The Revenue Cycle
Runs Both Ways.

Cuban's role in this is less about the pharmacy he built than about the permission structure he has been creating. By naming CVS Caremark, Express Scripts, and OptumRx in public, by publishing the list prices his facility pays, by calling out employer CEOs for not negotiating better, he has made it normal to treat healthcare pricing and denial the way we treat any other supplier negotiation: as a leverage problem. That framing is what this issue tries to extend. The hard part is not writing an op-ed. It is building the infrastructure that lets a family who cannot afford a $14,000 emergency department bill have a chance of not being crushed by it.

The Himmelstein team's 2019 AJPH paper is still the cleanest number we have: roughly 66.5% of personal bankruptcies in the U.S. cite a medical cause, which translates to about 530,000 families filing for bankruptcy each year with medical bills or illness-driven income loss contributing materially. Subsequent work by the Commonwealth Fund and KFF finds that the underlying affordability crisis has widened, not narrowed, since then.

Three things about that number are worth sitting with before we go any further.

First, most of these families have insurance. The Commonwealth Fund's 2024 Biennial survey found that 23% of insured adults are underinsured — meaning their deductibles and out-of-pocket spending consume more than 10% of household income (5% for lower-income families). Medical bankruptcy is largely an underinsurance story, not an uninsurance story. That is important because the "just get covered" framing — the default political answer for a decade — does not describe what is actually breaking.

Second, the triggers are routine. A single hospitalization is the modal event — one in four bankruptcy filers cite a specific stay. Cancer and chronic disease account for another large share. Commonwealth Fund analyses and KFF data consistently show that unpaid hospital bills are the single largest category of medical debt, followed by out-of-pocket prescription costs and ambulance/ED charges. Dental and behavioral health debt is also material and often invisible in official statistics.

Third, the demographic pattern is not random. Black and Hispanic households are disproportionately represented. Single-parent households are over-represented. Residents of non-Medicaid-expansion states are over-represented. And — this is the quiet one — the age group most likely to carry medical debt in collections is working-age adults between roughly 35 and 54, not retirees. The system is not mostly failing the elderly. It is failing people who are in the middle of building their working lives.

One more piece of the picture — policy has moved in the wrong direction over the last year. The CFPB's January 2025 rule would have removed roughly $49 billion of medical debt from the credit reports of an estimated 15 million Americans. It was vacated by a federal court in the Eastern District of Texas on July 11, 2025, which also signaled that the Fair Credit Reporting Act preempts state-level medical debt credit bans — putting about fifteen state laws in legal jeopardy. Add to that the post-pandemic Medicaid unwinding (over 25 million redeterminations, many procedural disenrollments), the expiration of enhanced ACA subsidies on the current schedule, and eligibility redeterminations every six months under the "One Big Beautiful Bill" — and the forward trend on medical bankruptcy exposure is worse than the historical data says.

Take the cleanest example. Nonprofit hospitals — about 3,000 facilities, roughly 60% of all U.S. hospitals — receive federal, state, and local tax exemptions collectively worth an estimated $28 billion a year. In exchange, under IRS Section 501(r), they are required to publish a financial assistance policy, screen patients for eligibility before collection actions, and limit what they charge eligible patients.

The Lown Institute's 2024 Fair Share Spending analysis found a $25.7 billion gap between the value of nonprofit hospital tax breaks and the community investment those hospitals reported. Separate research from the nonprofit Dollar For quantifies the charity-care-specific piece: hospitals collectively fail to deliver at least $14 billion a year in charity care they were legally obligated to provide, and only about 29% of eligible patients actually move from discovery to approved financial assistance. More than half of patients surveyed were not given any information about financial assistance by the hospital at all. Black patients were 62% less likely to be approved than patients of other races.

Charity care is only one of the leaky resource pools. A similar pattern shows up across the system:

The pattern is consistent. Resources exist. Take-up is low. The infrastructure that could raise take-up — eligibility checks, application prefill, subsidized enrollment, automated charity-care screening — is technically trivial. It is just not built by anyone whose revenue depends on it.

That last sentence is the one worth reading twice. The same data flows that feed a collection engine could, with different incentives, feed a benefits engine. Eligibility for Medicaid, for hospital charity care, for marketplace subsidies, for state assistance programs, for manufacturer patient-assistance drugs — all of it is computable from information the health system already has: income, household size, residency, insurance status, diagnosis codes, discharge data. The pay-side revenue cycle is not a technical problem. It is an assignment problem: who is accountable for running it, and whose side they sit on.

There is a second myth worth naming. Medical bankruptcy is not, mostly, a problem of educated patients making bad choices — it is a problem of ordinary people hitting an intentionally complex administrative wall with no one on their side. The people who are already on that side are underfunded, overloaded, and mostly working on paper and phone calls. Before we ask who the next Dollar For is, we should ask how much more the existing field can do with better tooling.

There is a version of this story that ends with "and then a brilliant AI startup solved healthcare." Nothing about the evidence above supports that version. 530,000 families a year go bankrupt in this country because of a financing and enforcement structure, not because of a missing app. If anything, the last three years have shown that frontier AI, left to the current incentive structure, tilts the table further — because the side with money to spend on AI is the side already winning, and the side that needs it the most has no budget, no distribution, and no legal access to the data.

The more honest version is quieter. The same signal that powers an RCM engine can power a benefits engine. The same eligibility logic that drives coverage denial can drive automatic charity-care screening. The same natural-language generation that drafts clawback letters can draft appeal letters. The same claims data that identifies high-cost patients for aggressive collection can identify them for presumptive Medicaid enrollment. There is no technical invention missing. There is a set of policy decisions about whose side the infrastructure sits on — and those decisions, unlike physician pipelines or hospital construction, can be made in months.

Cuban's framing, at the end of the day, is the right one. If you are the CEO of a self-insured employer, an elected official with a healthcare committee seat, an attorney general, a hospital CFO, a nonprofit funder — you do not need to wait for the next model release. You need to decide whether the revenue cycle you authorize runs in one direction or both.

Don't be a wimp.