Medicare's AI Payment Model Will Reset Health Tech

The Centers for Medicare and Medicaid Services just did something the tech industry mostly missed: it built a payment model from the ground up designed to pay AI companies for what AI actually does. The ACCESS program, Advancing Chronic Care with Effective, Scalable Solutions, launches July 5, 2026, with 150 participating organizations and a 10-year runway. Medicare controls $900 billion in annual payments. When it decides something is reimbursable, the rest of the US healthcare market follows within 18 to 24 months. That's the mechanism most of the tech world isn't watching.

What Actually Happened

The Centers for Medicare and Medicaid Services selected 150 organizations to participate in ACCESS, a 10-year program designed to test AI-driven care coordination at federal scale. The program goes live July 5, 2026. Pair Team, a care management company focused on high-cost complex patients, was among the first participants announced. ACCESS was designed by Abe Sutton, director of the CMS Innovation Center, and Jacob Shiff, chief AI and technology officer of the CMS Innovation Center. Both have startup backgrounds, and the program's structure reflects that orientation: outcome-based payments, direct-to-consumer enrollment, and deliberate competitive pressure among participants to produce better results.

The core structural shift is what Medicare is agreeing to pay for. Traditional Medicare reimbursement covers services delivered during clinical encounters: office visits, procedures, diagnostics. ACCESS pays participating organizations for AI-driven care coordination that happens between those visits. Phone calls where an AI identifies that a diabetic patient hasn't refilled their medication. Automated outreach when a care gap opens. Proactive risk stratification that routes a high-risk patient to a care manager before they hit the emergency room. These are activities that AI systems are well-suited to run at scale, and they have historically fallen outside Medicare's reimbursement framework. ACCESS closes that gap on a 10-year experimental timeline.

Why This Matters More Than People Think

Medicare's role in setting payment standards for the entire US healthcare system is not widely understood outside the industry. The $900 billion in annual Medicare payments functions as a pricing signal that ripples through every major insurer in the country. When Medicare decides a category of service is reimbursable, private insurers typically add it to their coverage frameworks within 18 to 24 months. When Medicare decides something isn't reimbursable, the entire market treats it as financially unviable regardless of clinical evidence. The ACCESS program is not a pilot in the ordinary sense. It's a 10-year, federally funded determination of whether AI-driven care coordination generates enough measurable outcome improvement to justify a permanent payment category.

The implications for AI health startups are direct. The single largest obstacle to scaling AI-powered healthcare companies in the US has not been technology or clinical evidence. It's been the absence of a reimbursement pathway. Investors have historically been reluctant to fund healthcare AI companies at venture scale because the revenue model depended on either employer-sponsored insurance contracts that required lengthy sales cycles or out-of-pocket consumer payments with limited market size. ACCESS creates a third pathway: direct Medicare reimbursement for AI coordination services, at federal scale, with a defined outcome measurement framework. That changes the unit economics of healthcare AI businesses in a way that no amount of clinical proof alone could achieve.

The Competitive Landscape

The 150 organizations selected for ACCESS span a range of types, from traditional health systems using AI tools to purpose-built AI-native care companies like Pair Team. The selection criteria rewarded organizations that could demonstrate existing relationships with complex patient populations, infrastructure for outcomes measurement, and AI capabilities already deployed in some form rather than planned. That selection dynamic means the first ACCESS cohort is likely dominated by companies that were already running some version of AI-driven care coordination on their own economics, before reimbursement existed. Their competitive advantage now extends to using ACCESS revenue to scale operations that slower competitors will only begin building once the payment model is confirmed.

The risk is real, however: critics point out that Medicare's history with outcome-based payment experiments is not uniformly encouraging. Several previous CMS innovation programs produced strong early results that didn't hold up as participants scaled and adverse selection effects dissipated. The ACCESS program's 10-year runway is designed to mitigate that risk by requiring sustained performance over time rather than rewarding early efficiency gains. Skeptics argue that the fundamental challenge of AI-driven care for complex patients is not coordination quality. It's that the highest-cost patients have social determinants of health, including housing instability, substance use, and mental illness, that AI coordination systems aren't equipped to address regardless of how well they identify care gaps. The risk is that ACCESS proves AI can find problems that still can't be solved.

Hidden Insight: This Is the Missing Revenue Model for Healthcare AI

The healthcare AI market has produced dozens of companies with compelling clinical evidence and almost no sustainable revenue. The pattern is consistent: a startup demonstrates that its AI reduces hospital readmissions by 15%, or catches medication errors that human review misses, or identifies high-risk patients before they deteriorate. Health systems express interest. Procurement moves slowly. The pilot succeeds. Scaling stalls because no one will pay for the ongoing service at a price that supports the business model. ACCESS doesn't solve that problem for every category of healthcare AI. But for care coordination specifically, the largest category of AI-assisted activity in chronic disease management, it creates a reimbursement mechanism that didn't exist 12 months ago.

The deeper strategic shift is what happens when private insurers respond. United Health Group, CVS Health's Aetna, and Cigna collectively cover more than 100 million Americans. All three have been building or acquiring AI capabilities for care management for the past three years. When Medicare's ACCESS results start showing measurable outcome improvements, those insurers face a choice: add similar reimbursement frameworks to their own commercial plans, or watch the AI-native care companies that ACCESS funds become structurally more competitive in the Medicare Advantage market. That competitive pressure is what converts a CMS pilot into an industry standard. The 18 to 24 month private insurer follow-on window from ACCESS's July 2026 launch puts the first commercial replication decisions in late 2027 and early 2028.

There is a dimension of ACCESS that deserves attention beyond the payment mechanics. The program explicitly rewards competition among participants. Organizations that demonstrate superior outcomes can access higher payment rates over the program's 10-year timeline. That structure is unusual for a federal healthcare program and reflects the startup sensibility of its designers. It means ACCESS will generate longitudinal performance data comparing different AI approaches to care coordination at scale, across diverse patient populations, over a decade. That dataset, if CMS makes any version of it available for research, could become one of the most consequential inputs to AI health research in the 2030s.

What to Watch Next

The July 5, 2026 launch date is the first concrete milestone. Watch for which participant organizations publish their initial enrollment numbers and baseline patient population characteristics in the first 90 days. The organizations willing to be transparent about their starting conditions are likely the ones most confident in their results trajectory. Watch also for which major health systems not selected in the first cohort apply for expansion slots, which CMS has indicated will be available in 2027. The expansion applications will reveal which health systems view ACCESS as a competitive threat rather than an optional program.

Watch the private insurance market in the second half of 2026 for any announcement of outcome-based AI care coordination payment frameworks. If any major insurer moves before ACCESS has published 12-month data, it signals that internal evidence from their own AI investments has already convinced them the model works. The long-term indicator: whether the $900 billion Medicare payment system creates a new category of AI-native care companies worth more than $10 billion each by 2030, the threshold at which the ACCESS experiment can be called a category-defining policy decision rather than a well-funded pilot.

When Medicare decides something is reimbursable, it doesn't just change one program. It rewrites the business model for an entire industry.

---