies in our portfolio that are navigating this well share a common characteristic: they entered clinical validation with payer evidence requirements already modeled into their study design. That discipline is not accidental — it is the result of founders who understood the full commercialization path from the beginning, not founders who discovered the reimbursement problem after clearance. For investors evaluating diagnostic AI companies, the question is not just "what is your FDA strategy?" It is "what is your CMS strategy, and when did you start building evidence for it?"