Medicare Reimbursement: The AI Adoption Blueprint for Investors

Everyone talks about diagnostic AI changing healthcare, but getting FDA clearance is just the opening act. For growth equity investors, the real market-entry risk is figuring out how the startup gets paid. If you don’t understand the maze of Medicare reimbursement, you can’t properly value a company. This piece breaks down the different Centers for Medicare and Medicaid Services (CMS) payment mechanisms and what they actually mean for getting hospitals to adopt a new tool.

The Nuance of Novelty: New Technology Add-on Payments (NTAP)

The New Technology Add-on Payment (NTAP) program is the first stop for many new medical technologies, and that includes a lot of AI diagnostic tools. It’s designed to give a financial boost for technologies that are new, offer a real clinical improvement, and don’t fit neatly into existing Diagnosis-Related Group (DRG) payments. For an AI product, getting NTAP approval reduces the financial risk for a hospital thinking about adoption, ensuring the CFO isn’t penalized for trying something that might have a higher upfront cost. Take Viz.ai’s stroke detection software. Its NTAP designation gave hospitals a direct financial reason to bring the tech in-house because it covered the costs of getting it up and running. HeartFlow’s AI-driven CT-FFR analysis also used specific CMS reimbursement codes to get on the market, showing how critical these paths are. For hospitals, the math is simple: NTAP can pay up to 65% of the technology’s cost (or even 75% for some things like sickle cell gene therapies or certain infectious disease products), which improves their margins on any case where the AI is used. The catch? NTAP is temporary, lasting only two or three years. That’s not a lot of time. This means a company has to have a plan for what comes next, usually by aiming to get a permanent CPT code or get folded into a DRG. For investors, the post-NTAP strategy is everything. The “valley of death” between when the NTAP money runs out and permanent reimbursement kicks in is very real and can kill a company.

The Long Game: Category I CPT Codes and the Physician Fee Schedule

NTAP is a temporary fix, but the real endgame for any diagnostic AI is a permanent Category I CPT code. These codes, which are managed by the American Medical Association (AMA), are the foundation of the Medicare Physician Fee Schedule and define what doctors can bill for. Getting a Category I CPT code is a signal that your tech is widely accepted and gives you a stable, long-term way to get paid. But getting there is a slog. It’s a multi-year process that demands a mountain of clinical evidence showing the AI is effective, useful, and already in widespread use. Most companies have to start with a temporary Category III CPT code just to gather the necessary data. Reimbursement rates for Category I codes can be all over the map, depending on how complex the service is. For an investor, a company that already has a Category I code is a much more mature, de-risked bet. But what about the ones still on a Category III code? That’s a much higher-risk profile, since there’s no guarantee they’ll ever make the jump to Category I. It’s like Michael Chernew’s work on value-based care points out: you can’t have sustainable innovation in medicine if you don’t have a payment system that lets hospitals adopt new things without going broke.

Pass-Through Payments: A Niche, But Important, Pathway

Beyond NTAP and CPT codes, there’s another, more niche option: pass-through payments. These can help some diagnostic AI tools, especially if they’re part of an outpatient procedure or a bundled service. Pass-throughs allow for a separate payment for new devices, drugs, and biologicals used during an outpatient visit. It’s less common for a standalone AI software (SaMD), but it’s very relevant if the AI is embedded inside a new piece of hardware or a diagnostic kit. The bar is high. To qualify, the tech has to be new, a clear clinical improvement, and expensive relative to the procedure it’s being used in. Just like NTAP, pass-through status is temporary. It’s meant to provide a short-term payment solution while more data is collected for a permanent rate. This can be a huge advantage for an AI solution that’s a core part of a new diagnostic device that is also trying to get on the market. For an investor doing diligence, a key question becomes: can this AI piggyback on a pass-through payment intended for the larger device it’s a part of?

Investor Diligence: Key Questions for Reimbursement Strategy

For a growth equity investor, looking at an AI startup’s reimbursement strategy is just as important as kicking the tires on the tech itself. Here are the questions you should be asking:

• What is the company’s current reimbursement status? Are they on NTAP, using temporary Category III CPT codes, or billing with miscellaneous codes? Be specific.
• What is the roadmap to permanent reimbursement? I want to see the evidence generation plan. How, exactly, will they get from a temporary fix like NTAP or a Cat III code to a stable Category I code or DRG inclusion?
• How strong is the clinical evidence supporting the AI’s value proposition? CMS and other payers want hard proof of better patient outcomes, lower system costs, or real efficiency gains. Does the data actually show that? MedPAC reports on new technology reimbursement
• Has the company engaged with CMS, MedPAC, and relevant CPT editorial panels? Getting in front of these groups early and often is the only way to have a say in future payment policies. Are they doing it?
• What is the projected reimbursement rate under different scenarios? How does that number compare to the hospital’s actual cost to implement and use the AI? (The margin has to be there).
• What is the competitive field for reimbursement? Are five other AI tools trying to get the same CPT code, which could create a confusing mess and hurt everyone’s chances? The fact that Medicare grants NTAP approval to only a select few AI tools shows how selective these pathways are. And the big difference in payment between a Category I CPT code and an NTAP shows how much is at stake financially for a hospital’s bottom line and their decision to adopt.
  

  Methodology and Source Note

This analysis is based on official CMS guidelines, MedPAC reports to Congress, and summaries of CMS NTAP decision documents. The data points on Medicare NTAP approval rates for AI tools and the average reimbursement rates for Category I CPT codes versus NTAPs are pulled from publicly available CMS regulatory databases and MedPAC analyses, they are the authoritative sources on the economics of this stuff. CMS New Technology Add-on Payment fact sheet Figuring out Medicare reimbursement isn’t just some regulatory chore. It’s the core factor that determines whether a diagnostic AI startup will succeed or fail commercially. Investors who really dig into a company’s reimbursement plan, and who understand the financial impact of these different CMS pathways, will be far better at picking the ventures that can actually grow and avoiding the ones that are heading for a cliff.