PitchBook's institutional research dropped yesterday with a headline that shouldn't surprise anyone who's spent time in this industry — but now has a number attached. The RCM revenue pool shrinks more than 50% by 2040. Cost-to-collect collapses from 4–5% to below 1%. Forty-two point nine billion dollars in PE bets are now facing a structural question they can't underwrite their way out of.
The report — titled "AI Kills the RCM Star," published June 24 by PitchBook's Institutional Research Group — doesn't bury the lead. Agentic automation combined with intensifying competition from AI-native entrants will compress cost-to-collect from the 4–5% range to below 1%. Some participants PitchBook spoke with went further: they project a full transition to a subscription-based model, eliminating percentage-based take rates entirely.
That's not a modest efficiency gain. That's a revenue model getting repriced from the ground up. Every outsourced RCM contract signed at today's rates is being written against a floor that may not exist in seven years.
Projected shrinkage in the total RCM revenue pool by 2040, per PitchBook. Driven by agentic AI collapsing cost-to-collect from 4–5% today to below 1% — with some projecting a full shift to subscription pricing.
Since 2017, PE sponsors have deployed $42.9 billion into RCM deals — buyouts, growth equity, platform builds — all predicated on the same core thesis: take a services business running 20–30% gross margins, automate the labor, and push margins into 60–70%+ software territory. $22.5 billion in exits have already cleared. The remaining capital is the question.
The automation thesis isn't wrong. Gross margins do expand when you replace headcount with software. The problem PitchBook is surfacing is the denominator: if agentic AI compresses the fee that RCM vendors can charge, not just the cost to deliver the service, then margin expansion at the unit level doesn't save you if total revenue per claim is falling simultaneously.
The Thoreau Group's $12 billion acquisition of Ensemble Health Partners — announced this month and excluded from PitchBook's deal tally — is either the last great RCM buyout at legacy multiples, or the first one priced into the new reality. That distinction will matter a great deal in 2029.
The most concrete signal in the report isn't a forecast — it's Waystar's stock price. Waystar is a near-AI-native RCM vendor with 11% organic revenue growth in Q1 2026. By any traditional measure, that's a healthy business. Its shares are down 43% year-to-date.
Waystar — 11% organic revenue growth, near-AI-native, down 43% YTD. CommonSpirit Health — paid Conifer Health Solutions (Tenet's RCM subsidiary) $1.9 billion to exit their contract six years early, then immediately announced an insourcing partnership with Midstream Health. When health systems are paying nine-figure termination fees to get out of RCM outsourcing deals ahead of schedule, the pricing pressure is no longer theoretical.
CommonSpirit's $1.9 billion exit from Conifer is the kind of transaction that defines a turning point. Health systems don't hand over $1.9 billion to escape a contract early unless the alternative — staying in at current rates while the market moves — looks worse. That calculus is now visible to every CFO evaluating their outsourced RCM arrangements.
PitchBook's analysis identifies the competitive dynamic clearly. The threat to traditional RCM isn't just that AI-native vendors can automate better. It's that the most dangerous competitors — EHR vendors like Epic and ModMed, clinical AI platforms, ambient scribes — treat any RCM revenue as purely incremental. They're not building RCM businesses. They're extending platforms they're already paid for.
Innovaccer's May acquisition of CaduceusHealth illustrates the vector: a data and analytics company that already lives inside the clinical workflow simply acquiring its way into agentified RCM. For Innovaccer, that revenue is upside. For a standalone RCM outsourcer, it's their core.
The report is direct about the endgame for EHR-integrated players: Epic and ModMed have the clearest medium-term path to winning RCM because they hold the system-of-record moat. Switching costs are enormous. And critically, PitchBook notes that AI-native solutions not integrated into clinical workflows will struggle — the burden of clicking in and out of systems of record is too high for care providers to sustain. Integration isn't a feature. It's the filter.
If you're running an RCM department, a billing company, or managing a portfolio of outsourced RCM vendors, a few things are worth stress-testing today:
The report frames this as a question of timing, not outcome. The direction is settled. The variable is how fast the transition moves — and whether the organizations running on current-model RCM get ahead of it or react to it.
PitchBook publishes reports like this once a quarter. We publish every day. Get the trends that matter before they hit your AR.