Why healthcare's massive middle layer is a builder's goldmine
The US spent $4.9 trillion on healthcare in 2023.
For that sum, you'd expect to be getting a lot of care. You're not.
For every dollar that enters the system, only about 20 cents reaches a physician. Another 10 goes to prescription drugs. The remaining 70 cents gets consumed by hospitals, insurance companies, billing departments, and administrative overhead that exists primarily to move money through the system, not to treat anyone.
That's the 70-cent problem.
And it's where most of the interesting building is happening right now.
The service economy created this mess
The US economy runs about 79% services. In sectors with real competition (e.g., retail, tech, logistics), that shift forces greater efficiency. Tight margins and actual market pressure either made companies lean or killed them.
In protected sectors, the opposite happened.
Healthcare, along with education and financial services, all developed bloated middle layers insulated from those same forces.
Healthcare is protected by regulation, locked into multi-payer complexity, sand hielded from any normal market mechanism. It’s an entire system of overhead. Insurance administrators. Hospital billing departments. Prior authorization teams. Medical coders. Claims processors. Each layer adds its own cost, delays, and paperwork. None of that delivers care.
Where the money actually goes
Back to our 70-cent problem.
That 70 cents isn't all waste. It includes nursing homes, home health, and dental care. But buried within it are those overhead divisions not focused on delivering care.
Research quantifies pure administrative bloat at 22-34% of total healthcare spending. A 2017 study in Annals of Internal Medicine put it at 34.2% — roughly $812 billion, or $2,497 per person in the US. Canada spends $551 per person on healthcare administration.
What that overhead looks like in practice:
Prior authorization. Physicians handle an average of 39 prior authorization requests per week, with 13 hours of administrative time before staff support is factored in. Studies show 93-94% of doctors report these delays harm patient care. And roughly 70% of denials get overturned on appeal (over 80% for those on Medicare Advantage). It's friction generating zero clinical value, sustained entirely by how the payment system is structured.
Billing. The per-claim cost of billing and insurance-related activities ranges from $20 to $215, depending on the setting. For professional revenue, that can eat 3-25% of total collections. In 2023, providers spent $25.7 billion just contesting claim denials.
Documentation. Time-motion studies consistently show that for every hour of direct patient care, physicians spend two hours on EHR work and desk tasks. Then another 1-2 hours after hours. The system turned doctors into data entry clerks.
Pharmacy middlemen. Three pharmacy benefit managers control roughly 80% of prescription claims. The FTC's 2025 interim report documents how these vertically integrated players create opacity and extract rent through specialty generic markups and steering.
Policy is finally creating infrastructure
The CMS Interoperability & Prior Authorization Final Rule (2024) mandates 72-hour turnarounds for expedited prior authorizations and 7 days for standard ones. It requires FHIR-based APIs for patient, provider, and payer-to-payer data exchange by January 2027. It forces transparency on denial reasons starting in 2026.
For builders, the relevant shift is this: the mandates are converting the paper trail into machine-readable data. The whole system is being forced to digitize in ways that make automation necessary to remain compliant.
Where AI fits
The tasks consuming that 22-34 cent administrative wedge aren't cognitively complex. They're tedious, rule-heavy, and volume-intensive. Pattern matching. Consistency checking. Navigating byzantine regulatory requirements. These are exactly what current AI systems do well.
The specific opportunities are measurable:
Medical coding and revenue cycle. Early AI system deployments show 30-60% reductions in coding costs, faster claim submission, and fewer denials when documentation and coding align. Against $25.7 billion in annual hospital denial costs, even modest improvements matter.
Prior authorization. The new CMS rules create forcing functions for automation. AI can ingest machine-readable payer policies, pull clinical data from the EHR, match against requirements, and generate compliant justifications.
Clinical documentation. Ambient AI tools that capture patient encounters and generate structured notes are cutting documentation time. The goal is to return patient time to physicians.
Claims processing. Rules-based systems combined with ML can pre-edit claims, flag likely denials before submission, and auto-generate appeal rationales. The $25.7 billion providers spend fighting denials is an opportunity to improve that number.
What we're backing
Home healthcare is one of the most documentation-heavy domains in the system. OASIS-E assessments, PDGM payment logic, and EVV mandates for visit verification all require meticulous paperwork. Nurses routinely spend 60-75 minutes per visit on documentation alone. That's not clinical time.
That's why we invested in Enzo Health. They're building AI-native documentation and coding tools specifically for home health agencies.
The ROI for customers is concrete: time recovered, faster cash flow from cleaner claims, lower denial rates, and compliance maintained through structured workflows. This is a product that measurably reduces a specific slice of that 70-cent overhead.
The 70-cent problem isn't going away without builders who can see it clearly enough to pick a piece of it and ship it. If that's you, we want to hear what you're building.
