What Judy Faulkner’s Freakonomics interview tells us about Epic’s Ten Commandments, the Purpose Trust succession, the 30 percent R&D rate, and the live antitrust fight in healthcare

May 04, 2026

∙ Paid

Video Preview

🎧 Part I Podcast free on Spotify (https://onhealthcare.substack.com/podcast)

0:00

-9:40

🎧 Part II Podcast episode for paid subscribers only (see podcast feed and below paywall)

Abstract

Episode 672 of Freakonomics Radio, April 24, 2026, ~1h08m, recorded on the Verona campus.

Faulkner is 82, has done basically zero longform press in years, founded Epic in 1979, still CEO.

Reported FY24 revenue ~$5.8B per Dubner; ~$410K revenue per employee against ~15K headcount.

R&D at 30 to 35% of revenue, no formal budgets, only “projections.”

Footprint: 80%+ of Americans have an Epic record, 16 countries, 350K job applicants per year (harder odds than Harvard or med school per Dubner).

Ten Commandments include “do not go public,” “do not acquire or be acquired,” “be frugal, do not take on debt for operations,” and “what you put up with is what you stand for.”

Purpose Trust splits Faulkner’s stock into value (charity, via Roots and Wings) and vote (4 family + 5 Epic staff), with 3 health system CEOs from the CEO Council pre-committed to sue any trustee who breaks the rules.

Active antitrust: Texas AG suit on the state Free Enterprise and Antitrust Act, plus Particle Health civil case.

Key AI plays disclosed: sepsis prediction (one customer reportedly went from 2.5% sepsis incidence to 0%), a diagnosis checker (cohort-matched against ~32,500 similar patients), and Best Care Choices for My Patient.

Seth Howard’s stated priority order: customers first, Epic second, employees third, shareholders last (and employees are shareholders).

Care Everywhere shipped in 2007 after Faulkner’s pediatrician husband lost a young patient who died because her records weren’t accessible across systems.

Faulkner advised the Obama administration on HITECH and lost the vote on letting scribes enter clinician notes; ambient AI is now her quiet revenge.

Body of the essay below, no markdown symbols, plain text, written for a healthcare investor and operator audience.

1. Video Preview

2. Podcast, Part I (Free)

2. The Tesla story tells you everything you need to know about how Faulkner thinks

3. The Ten Commandments are not a wall sticker, they are the operating system

4. Care Everywhere, a dead kid, and the origin of US clinical interop

5. The antitrust fight is real, but the framing is asymmetric and probably bounded

6. The Purpose Trust as engineered immortality, and the CEO Council enforcement layer

7. Profit as a side effect, 30 percent R&D, and what that does to every competitor

8. The customer rejection model and what it means for digital health and rural systems

9. Paywall

10. Podcast, Part II (Paid)

11. AI in the EHR, the sepsis numbers, and why ambient documentation is HITECH revenge

12. The Zeke Emanuel API question and the “it’s not our data” defense

13. What changes if you’re building, investing, or selling into a system with Epic in the stack

The Tesla story tells you everything you need to know about how Faulkner thinks

The single most useful diagnostic moment in the whole interview is Faulkner’s Tesla story. She bought one for the self-driving feature, and shortly after she got it the car would, on hilly roads near her house, accelerate from 35 to 60 mph and steer itself across the double yellow line into oncoming traffic. Most CEOs telling that story go straight to the drama. Faulkner narrated it like a bug report. She gave the parameters (hilly road, 35 mph baseline), proposed a hypothesis (the system probably interpreted the elevation profile as the road ending and was attempting to “save” her by changing lanes), described the replication step (Tesla sent someone to ride with her, confirmed it, then brought a second car and reproduced the bug), and noted that Tesla “reacted quite appropriately.” Dubner basically built the entire framing of the episode around this single anecdote, and for good reason. It’s not a story about a car. It’s a story about a coder who is still, at 82, narrating reality in terms of inputs, expected outputs, observed deviations, and root cause.

That temperament is the explanatory variable for almost everything else Epic does. The same person who reproduces a Tesla bug systematically is the person who built a company where every employee is required to write up “red flags” when something goes wrong. Most of those red flags get triaged as low impact, but the ones that turn out to be clinical or financial errors get jumped on immediately. That’s not a slogan, it’s an actual operational discipline that survives at 15K employees. For comparison, the Cerner-into-Oracle integration is famously losing institutional knowledge of exactly this kind of system, and Meditech has historically struggled to retain it. Epic has the luxury of being a single codebase with a single founder who genuinely thinks like a debugger. Hard to compete with on substance, harder still to compete with on cultural transmission.

The Ten Commandments are not a wall sticker, they are the operating system

Faulkner’s Ten Commandments are posted around campus, including, on her telling, in every bathroom and break room. The list, as she walked through it on the show: do not go public; do not acquire or be acquired; software must work; reality equals expectations; keep commitments, even the unspoken ones; focus on competence, do not tolerate mediocrity; have standards, be fair to all; have courage, what you put up with is what you stand for; teach philosophy and culture; be frugal, do not take on debt for operations.

It is tempting to read these as founder-tchotchke. They aren’t. Each one of them is doing real work as a defensive moat against the standard SaaS playbook. “Do not go public” eliminates the entire quarterly-earnings flywheel that pulls public software companies toward customer extraction. “Do not acquire or be acquired” eliminates the M&A tax that makes integrated platforms degenerate into bolted-together messes (see: Allscripts/Eclipsys, Cerner under Oracle, the entire revenue cycle vendor stack). “Be frugal, do not take on debt for operations” eliminates leverage as a forcing function on revenue maximization, which is the mechanism by which most PE-backed health tech companies eventually start squeezing customers in ways the founders never would have. “What you put up with is what you stand for” is, frankly, the best one-line explanation of organizational drift in healthcare anyone has written, and probably belongs in the sidebar of every health system board deck in America.

The interesting empirical question is whether these are actually causal or just rationalizing what Epic was already going to do. The honest answer is probably both. Faulkner herself told Dubner that some of the original Epic co-founders saw the company as a cash-out vehicle, and only revealed that interest once the company started getting valuable. She had enough ownership to fight them off, and the Commandments are partly a way of formalizing that fight into a perpetual rule set so that future trustees, board members, and senior managers can’t drift into the same posture. That’s why the Commandments aren’t just culture. They’re constitutional law for a private company that intends to outlive its founder. The fact that they get reposted in bathrooms next to mirrors (Faulkner said she wanted to put them in the stalls but didn’t) is a tell that this is treated as catechism rather than mission-statement filler.

Care Everywhere, a dead kid, and the origin of US clinical interop

The Care Everywhere origin story in the interview is the kind of thing that gets glossed in most coverage of Epic, but it’s worth dwelling on because it’s the only honest answer to the “why did interop happen at all in the US” question. Faulkner’s husband is a pediatrician. Around 2007 he came home extremely upset because a young patient of his, who had a known condition that was under control, had traveled with her parents to another city, gotten sick, ended up in an ED that did not have access to her records, was treated as a generic patient instead of someone with her specific condition, and died. He kept telling Faulkner that if the records had been there, she’d be alive. Faulkner went back to Verona, asked her developers to build a system to move clinical data between Epic instances, and then, in a detail that nobody outside Epic ever talks about, sent her own people across the country to physically install Care Everywhere at customer sites and refused to let any customer not adopt it.

That last detail is the one that should land hardest for anyone who’s been frustrated by the multi-decade slog of TEFCA, Carequality, eHealth Exchange, and the broader ONC interop agenda. The federal government has been trying to coordinate clinical data movement since the HITECH Act in 2009, and progress has been incremental at best. Epic decided unilaterally, in 2007, that interop wasn’t optional, and got it deployed across its installed base by mandate. By Epic’s count, 100% of Epic organizations are now interoperable through Care Everywhere. Whatever you think of Epic as a market participant, the empirical fact is that the largest single chunk of US clinical data exchange happens because one private company decided to force it on its customers two years before the federal government started trying. That’s a real historical contribution, and it complicates any clean “Epic is anticompetitive” narrative.

It also sets up the antitrust framing in a particular way. When Faulkner says her customers should be free to work with any third party they want, she’s drawing on a real history of Epic doing the unglamorous data-portability work first. When Particle Health alleges that Epic interferes with third-party data flows, they’re drawing on a different real history of Epic being protective about who gets clinical data and for what stated purpose. Both are true at once, and the regulatory and legal fight is about where the line sits in 2026 between protecting clinical data integrity and gatekeeping a third-party developer ecosystem.

The antitrust fight is real, but the framing is asymmetric and probably bounded

Two cases worth watching. The Texas AG suit alleges that Epic violates the state Free Enterprise and Antitrust Act. The Particle Health civil case (NY) alleges that Epic interfered with Particle’s customer contracts and used market power to squeeze out competition. Faulkner’s framing on the show was, paraphrasing, “it just seems so weird to have begun with three halftime people in a basement.” That’s a rhetorical move, but it’s also pretty effective, because Epic’s market story is genuinely more “patient win-by-win for fifty years” than “rolling acquisition-driven consolidation.” The antitrust theory of harm in software has historically been about exclusionary contracts, tying, and predatory pricing, and Epic’s pricing is not classically predatory and its contracts are not classically exclusionary. The case will probably turn on specific data-access and integration-partner decisions, not on the broad “they have 40+% share” point.

What’s harder to dismiss is the API access question. Faulkner herself addressed this when Dubner read out a question forwarded by Zeke Emanuel: “Why does she inhibit independent app development and easy access to the platform? She plays favorites that inhibit innovation.” Faulkner’s answer was that Epic doesn’t inhibit app development, has more APIs than any other vendor, and that the rate-limiting step is permissions, which sit with the health system rather than Epic. Her exact framing: “We write the code to allow the data to go out to the app, but we don’t give the permission, because it’s University of Wisconsin’s data, or it’s Cleveland Clinic’s data. It’s not our data.” This is technically correct under HIPAA and TEFCA. It is also commercially convenient. Both can be true.

The likely outcome of all this litigation, fwiw, is some specific behavioral remedies (probably around treatment-purpose data exchange, possibly around Particle-style aggregator access), no structural separation, and Epic ending up marginally more open over the next 24 months than it would otherwise have been. That’s the boring base case, and it’s also the one most consistent with both the legal record and the political optics of suing a private Wisconsin company that gives away $100M a year in philanthropy.

The Purpose Trust as engineered immortality, and the CEO Council enforcement layer

The succession structure is the most underdiscussed piece of the interview, even though Faulkner laid out the mechanics with unusual clarity. Her stock is bifurcated. The value side gets given for charitable purposes, mostly through Roots and Wings, the foundation she named after asking her young kids what parents owe their children (their initial answer, hilariously, was “food and money”). Roots and Wings receives stock annually, sells it back to Epic, and Epic redistributes those shares to employees. So the company runs an internal liquidity recycling mechanism that generates ~$100M a year for philanthropy and continuously reseeds employee equity, all without any external capital event ever happening.

The vote side is where it gets governance-nerd interesting. Voting power on Faulkner’s stock goes to a Purpose Trust. The Trust has two voting blocs: four family members (with succession plans for each seat) and five Epic staff members (also with succession plans). They vote the stock under a written rule set that prohibits, irrevocably, going public, being acquired, or creating new stock to engineer around either of those rules. On top of that, the CEO Council, which is Epic’s governance-y customer body that meets annually, designates three health system CEOs whose explicit job is to sue any trustee who votes against the rules. Faulkner declined to name them on the show because the slate had recently rotated.

This structure is genuinely novel for a software company at this scale. The closest analog is Patagonia’s 2022 reorg into a perpetual purpose trust, which Dubner’s team reportedly couldn’t find a better comp for either. The differences matter though. Patagonia’s structure is mostly about climate purpose. Epic’s is about preventing financialization specifically. The CEO Council enforcement layer (customers with legal standing and stated intent to sue) is the unique piece, and to a healthcare-investor reader it should land as a strong signal that any 10 to 15 year competitive scenario built on “Epic eventually IPOs,” “PE rolls it up,” or “Microsoft acquires after Faulkner steps down” is structurally betting against the constitution of the company. The structure is untested if Faulkner dies tomorrow, which is the legitimate counterargument (Hershey’s trust has had decades of board fights), but the design intent is unambiguous and the enforcement mechanism is unique enough to deserve real weight in any long-horizon model.

Profit as a side effect, 30 percent R&D, and what that does to every competitor

The most quoted exchange from the episode, the one that’s been bouncing around healthcare LinkedIn for a week, was Dubner saying “my conclusion is, you are not profit maximizing at all” and Faulkner replying “I agree with you entirely.” Then she added that profit “isn’t a goal. It’s a side effect.” Dubner ran the math: ~$5.8B revenue, ~14K to 15K employees, gives roughly $410K revenue per employee. That’s high in absolute terms, low compared to a Google or Meta. Faulkner’s response was, basically, “but it’s fine.”

Underneath the soundbite, the substantive number is the R&D rate. Faulkner confirmed on the show that Epic spends “30, 35 percent” annually on R&D, and that there is no real budget process. The financial team makes projections, but those aren’t operating constraints. For context, Microsoft’s R&D rate is roughly 13%. Salesforce sits around 14 to 15%. Veeva, the closest “stay private until forced public, build everything” comp in vertical SaaS, spends about 22%. Epic’s R&D rate is roughly double the public-comparable median. Sustaining that rate requires either a private cap structure (no shareholders to satisfy) or genuinely unsustainable financial engineering. Epic has the first, and that’s the real moat.

The downstream implication is that any vendor competing with Epic on a feature-by-feature basis is competing against a company that can outspend them on R&D in any given category, indefinitely, without ever needing to justify the spend to a board on margin grounds. Hospital CIOs evaluating “Epic vs. point solution” decisions on a 5-year horizon should weight this heavily. The point solution will probably ship faster initially and Epic will probably catch up over the following 18 to 36 months. That’s not a guess, it’s the empirical pattern across ambient AI scribes, virtual care, patient communication, RCM modules, and most of the spaces where third-party vendors got an early lead.

Seth Howard, Epic’s EVP of R&D, made the priority order explicit on the show. Customers first, Epic the company second, employees third, shareholders last (and employees are shareholders). That stack is exactly the inverse of a public company’s. Whether you think it’s good for American healthcare or not, it explains the R&D rate, the customer satisfaction scores, and the relative absence of customer churn. Faulkner’s claim on the show that exactly one customer has ever left Epic, and they came back, is the kind of metric that no public software company at scale could survive saying out loud, because it would invite a conversation about why the company isn’t extracting more revenue per seat.