Epic Systems (MyChart)

The answer is somewhere in the middle. You texted me last night that you've made it to Singapore.

I made it to Singapore.

Yes. Anytime you are researching anything in U.S. health care, it is time to stop your research process and start the episode once you've found Singapore.

Welcome to the spring 2025 season of Acquired, the podcast about great companies and the stories and playbooks behind them. I'm Ben Gilbert. I'm David Rosenthal. And we are your hosts. Listeners, today's episode is about a quiet company in rural Wisconsin that plays an enormous role in our lives: Epic Systems.

Indeed. Whether you know it or not, you probably know them from their medical patient software, MyChart, which, if you're listening to this, you most likely use. Epic is a very unusual company in so many ways. They do no marketing. They basically don't do any sales, either.

They often say no to potential customers who approach them. They don't negotiate, they don't discount, they never raised any venture capital, and they've never done any acquisitions in their 47 years of existence. They don't work remotely. Everyone is in person all the time.

They notoriously have one gigantic campus on a farm, with buildings designed to look like the Land of Oz, a Wizard's Academy, a treehouse, a barn, a replica of New York's Grand Central Station, and an 11,000-seat auditorium underground. They have the majority of the major hospital systems in the U.S. using their software. Of their more than 600 customers, they have never lost a single one.

Yeah, that is the craziest thing to me about this company. It's 47 years old, and they have never lost a customer.

Actually, we found out that's not totally true. They lost one customer once for 6 months, and then that customer came back 6 months later.

Yes. The company's founder, Judith Faulner, is undoubtedly one of the great founders of our time. You probably don't know much about her or the company because the company is still privately held, and Judy and her family foundation own about half of it, despite its size. I think at this point they're close to $6 billion in revenue and have more than 14,000 employees.

They have a stated goal to never go public and never be acquired. Judith, at age 81, has created a succession plan and a trust structure for her voting shares to ensure that will stay true forever.

Yes, we heard all sorts of stories about companies sniffing around Epic over the years, trying to buy the company—GE, Microsoft, Google, everybody you would imagine wants to buy this company—and it's never going to happen.

Yep.

And we'll dig into this at the end of the episode, when we have all the context and all the numbers, but I believe that Judith Faulkner, in starting one of the most valuable companies in health care, is the most successful female entrepreneur in history.

Almost undoubtedly.

Well, all right then. Spoiler alert, listeners. We'll discuss that at the end.

Yes. So, the health care industry: There is so much wrong with the American health care system. That is an incontrovertible fact. There's nobody who's going to tell you, “Oh, actually, it's pretty good.” It's not pretty good. It's a disaster. There are runaway costs, burdens of administration, and so much excess and waste, causing health care costs to reach, I think, 18% of our GDP.

So rather than trying to eat that whole elephant today and unpack the entire system, today's episode is about understanding Epic's role within it and how Epic became so dominant.

Yep. If you want to understand the system, you have to understand Epic.

David, take us in.

1. Judy Finds Computer Science

All right. Well, we start our journey in August 1943, when Judy—today Faulkner, then Judy Greenfield—is born in Earlton, New Jersey, which is part of Cherry Hill, a suburb of Philadelphia, right across the Delaware River, not too far from where you and I grew up.

That's true. And Taylor Swift.

That's right.

And DuPont. So much great American entrepreneurship in that area.

So much indeed. It is a pretty auspicious time and place to be born because just about 4 years after Judy is born, up the road a little way from Cherry Hill, in Murray Hill, New Jersey, William Shockley and his colleagues would invent the transistor at Bell Labs. That would enable Microsoft, Epic, Intel, and all of this. For a long time, the early pioneers of electronic health records were hardware companies.

Yes. Lockheed, GE, Siemens.

Incredible. But for the moment, Judy probably didn't know anything about this because her family was not in the tech industry growing up. Her father, Lou, was a small-town entrepreneur. He ran a local pharmacy and soda fountain there in Erlton called LSE Lou Soda Fountain. Perhaps this is where Judy would later get her own entrepreneurial bent from.

Could be.

So that's Judy's father. Now, Judy's mother, Dell Greenfield, was an absolute freaking dynamo. She graduated high school at age 15. She worked first as a secretary, then with Lou at the pharmacy and soda fountain, and later got really involved in peace advocacy during and after the Vietnam War, which I assume was not typical for her generation that had lived through World War II.

She ended up becoming the director of the South Jersey Peace Center.

And later in life, after the kids were gone, she and Lou moved to Portland, Oregon, where Dell became the executive director of an organization called Oregon Physicians for Social Responsibility. Get this: In 1985, this group, in partnership with a broader international group called Physicians for the Prevention of Nuclear Wars won the Nobel Peace Prize. Judy Faulner's mom was part of a group that won the Nobel Peace Prize in 1985.

So, David, where did you find this? Because in everything that I read about Epic—and you and I basically read everything you possibly can read about Epic that's out there on the internet—nobody knows this. This is not cited anywhere. So how did you find your way to it?

I was curious. I was trying to learn more about Judy's family growing up. I was trying to learn about the soda fountain and the pharmacy and the impact that her dad had on her future entrepreneurial career. So I started Googling obituaries, and I came across her mom's obituary, where I learned this. The company actually confirmed this to us.

Yes, Judy's mom won a share of the Nobel Peace Prize in 1985. Incredible. So cool.

All right, so you've got entrepreneurial DNA. You've got sort of peace-oriented, divergent-thinking DNA.

Yep. Then you've got Judy. At this point in time, growing up, she's mostly just interested in math. She loves to tell the story about how, when she was in seventh grade, a teacher asked the class a number theory problem—of course, probably didn't frame it as a number theory problem—but asked the class, “Why is it that numbers that are divisible by 3, if you add up all the digits of that number, the sum of its digits is also divisible by 3?”

This is like a law of number theory.

So Judy hears this problem and is just like, “My future is in math.” In 1961, Judy graduates high school and goes to Dickinson College to major in math. While she's there, she gets a summer job one year at the University of Rochester, just a little bit farther north, in their particle physics lab.

For the work that she had to do over the summer, she needed to learn computer programming. This is in the early to mid-’60s, and they said, “You've got to learn Fortran for the work that we're doing here and running these experiments over the summer.” So they give her a book, a manual on Fortran, and Judy teaches herself Fortran in a week and becomes one of the best programmers in the lab.

If you're not getting the picture here, Judy is a genius. She is an incredibly talented person.

Yeah, and programming at a time when programming wasn't a thing. The field of software engineering was not a field. These were math people taking programming languages and using them.

But there are very few of those people in the world. Universities didn't have computer science departments until this point in time. Judy is intertwined with the beginning of computer science as a field.

Makes sense.

So she absolutely falls in love with programming that summer in Rochester. She'd later say that she felt like a kid playing with clay, and that programming a computer was an amazing combination of math—which, of course, she loved—but also language and art together with math. For as mathy as she is, she also has a hyper-creative streak.

Obviously, her story is very different, and she is a very different person than him. But the echoes that you're seeing here in Judy—in what she's exposed to as a kid, how she thinks, how she operates as an entrepreneur—you should be getting some Bill Gates vibes here.

I thought you were going to say Steve Jobs, with the peacetime orientation around her upbringing and the sort of—I don't want to say hippie for Judy, but almost hippie-esque—movement, too.

Yeah, that's actually a good point. Well, Jobs and Gates. The reason I said Gates is to foreshadow some Microsoftian DNA and analogies that come into Epic here. They're contemporaries. She is going to build her company in almost the exact same time frame that Apple and Microsoft were built.

Totally.

So when Judy gets back to Dickinson College, she decides that she's going to apply to graduate schools in math. She applies to 5 PhD programs and, of course, gets into all of them. On her CV, she lists her Rochester experience and her Fortran programming experience.

Two of the schools she applied to, Stanford and the University of Wisconsin, are just starting their computer science departments. When she applied, they didn't have computer science departments, and in the interim, they started them. They saw Judy's application and thought, obviously, this is a brilliant person who we want here at the university. We're starting these CS departments, so they unilaterally shift her applications to their new computer science schools.

Judy's like, “Oh, I didn't even realize I could go get a PhD in computer science. Amazing. This is what I'm going to do.” So she ends up choosing Wisconsin and goes off to Madison, Wisconsin, to start her PhD in computer science, which, as we shall see, she never finishes, but she also never leaves.

What a sliding-door moment. If Judy had gone to Stanford instead, we would probably still be telling this story about her, but it would be a very different story.

For sure. She would have been indoctrinated by a very different type of DNA—what computers are for and what types of companies you should be building with computers—if she was in Silicon Valley at that point.

Totally.

So while she's at Wisconsin, Judy takes a class called Computers in Medicine, taught by a faculty member from the Wisconsin Medical School named Dr. Warner Slack. Probably this is the first course of its kind anywhere in the world. Computer science departments themselves are new, and the idea of applying computer science and computers to the practice of medicine is new.

Yep. It had to be one of the first classes of its kind because mainframes were really becoming a thing in the 1960s. I think ENIAC was only 15 or 20 years before.

Yeah, right. It's a good point. This is the first time anybody could use computers for anything outside of government and the defense industry.

Right. Yep.

So Judy takes this class from Dr. Slack and, as you might expect, she's the star student in the class. Afterwards, Dr. Slack asks her to work on writing a program for use in the hospital and the medical school to optimize the on-call schedules for doctors. Doctors have to have on-call schedules, and optimizing that is the perfect thing that a computer application could do.

Yeah.

So Judy says, “Yes, great.” She starts working on it, and this is a part-time job for her as a grad student. She's getting paid $5 an hour for her programming time—trading time for money here.

Apparently, the story goes that her programming was so good and she was so efficient at writing these applications—because lots of departments in the hospital would ask her to write various applications for them—that she actually didn't make that much money because she just wrote them so fast and she was getting paid by the hour. They gave her a raise at one point; they doubled her salary to $10 an hour, and she still didn't make that much.

Anyway, as she's going around throughout the Wisconsin Medical School, she's working with psychiatry, OB-GYN, rehab, inpatient care in the hospital, and the intensive care unit. She starts to get these requests from all the different doctors and practices.

A big problem we have is that we're seeing these patients, and other departments are also seeing the same patients, but there's no way for us to know what's happening to those patients in other departments across the hospital as they're being seen. We really would love it if there were a single database that could keep records on every patient we have across the whole longitudinal course of their care at the University of Wisconsin Medical Center—an electronic sort of health record for these patients, you might say.

And this is the origin of Epic.

2. The Paper Record Problem

But before we go further on what Judy does next, I think it's worth taking a step back and talking about what health records are, what medical records are, what the state of them was in America here in the mid-1960s, and why this idea of an electronic version of them is so appealing.

So patient health records, at least here in the American medical system, were a ragtag, informal process, and they were all paper-based, of course. I think attempts at unifying them or creating standards dated all the way back to 1912.

Yes, at Mass General.

So all of the efforts to standardize them—and you can imagine all the reasons why standardizing them is important—go back to Massachusetts General Hospital in Boston. Patients see different doctors at different hospitals, and they move. Even within a hospital, like here at the University of Wisconsin, cardiology really wants to know what has happened to this patient in other departments that they've been in before.

Or if you have to see a different doctor for the same thing, it'd be great if they're calling the same condition the same condition.

Exactly. Mass General, or MGH, is the main and largest teaching hospital of Harvard Medical School. It would make sense that this is where a lot of research is happening.

In 1912, a few members of the American Medical Association and the American Hospital Association get together there at Mass General in Boston. They start working on whether they can create some standard entry practices for physicians that they can distribute across the country—something physicians can use as their patient diaries, as they're known back in the day.

In 1919, after 7 years of work, they finally introduce a standardized treatment diary, quote-unquote, for distribution across the country. Now, there's no incentive for doctors to actually use this. Doctors, as we shall see, are a notoriously independent-minded group and profession, and they don't like being told what to do or how to design the notes that they take.

Which makes sense. They've been the smartest people their whole life. They're doing this thing that required an incredible amount of education. It's a very prestigious, high-paying job.

Totally. And I think there's a good argument for this here, too. The practice of medicine, especially then and still now, is equal parts art and science. Telling me how to standardize what is going on with my patient probably seemed anathema at the time. I'm the doctor. I know best. I'll figure out what delivers the best care.

Yep.

Exactly. So this continues for a while. Then finally, in 1928, the Association of Health Record Librarians of North America gets set up as an official body to standardize the collection, storage, and retrieval of patient data.

This is a fun callback. I believe the Rockefeller Foundation was behind funding all this—a callback to our Standard Oil episodes. I believe that because wasn't our conclusion on Standard Oil that the Rockefeller Foundation was the initial funder of the nation's medical schools, or sort of kicked off the movement of having real medical schools?

Yes. One of, if not perhaps the main, goals of the Rockefeller Foundation was to improve the state of medical care in America. It makes total sense that, in the late 1920s and early 1930s, this is getting funded by them.

So all this is great, but the reality is that until this time in the 1960s, when computers start to arrive, even the best intentions—and even if doctors really want to follow all this—as long as you're dealing with paper records, there's a limit to how helpful this can all be. You could standardize it as much as you want, but you still have to get the reams of paper from one physical location in a hospital to another. God forbid you're trying to get them to a different hospital, a different health system, or a different state. It's a mess.

Not to mention, paper doesn't lend itself well to structuring data because, at the end of the day, you can write on paper however you'd like to.

So we finally get to the current moment here in the mid-1960s, and 2 really important things happen. You could argue that they are both of equal importance. One is the arrival of the computer age. You can now digitize this stuff. It's possible to have portability and standardization, and the dream can really be realized here.

Yep.

The other, maybe more important, thing that happens is that in 1965, Congress creates Social Security and the Medicare and state-based Medicaid programs in America.

Ah, yes.

So far, we've given you the technological history of medical records, and now it's time to flip over to the policy side of things, which in medicine you have to understand before you can understand the business.

3. Payers Take Over Healthcare

All business structure in this industry is driven by the architecture of policy in America. So, we rewind the clock again to 1942.

Oh, all right. Educate me here. We're going to catch up to Medicare here.

But in 1942, America is at war. There is something passed called the Stabilization Act that imposed wage and price controls. An interesting thing to note is that, at this point in time, only 10% of Americans have health insurance. Period.

Yeah.

Epic Systems

Mostly, the paradigm in health care is that you pay for service. You go to the doctor, you pay the doctor, right? You don't yet have these runaway costs and, well, I might have a procedure that could randomly cost $1 million. You don't yet have these amazing breakthrough treatments that could be very expensive if you were able to achieve them. So, you mostly just paid out of pocket.

So, you've got these wage controls, but employers still want to be able to attract top talent. So, the way around the wage controls was to offer health insurance as a bonus to work around the system.

Markets always find a way.

Epic Systems

Yeah. Right. So, as soon as this happens, unions go, “Oh, this is awesome. Our people can make more money. Amazing. We're going to make capital pay for our health plans now.”

Exactly. Increase wages for workers. This is what we're all about. They lobbied to make it explicitly legal since it was kind of a gray area from the wage controls. And so, the National War Labor Board makes it official: You can do this.

2. Employers and the health insurance companies that are springing up then lobby to say, “Hey, these health insurance premiums—can we make these tax-deductible for the business?” Because the business shouldn't have to pay taxes on this money if it's going out and buying health plans for its employees with it. And on top of that, even though it's a form of compensation, can we make that not taxable as income? That would be really great. The businesses can deduct it, and the individuals shouldn't have it affect their income taxes, even though it's clearly compensation.

So, this is where the whole origin of pre-tax versus post-tax stuff in your paycheck comes from.

Guest

Exactly.

So, this sets it all in motion with incentives like that. Why would you want your health care any other way? What a dream. You can pay for your premiums with pre-tax dollars. And what a dream for employers that can now offer this benefit that was deductible.

Guest

So, at the end of the war in 1946, the percentage of Americans who had medical insurance was already up to 30% from that 10% number just 6 years before. Flash forward a couple of decades to 1964, and it's now up to 80%.

So, of course, it was going to run the table and become the default as soon as this regulatory framework was set up. And once we did this, the US did not have a chance of implementing any other system. We were just going to end up with employers primarily being responsible for health insurance and creating large insurance companies to provide it.

Yep. And then, of course, that takes us to 1965 and Medicare and Medicaid, because the big policy question is now: What do you do about all the people who don't have jobs? What do you do about people who don't have jobs, which actually aren't that many, but what do you do about people who are old or poor? America needs a social safety net. It's what we're all about.

Guest

Who also, by the way, are the highest consumers of health care.

Yes. So, here we are at 1965.

Guest

David, you're exactly right. Medicare and Medicaid enter the picture as a part of the Social Security Act. How did we get here is sort of interesting. FDR, Truman, and Kennedy had all tried to pass a single-payer system the way that the UK did right after World War II.

The UK is in this moment of sort of great nationalistic pride: We all have to band together and look after one another. They pass the National Health Service. The US does not have the political will to do it, and so we don't. You sort of have this coverage gap, to your point earlier, of all these people who are not currently in the workforce.

So, the compromise is: We will create Medicare for those over 65 and Medicaid if you have a low income or other special situations. And we can't really fund either one exactly right, so we'll sort of figure out how to fund Medicare out of the federal government. And Medicaid, let's make that the state's responsibility. We'll help, but we'll federate that out to the states.

So, all of that is our system today: private insurers if you have a job, buying directly through the ACA if you don't, but again, still from private insurers; Medicare or Medicaid if you sort of fall outside those buckets.

And the important thing to realize from all this is that the vast majority of patients do not feel the cost of their health care directly in the United States. Those costs are so laundered through private insurance companies and Medicare and Medicaid that most people think about any given health encounter as being paid for by someone else, by a part of some system.

Yep. Okay. So, in case it's not glaringly obvious, why is this so important to medical records and the fledgling electronic medical record industry here?

Guest

Well, it's because now, with these second- and third-party payers set up, you need documentation of what happened in order to get paid. So, if you're a hospital or you're a doctor or you're a clinical practice, before this system, you just see your patients and they pay you. Now, after this system, you see your patients and then you need to negotiate with the payer, whether that's the insurance company or the government, about getting paid for that.

And the insurance companies and the government are like, “Well, hey, I need proof. I need documentation of what you did. I'm not going to pay you if you just tell me that you did this stuff, and I need it in a really standardized form.”

Yep. I need an official, standardized medical record.

Guest

And so, all of a sudden, you start this Faustian bargain, or slippery slope, if you will, for the medical profession: “Okay, in order to get paid—and get paid more and more over time as my procedures become more and more complex—I need strictly codified, regulated, and standardized documentation of what I did. And I need really bulletproof workflows and data flows between what's happening in the patient room and then what bills get sent over to the payers.”

And thus, the existential need for an electronic medical record system. But really, also, and maybe more importantly, you need an electronic medical billing system.

Yep.

Guest

And the realization that I think the whole industry kind of had early is that insurance wants to pull together in a very large pool. So, the payers are always going to have a tremendous amount of leverage over individual physicians, small hospital systems, and even larger hospital systems.

So, you really need to be extremely buttoned-up, extremely standardized, and extremely auditable because you're negotiating with this large counterparty, whatever the given payer is—be it a government or a large insurance company—that's going to have leverage over you.

And for the hospitals, in terms of quote-unquote customers—I mean, they're not customers; they're payers—the reason they're called payers is they're not getting the delivery of care. They're just paying for it. Of that set of payers, Medicare and Medicaid are the big gorillas because they're covering the elderly population, who are consuming the vast majority of care and the vast majority of complicated and expensive care in the country.

Makes sense.

Guest

So, here we are in 1966, the year after Medicare and Medicaid get created by the government, and there's this existential reason now for health care to adopt systematized records. And so, a group once again at Massachusetts General Hospital in Boston starts work on the first real computerized medical record system, called COSTAR, or Computer Stored Ambulatory Record.

And it's used for patient scheduling and registration and clinical data and everything you would think in an electronic health record, but also, critically, billing and interfacing here with the payers. Development for that was funded by the National Institutes of Health.

The technical requirements for creating this, though, are actually pretty difficult, especially with programming languages available at the time, like Fortran or whatnot. This is a very high-concurrency transaction system. A lot of users need to use this across a health system, and it needs to interface with a lot of endpoints, especially with the limited storage and processing power of computers, like mainframes, at the time.

The group of programmers that were working on this found that existing programming languages couldn't really suit their needs to build what they needed. So, they ended up writing their own new programming language called the Massachusetts General Hospital Utility Multi-Programming System, or MUMPS, which is amazing—that there's a medically oriented programming language called MUMPS.

Yes. And Epic still uses MUMPS today—or, actually, it uses its modern descendant, Caché—but this is the standard programming language and database system for the industry.

Guest

I don't know about the industry, but for Epic.

Oh, yeah, which I guess now is the industry.

Guest

Yeah. Well, that's what I was going to say. Many of Epic's competitors do not use this, but Epic does. So, it's kind of de facto the standard now.

So, now there are 2 key features of MUMPS, and this is why these programmers had to create their own language. One, the language and the database are integrated. Right there within the programming language is a database structure, and that makes it very computationally efficient for handling all these high-velocity and data-intensive transactions that need to happen—or at least efficient for the programmer, because you're not switching over to write SQL queries in the middle of your program.

Everything is in the same language. If what you're doing is primarily building a wrapper around a database—to use modern parlance, a glorified database that's going to constantly be read from and written to—it's nice for it all just to be one standard language.

Yep. And then the other thing that they designed it for is multiple simultaneous users. Imagine you've got different departments within a hospital updating the same patient record at the same time. You've then got administrators in the hospital also updating that record to know what's been billed out to a payer, what's not, what's been challenged, and so on.

Right. You don't want collisions to create data loss or something like that. These are people's lives we're dealing with.

Exactly. Now, the lead programmer of this group here at MGH—at Mass General—that was creating COSTAR and MUMPS was a recent MIT graduate named Neil Pappalardo. And being the young, enterprising MIT grad that he was, a couple of years later, in 1968, Neil spun out and started a company around this to sell his software, originally called Medical Information Technology, or MIT. Get it? Name's taken. Yay.

But soon he changed the name, perhaps at the request of the real MIT, to MEDITECH. For folks in the industry, that'll sound very familiar, because MEDITECH is still in business today and is the number 3 player in the EMR space behind Epic and Cerner.

Yep. Pretty amazing.

4. Chronicles Creates Epic

So, back to Judy now, here at the University of Wisconsin. She's working with Dr. Slack and the medical department there, doing all these various application projects. And then Dr. Slack moves to Harvard, where he of course meets Neil and starts working with MEDITECH. But he and Judy keep in touch.

Judy's part of this community of programmers and computer scientists building applications for healthcare systems. And like we said earlier, what the people at Wisconsin really want is the same thing that the people at Harvard and MGH really want: this integrated system where they can get longitudinal patient records across the continuum of care and use them to bill the payers.

The holy grail.

The holy grail. So through this connection, Judy learns about MUMPS, the programming language, and starts working on this problem. Legend has it that one day in the mid-1970s, she's sitting in her living room and has an epiphany about how she is going to build a great system—a single database that can do all of this.

And her quote on this is, “The sun was shining. I was disattentive. I was just sitting there, and suddenly it all came to me. Here's how you build it: the integrated system. And I remember running to the kitchen, grabbing a pad of paper, and just writing code, code, code, code, code. And that code became Chronicles.”

She called it Chronicles. It's a chronicle of a patient's care journey. And that code and that database are still at the core of Epic to this very day.

Is it actually the same code?

No. I mean, she'll claim, “None of my code that I wrote is still in production,” et cetera, and I'm sure that is true. But it is true that Epic's core database—the core single database—is Epic's big differentiation. There's only one database that every application pulls from directly and talks to directly, whether it's the clinical side for EMRs, the billing side with the Resolute module that Epic has, Cosmos, Storine or Beans for Kidney, or whatever application Epic makes of its hundreds of applications. It all pulls from one single database in Chronicles. And that's what Judy writes here in the 1970s.

Okay, so it's just a database. Where does it go from there? She writes a bunch of code when she has this insight that there should be a database with the patient model at the center?

Well, that's a good question, because it is just a database. The idea of applications, and certainly graphical applications on top of it, doesn't make sense because we're still in the mainframe world here. So what the product is is this database, and then different departments in the Wisconsin Medical Center can write their own screens, like queries, on top of the database. Those can sit on their terminals, and they can read data out of the database directly into their patient rooms or departments, wherever they're sitting.

So these are terminals. I mean, these are 80-character-wide, text-only terminals.

Yes. Green-screen UNIX terminals.

Okay. So you run Judy's database—you run Chronicles—somewhere in your hospital building on a big mainframe, and then there are these text-only terminals that can query it.

Yep. That's the product.

Sweet. Long way to go.

So, just like the projects in Boston at MGH, where the intention was to make the systems for use there at MGH, Judy at first is just making Chronicles for the Wisconsin medical system. She's just a programmer employed by the medical center.

But the doctors at UW are going to academic conferences all around the country, and they start telling other doctors at other academic institutions about this great system called Chronicles that they have, which this programmer they employ, Judy Faulner, has written. Demand for it starts to spread virally. Wisconsin gets calls, and then Judy gets calls from all these other health systems around the country that are like, “Oh, hey, can you write this for me, too?”

Doesn't she get asked to start a company several times, and she's like, “No, I just made it for this one instance. We don't need to build a whole company around this thing”?

Exactly. And this happens again and again and again. Legend has it that, finally, she just breaks down. She's like, “Fine, I'll start a company. I'll be part-time on it. We'll get some other people who are working with me. We'll be part-time. This will be a small little thing.”

And so, finally, in 1979, the company is born: Human Services Computing.

Amazingly generic. The future Epic Systems.

All right, so David, Judy leaves her job. She starts—what is it?—Human Services Computing.

Human Services Computing really rolls off the tongue.

Yes. She's not planning, up until this point, to be the founder of a company. And yet, for the next 47 years of her life, she would be. How does that go?

Well, before she does anything, she needs to raise some money—not the least of which is to buy a computer to do this work on. Wasn't it something like $70,000 to buy a computer?

Yes, a computer being a Data General Eclipse 16-bit minicomputer, which is the size of, like, a washer-dryer system and sits in the basement.

That's right, because we're in this awkward era between mainframes and microcomputers. The microprocessor hasn't been invented yet, and so we have these quote-unquote minicomputers that are still washing-machine-sized.

Yeah. We're right before the PC era here. It's so interesting how sometimes you have the technology wave that eclipses all prior technology, but sometimes you have these half-steps along the way that turn out not to eclipse everything.

Yeah. Minicomputers—they were like the netbooks of their time, you know.

Yes. We talked about this a lot on our Microsoft episodes, but the critical thing about minicomputers, especially for Judy here and fledgling Epic, was that a consumer wouldn't buy these things, but small businesses could, and small departments of big companies could. An IBM mainframe system was going to cost you a lot of money.

Yeah. A Data General or DEC minicomputer was going to cost you about $70,000, and Judy needs to buy one. She also needs to fund the company.

So she goes to the bank and gets a bank loan for $70,000 to finance buying the Data General Eclipse minicomputer. Then she rounds up a bunch of friends, family, physicians, and other programmers in the University of Wisconsin system who are working with her on these projects to invest money in the new company. Together, they all put in about $70,000. They value the company at $70,000—a pre-money valuation of $70,000—with another $70,000 of new money on top.

Yep. Post-money of $140,000, taking 50% dilution from your $70,000 fundraise, and then you also have this $70,000 loan.

Yep. And those were the only primary investors ever in Epic Systems. There are some really fun stories about what happened to the shares that the other original investors bought over time. The company has bought a lot of them back, but not all of them. There are still some floating out there.

There are some floating around out there.

We can't share all the stories that we heard in the research, but one fun thing is that, at one point in time—I think this was probably in the 2000s—maybe a pretty good chunk of those shares made their way to Sequoia Capital.

Yes. Sequoia did not put that one on their website. It was almost the only way in which Epic intersected with Silicon Valley in the entire company's history. But this is a crazy point: They do $5.7 billion in revenue today. They dominate an industry, at least here in the United States, and in total they raised $70,000 of equity capital and $70,000 of bank debt.

Yes. And that's it. This is IKEA all over again.

Totally. And I think a huge part of what enabled this—building software is not necessarily a capital-light activity. It is really hard, especially in those days, to build great software. I think, similar to how Microsoft never really raised primary capital either, it's because Judy had generational talent as a programmer.

I hadn't made that linkage. That's interesting.

They had kind of a cornered resource as a startup. Microsoft did in Bill Gates and Paul Allen, too, but Epic totally did in Judy. She could write really, really great software. They didn't need to go hire an army of programmers.

No.

And the company stayed very, very small for a very long time. As legend has it, 3 half-time people in the basement of a building started the company, and I think they got pretty far on just that.

I think so. And for years, they didn't hire that many more people. That would obviously change over time.

Hm.

Okay, I'm 90% sure I'm going to stump you on this one. You mentioned the basement of the building. Judy gets the company's first office space in the basement of an apartment building at 2020 University Avenue in Madison, Wisconsin. Do you know what other great American company from Madison, Wisconsin, also started in that same office space? I believe they overlapped. I think they were both in this same shared office space concurrently.

David, there's only so much material out there on this company. We were both going to find this.

Listeners, this is the American Girl Doll Company.

Oh, I really thought I had you on this.

Ben and I don't do our research calls together. We do most of them separately, but we were on one together, and the person we were talking to was about to say this. I was like, “No, don't say it. I want to stump Ben on the episode.”

I did see it somewhere else, though. The crazy thing is, I think when American Girl moved out, Epic bought some of their furniture.

Yes, I think that's right.

So, listeners, what we're talking about is the American Girl Doll Company, founded by Pleasant Rowland in Madison, Wisconsin. My sister had a few of these growing up.

Oh man, Jenny had so many of these, and a lot of them have migrated now to our house and our girls have them. Already, my 3-year-old runs around our house with the American Girl doll magazine, and she tells me on a weekly basis, “These are the ones I want for my birthday.” She's pointing at a $250 doll.

Yeah, incredible company. It actually ended up being acquired by Mattel in the late 1990s. It was a big part of Mattel for a long time.

I'm so bummed I didn't stump you on this one.

Okay, back to the Epic story. Judy's got the financing, she's got the computer, but she still doesn't know anything about building and running a company. She's a software developer.

So Warner Slack says, “Okay, come on out to Boston. I'm going to set up a couple of days for you to spend time with Neil at MEDITECH, and I'm going to ask him to help you out starting this company.”

I've got to say, this is a pivotal moment, and not just for everything she learns from MEDITECH, which I know you're going to get into—all these cool things that she brought over. Instead of, “I'm going to introduce you to a businessperson,” he says, “I'm going to teach you as a programmer everything you need to know about running a business.”

It's not true, but it kind of rounds to true to say Epic basically never did hire any businesspeople. It is essentially a big, gigantic company of programmers, logicians, and implementation people who could be programmers, who think like programmers. That is the DNA of the company to this day. It could have gone in a super different direction: You need to go figure out a sales and marketing strategy and a business plan. That is just not what happened.

This is a history-turns-on-a-knife-point moment for Epic and Judy because the likelihood that she would get introduced to a business mentor who is also a software developer, and not a business-school graduate business guy—we're talking about the 1970s here.

The business playbook is: You bring the grown-ups in, right?

Yep. Bring in the suits, bring in the business-school graduates, fire the founders.

Well, right. Exactly. You wonder why Judy was so averse to venture capital. Everybody else she knew who was taking it in that era was getting fired and swapped in for a business guy.

Totally.

But yeah, Neil came out of MIT. He's a programmer. He was on the original MUMPS team. So Judy goes out, spends 3 days with Neil, and gets a total crash course in setting up and running a company.

And Ben, like you said, Neil ran MEDITECH like a software developer. All the processes were extremely standardized. He had hardened APIs for running the company. He had manuals and documentation for everything, and he shared the manuals with Judy: This is how you set up an HR system; this is how you do benefits and payroll; this is how you do college hiring; this is how you promote internally.

Well, yes, MEDITECH recruited from the universities there in Boston. They didn't hire experienced programmers. They were hiring fresh college graduates. Totally natural that Judy's going to do the same thing at Epic, first out of the University of Wisconsin and then at other schools. But yeah, that was the playbook. That DNA runs right through to this very day at Epic. It's mostly people from nonmedical majors. They're hiring from Midwestern schools, from people with technical majors, just operating under the assumption that you can learn this healthcare thing. It's the same thing as Microsoft, Google, Meta, or what have you. They are going to universities, all departments, and recruiting kids through on-campus recruiting at colleges.

Yep.

And she was doing this at a time when it wasn't the norm. I mean, I've got to keep drilling in how different the world was then in a bunch of different ways, but one of them was that you didn't have these college career fairs where you would just assume you could get this amazing, high-agency job right out of college. Microsoft was really on the frontier of hiring smart college graduates, empowering them, and letting them run free. Epic was doing the same thing, and this was not industry standard.

Microsoft wasn't even really doing it at this point in time. They had only just moved to Washington. They'd only just left Albuquerque. This is the timeframe we're talking about.

Yeah. What year are we in?

We're in 1979. So Microsoft really isn't starting to ramp hiring yet.

No. And to that point, there is no DOS yet. There is no real PC industry. Microsoft's main product is still the BASIC interpreter for PCs and the fledgling personal-computing industry. They have demand from other large academic hospitals and medical systems out there that have mainframes set up, that have university computing infrastructure that they can leverage and use. But your average hospital, medical clinic, or outpatient clinic in America doesn't have a minicomputer. They don't have a mainframe. The market isn't really there for this kind of stuff yet in a big way, except at these university medical systems.

And part of what she is doing is going after a different segment than MEDITECH was doing, a little bit out of deference to Neil and his company. They're kind of going after the small-hospital market. She's going to take the more upmarket, enterprise-y approach and go after the most complex institutions: these academic teaching hospitals, later the IDNs, or integrated delivery networks, which are huge hospital systems, or children's hospitals. Again, the most complex, upmarket enterprises have the most possible needs for the most complex software. That is where she's about to point the company.

Yep, which she has to for the Unix-based product that she's making, because those are the only institutions that have Unix infrastructure and can afford these big computers, right?

Yeah.

So, in 1979, they start out with 4 initial customers. 4 years later, in 1983, is when they renamed the company to Epic Systems. Ben, why did they rename the company?

Well, aside from the first name just being an awful name, you can start to see Judy's quirkiness come through. I think you saw it originally with Chronicles: her sort of abstract, creative thought, the notion of an epic. The Greek epic is this big story, this big longitudinal historic event.

And I think the way that she's thinking about a patient record is that the life of the patient is an epic.

Yep. Does that kind of jibe with your understanding?

Yeah, it definitely was not a declaration of aspirations as a company, because this was still a very small business. They started with 4 initial customers, and in 1983, when they changed the name, they only had 9 customers. In fact, by the end of the company's first decade of existence, at the end of 1988, they only had 24 customers. It was only doing $1.5 million in revenue, and they had a handful of employees.

That's the crazy thing. Unlike Microsoft, this was a small business from the get-go, right?

It took them a decade to get to $1.5 million in revenue.

Yes, that's kind of a slow-growing startup. A great software business for a local entrepreneur in Madison, Wisconsin, I think is how you would describe that.

Yep. So, yeah, for the first 10 years of the company, it's growing, but it's not exactly setting the world on fire. Part of the reason is that the computing infrastructure wasn't there at customers, like we talked about. The other big reason why they weren't getting crazy customer adoption is that they were only doing the clinical medical record stuff at this point in time. They weren't doing the really important stuff of the billing system until 1987.

So for the first, what is that, 8 years of the company, they weren't addressing the actual critical problem in the hospitals, which is, “Help us bill for this stuff.” If you can help them make more money, they're going to be a lot more excited to buy your software and pay a lot more for your software.

Yep. Totally. So in 1987, they launched a billing module called Resolute, again on top of the Chronicles single-core database, which still today is the company's revenue-cycle application for all their hospitals. And I know we've made this point a few times, but I really want to underline it.

It is still just an application built on top of the Chronicles database. And fast-forward to today, I think that is actually the single biggest reason why Epic wins over its competitors. I would argue there's 2 big reasons Epic wins today. One is reliability.

Yep. But the reliability also comes from the fact that it's all built on 1 database. You're not gluing multiple systems together. All their competitors along the way, or almost all of them other than MEDITECH, became 30 other companies glued together through M&A and taken public, taken private, and crazy transactions.

Epic's just been Epic the whole time. So you get this system that, when you buy it and they say it's going to take X dollars and X amount of time to implement, it does. Then you go live and it works, which sounds crazy, but it does the thing that they say it's going to do, on time and on budget, or as well as anyone does in this industry.

And then the second reason, in addition to reliability, David, as you're alluding to, is the fact that the clinical side and the billing side are completely stitched together in 1 codebase, working off of 1 database. It is perfect harmony. You don't have information dropped when 1 system is talking to another.

It's funny, I'm laughing as you were saying earlier that it sort of does what it's supposed to do as an EMR. I assume you're referring to the actual clinician- or doctor-facing EMR side of it.

If you're looking for a system as a hospital that ties what happens in your medical practice to your billing and your revenue events, there is no question. That's all you're trying to do as a hospital. That is what you're looking for. That is the product.

Yes. And Ben, like you said, all the competitors out there, except arguably MEDITECH, in many cases, have a billing system that is a separate and often separately acquired product from the medical record. Not only is it maybe not ideal if some of the information doesn't pass quickly, efficiently, or accurately between those 2 systems, it's incredibly not ideal. You're not getting paid for the work you're doing. Or even worse, maybe you're submitting documentation that is wrong, which is a federal crime, right? Or, even if it's bad in the other direction, then it's not just that your billing system is not making it into clinical; you're potentially causing patient harm and costing lives. Any data flowing in either direction that's bad is really, really bad in this particular use case.

So there's a great Substack called Health API Guy, which I'm going to reference a few times because it's some of the best writing on Epic you'll find. He put it perfectly: “Epic becomes the natural choice for enterprise decision-makers precisely because of its integrated system architecture. Rather than managing multiple vendors and systems, buyers get a comprehensive platform with a single database, unified workflows, and built-in interoperability through Care Everywhere, which we'll talk about later.”

Care Everywhere is the magical button that makes it so that your hospital records at one hospital are easily integrated and viewed in any other hospital that is, of course, also an Epic customer.

Yes. Okay. So after they launched Resolute, and now we're in the late '80s into the early '90s, we're finally entering the PC era. It's now possible for hospitals and health practices to adopt computers for a few thousand dollars instead of a few tens of thousands of dollars. The market really starts to take off for Epic.

And on the back of that, in 1992, they launched EpicCare, which I believe is the first graphical user interface—Windows-based EMR application—in the entire industry. Certainly, Epic definitely says that is true.

Yep. So what does that mean? Go back to what Chronicles was originally. It was terminal access directly into a UNIX-based database on a mainframe or on a minicomputer. This is not something that your average doctor, nurse, or medical assistant is going to use. Now, here in the '90s with EpicCare, they've created a Windows application that any PC user can fire up on their Windows machine and use a graphical interface for their patient interactions in their EMR. And it's tied directly into your billing system for your hospital or your health practice.

We're approaching the holy grail here if you're a hospital administrator, especially with scheduling. Their scheduling thing is called Cadence, but once you have billing, scheduling, and EpicCare handling the actual clinical part of it—and this is all ambulatory, right? We haven't gotten to inpatient yet?

Yes. The initial launch of EpicCare, the GUI EMR application, was only for ambulatory, only for outpatient settings, not overnight stays in the hospital.

As a non-healthcare person, that's always how I define what inpatient is.

Yes. You hear “ambulatory,” and you just hear “non-overnight stays.” Too confusing otherwise.

But this is it: scheduling, billing, and EpicCare on the clinical side for ambulatory. It's a pretty amazing product that is trajectory-changing for the company. It's now got enough functionality that it's not going to stay a small business for long.

So by 1995, they hit $18 million in revenue, up from $1.5 million in 1988. So, pretty phenomenal 7 years there where they more than 10x'ed. By most measures, you'd look at this and be like, “Business is going well.”

Yes. No longer a small Madison, Wisconsin, business.

However, what we were comparing it to before was Microsoft. You made the Bill Gates comparison. This company was started in 1979, so only a couple years after Microsoft—4 years after Microsoft.

Right? Very different trajectories.

In 1995, Epic did $18 million. In 1995, Microsoft had gone public, shipped Windows 95, and did $6 billion in revenue. The primary reason here is that one is a small, vertically focused healthcare company, and one is creating the horizontal platform of the future. But it's worth contextualizing “businesses going well” by making sure that we stop making the Microsoft comparison from here on out.

Well, I think I want to come back to it later in the episode and in the analysis because, yes, Epic is and always will be constrained by being a vertical software provider instead of a horizontal software provider like Microsoft, Google, Oracle, or what have you.

Vertical being 1 industry.

Vertical being 1 industry. However, the 1 industry that they operate in is 18% of American GDP. So how big can this get? That's a question we'll revisit.

Yes. Okay. So then finally, in 2001, they launched the inpatient version of EpicCare. This is for inpatient hospital stays overnight. Now you finally have the holy grail. You've got Chronicles as the 1 single database, you've got EpicCare Ambulatory for all your outpatient clinics, you've got EpicCare Inpatient for all your inpatient activities, and you've got Resolute, the billing system—all on a single database, all tied together, all built on top of it.

By the way, again, inpatient activities are probably going to be your majority revenue stream because that's where the most expensive, most complex care is happening. It's kind of similar to Medicare being your most important payer relationship because old people are where the most complex, most expensive care is happening. So you've got that and then you've got Resolute.

Yeah, if you're a hospital system administrator, this is the best thing you could possibly imagine. And Epic is really starting to get religion around this point in our future: breadth. Our customers—these hospitals—do not want to buy piecemeal solutions. They want to buy everything from 1 vendor, and they want that 1 vendor to provide the very best, most integrated experience possible. So we need to continue to orient the company around that philosophy.

Yep. Now, right around the same time—actually, before EpicCare Inpatient launches—Epic also launches MyChart.

This is crazy. If you had asked me before I did research for this episode, when would I have guessed that MyChart launched? This is an internet-based, consumer-facing medical-records access and interaction platform on the web in a highly regulated, HIPAA-regulated industry. I would have guessed 2010, maybe the mid-2000s at the earliest.

No, Epic launched this in 2000, which is wild—right around the dot-com bubble.

Yeah, crazy. It's absolutely wild that they launched this and that their customers launched this.

Well, and it's incredibly innovative. It is truly cutting-edge. It wasn't 1994 like Amazon, but the fact that, to your point, the medical record was being surfaced on the web—that's pioneering. Consumers had direct access to it. I want to come back in a minute to how important this is for Epic and its customers, but the origin story of MyChart is kind of fun.

It actually started as an outgrowth of what was called EpicWeb in 1997, which was a project they were working on for remote access to the EMR—to EpicCare—for doctors at home. The idea was, you're a doctor, you go home, you wake up in the middle of the night, you're thinking about a case, and you want to check the medical records. Maybe you want to update them.

Some foreshadowing here of how doctors are going to spend their time in the future.

Yeah, exactly. Little did they know how much doctors would hate what we're saying here. But you want to be able to access the medical records of your patients remotely from your home. So they start working on EpicWeb, and a young, right-out-of-college programmer is working on this by the name of Sumit Rama. Sumit today is the president of Epic.

He goes to the then-president of Epic, Carl D'vorak, and says, “This is good that I'm working on this, but I'm kind of bored. Can you give me something hard to do?”

This says a lot about the culture of Epic. This is an early-career programmer going directly to the president, and that says enough on its own.

Yes. The president is running this EpicWeb project, right? He's kind of a computer-architecture person himself. The conversation is, “I want a more interesting project. Can you give me something more cutting-edge to work on?”

Yeah. Give me something more challenging.

And out of that is born the initial idea for MyChart. I think they started working on it in 1998 and launched it in 2000. So it becomes the first integrated patient portal.

I mean, truly, this is a very fair thing to say: This is insanely innovative. For as many reasonable barbs as get thrown at the company—some around the technology, and the UI being clunky and all this stuff—come on. MyChart is really cutting-edge. People immediately got it.

Once you could access your own medical records from home, from your own computer, directly, without talking to anyone, you were never going back. The world changed overnight.

It was a little bit like Zillow in real estate.

Yes. As soon as you can look up how much the homes in your neighborhood and your friends' and neighbors' homes sold for, you're never going to go back to not being able to do that. We're going to have some rich debate later about the pros and cons of EMRs and whether we're better off today than we were. But I just can't fathom being in a world where I don't have a way to access my records other than going to the physical building and asking for them, or placing a call and asking them to call me back.

Records used to have to get faxed back and forth when you moved or changed providers. It was brutal.

But the other really compelling use case for MyChart, especially when it first launched and even through today, is managing family members' care. You have elderly parents who are using the health system, as you do when you get older, and you need to help manage that, but you don't live in the same city. Even if you do, having MyChart access to family members was huge and starts this whole patient-side virality for the company and then for its customers.

I imagine initially it was very scary to roll this out. Once they added self-scheduling to MyChart, this became the greatest thing for hospitals.

The workflow savings from calls that had to happen to schedule appointments are huge. The other big thing is no-shows. Before MyChart, self-scheduling, and the waitlist that MyChart manages for patients, if a patient no-showed, that was lost revenue for you as a hospital system. That was a big hole in your daily revenue operations.

Yep. And now you can automatically fill that in with another patient on the waitlist.

Yep.

So we're foreshadowing this, but everyone adopts this, basically. Today there are 191 million active users of MyChart, and this is deduplicated. These are active, unique users of MyChart today.

Yep.

So on the back of this, they launch MyChart in 2000 and EpicCare Inpatient in 2001. The company crosses 50 million in revenue. They feel like, “Okay, we're finally ready for the big leagues.”

And just to contextualize that 50 million in revenue, they're up to 88 health systems now. They really are starting to penetrate the market, just going one by one by one by one to all these different hospital systems and selling them their software.

Yep. So they're ready for the big leagues. Then in 2003, they get a call not just from the biggest player in the biggest league, Kaiser Permanente in California.

All right. So, David, Kaiser Permanente—the biggest of the big fishes—comes knocking.

Yes. Kaiser Permanente is this fascinating organization that was started and is headquartered here in California. It's a fully integrated, quote-unquote, managed-care consortium. What that means is that, essentially, they are both your health plan—your insurance—and your hospital system, all in one.

It'd be like if an insurance provider married up with a hospital system and said, “This is our captive hospital system. You, as our members, are only going to get your care here, and we control the whole system.”

Yeah. It's the closest thing that you can sort of have to a single payer here in the country.

But importantly, they do also need to work with and interface with Medicare, as they have their own Medicare Advantage plan. For their older patients, once they get older, they transition to Medicare Advantage through Kaiser.

Gotcha.

At the time, in 2003, Kaiser was the largest single hospital system in the entire country: 30 hospitals, 400-plus clinics, 11,000 physicians, and 8.5 million patients who were part of Kaiser in 2003. They decide that they are going to put out an RFP for a whole new, integrated, entire Kaiser-system-wide EMR system.

This is when you really should start to think, “Okay, EMRs are not just medical records, or medical records tied to billing and scheduling. This is the operating system for this industry.”

Someone described it to me as the nervous system for a healthcare system. You've got 20 hospitals, a bunch of doctors, and a bunch of administration people. The whole thing is tied together by this unbelievably complex, tentacles-everywhere piece of software with thousands of different screens and levels of authorization, authentication, roles, and permissions. It is incredibly hairy and is the single nervous system that the entire organization runs on.

Calling it an operating system is taking it too lightly. It's your operating system plus your ERP system plus your applications on top of your operating system. It's your everything.

It's quite reasonable, even though we refer to these things as EMRs, to start thinking about an EMR as a feature for a constituency of the whole system.

Yes. So in 2003, Kaiser puts out an RFP for a new, again quote-unquote, EMR—but really a new nervous system for all of Kaiser.

Yeah. And Epic wins the deal. Epic was a little company just a few years before. They were a 50-million-dollar company.

Yes. After they sign Kaiser, they go to $162 million in annual revenue. That's transformative. They probably more than doubled their revenue overnight.

Yep. Interestingly, let's talk prices for a second. The headline number, as everyone reports it, is a $4 billion deal. They call it a $4 billion, three-year project, and Epic's portion is around $400 million. Of course, it's not all in one year, but the way these things work is that there's a big implementation that costs a bunch of money up front, and then there's the ongoing license that I think eventually would transition to a subscription. At this point, though, it's licensing to use the software.

It's just funny to see these headlines because the $4 billion number not only captures many years of the deal and the implementation, but also rolls in the headcount of the hospital employees who have to do the work. It also rolls in the potential lost productivity from all the doctors across the health system who have to ramp up on the new software, which is a real economic impact, for sure. But it's not like Epic got $4 billion out of this.

No. I always chuckle because every single one of these numbers looks huge: multi-billion-dollar deals. Even if it's a small health system, it's a $300 million project, and it ends up resulting in nowhere near that much money for Epic. But this is how the industry has decided to talk about the size of these deals.

Yeah, it's funny. We should start talking about the size of Acquired and its fourth- and fifth-degree, tertiary impacts. We are a billion-dollar business. Just think about all the business that our customers do with Acquired listeners. But to your point, it is actually fair. If Kaiser is going to engage in switching its central nervous system over several years, it's going to be a net $4 billion of impact to them.

Yes. So, the story of how this goes down is wild. At the time, Kaiser's two main centers of gravity were Northern California and Southern California. They were almost like separate companies under the Kaiser umbrella. They had different systems, their own EMRs, and different management. Of course, they talked to each other and were part of the same parent organization.

Wasn't there almost like a cousin organization in the Northwest?

Yes, like this stepchild. They had—I don't know if it was just the Northwest region or if they had some other smaller regional operations at the time—a small Pacific Northwest region based in Portland, Oregon. Today, Kaiser has large regions through a large part of the country. They've grown a lot since then, but in this small Portland region, they had started using Epic for their ambulatory clinics. Not even inpatient stuff in the hospital, but their outpatient clinics.

At the time, the Northern California and Southern California factions were battling each other, and they were each trying to develop their own proprietary EMR systems with software consultants, with Accenture and stuff. There was this era where hospitals thought that EMRs should be their IP—that they should develop them and have a sort of competitive advantage over other hospitals because their EMR was better. I don't know what the thinking was, but people wanted to own their own EMRs. I think there may even have been some pipe dreams of, “Oh, we're going to commercialize this and sell it to other hospitals.” It seems like that's not a core competency that hospitals should be doing.

Anyway, within the Kaiser system, there was a fairly high degree of rotatability of physicians, of doctors. If you were a doctor in Portland with Kaiser Pacific Northwest and you wanted to move, or your family had to move down to California, you could transfer pretty easily to Northern or Southern California Kaiser. So this was happening, and as physicians from the Northwest started coming down to California, they'd say, “Man, what are you guys doing? You're spending all this money with Accenture and all these consultants and trying to roll your own. We've got this thing called Epic up in Portland that we're not even using at the hospital, and it's way better than the stuff you're trying to build.”

So finally, after a year or two of this battle between North and South, they agree: “All right, ceasefire. It's come to a truce. We're going to ditch our competing projects, and we're going to bid this out to third-party vendors.” They hold an RFP for a new EMR for all of Kaiser—one integrated system—and they pick IBM. IBM is going to come in and do this big $4 billion project for Kaiser. Nobody gets fired for buying IBM.

We're at the end of that era, but it's still a little bit of that era. This is only 2003. This bid might have even happened before 2003.

IBM comes in, and the project fails. It doesn't work. This is not uncommon, especially if you think about ERP. The number of times you've heard, “Oh, failed ERP implementation,” and some CEO explaining on an earnings call, “Yeah, we lost hundreds of millions or billions of dollars, and we actually didn't even switch systems.” This sort of thing happens in the hospital world, too. Epic has bet the whole company on having a reputation for, “We don't have failed implementations,” and that wins deals.

Totally. I think this is one of the most important reasons why their customers love them. We're going to say this a bunch of times on the episode: their customers love them. When we say “customer,” we mean the hospital CEO, CIO—the chief information officer—and CFO. When we refer to their customers, that's who we're talking to. Obviously, the chief medical officer and all of the physicians, nurses, and care folks are a part of that. But a part of my research has revealed that the customer is the hospital administration. I doubt that's even a controversial statement, probably not even at Epic themselves. I think they consider the customer to be the CEO, CIO, and CFO of their customers.

Yep. So the IBM project fails, and maybe it's worth another word on that, too. It's not like this doesn't impact the physicians and the staff of the hospitals. Poor performance for the hospital, as in the case of this failed IBM implementation, is a massive detriment to the physicians. This must have really sucked for them. They want stability, too.

The physicians provide the value at hospitals, and therefore you need to make the physicians happy to retain your administration job. They have the leverage in the organization because they provide the core competency.

Still, you're not going in and pitching the doctors when you're trying to land a customer. You could think of a hospital almost like a media company. The doctors, nurses, and clinicians are the on-air talent. The on-air talent are not the ones at Disney or Universal or wherever who are making the business decisions, right? In this case, the medical staff is involved, but they're not the decision-maker.

Yeah. So, okay, Kaiser's now got to rebid this project. By this point, enough Portland doctors had come down to California and sung Epic's praises that they're saying, “All right, we should take this little company seriously.” They start the RFP, and it comes down to Epic and their big main rival, Cerner.

Now, we haven't really talked about Cerner so far in this episode. It's part of Oracle today, as we'll get into later. Cerner was a much bigger company.

Yep. Interestingly, it started right around the same time, right?

Wow. I think it was started the same year as Epic, right? 1979.

Yep, by Neal Patterson in Kansas City, Missouri.

Unlike Epic, which took the no-venture-capital, no-acquisitions, single-platform, stay-private-forever route, Cerner took the lots-of-acquisitions, raise-capital, go-public, get-big route. That could have worked, too. It did work for a long time.

At this point in time, Cerner is almost a $1 billion-a-year revenue business. They're a public company, way bigger, and they're international. The RFP comes down to the two of them.

But it's worth saying that before Cerner was acquired by Oracle, it was a merger of 24 different companies. Put a nice little wrapper on it and call it Cerner, but there were a lot of companies along the way.

Yes, a different path than the one Judy took, let's say.

But that's not to knock it. It was, and especially back then, a good, really competitive product. The obvious choice for Kaiser here in 2003 would have been to go with Cerner, and they actually tried to go with Cerner.

So supposedly, at one point they come to both companies—to Cerner and Epic—and say, “Look, Cerner, we want to go with you for inpatient, in the hospital, since that's your bread and butter. This is the most important thing. This is the big business. We trust you. You've been around forever. And Epic, hey, you've got inpatient now, but you only just launched that in 2001. You're new at this. You're good at ambulatory, at outpatient. You're doing a really good job for us in Portland. We want to split the baby here and do one system with Epic for our outpatient clinics and one system with Cerner for our inpatient systems.”

Oh, yeah. That always works.

And Judy says, “No, that is a bad choice. That is the wrong choice to make. I don't care if you go with us or them, but to do the right thing for your patients and for your whole system, to make billing work, to have this all function correctly, and to have patient records transfer between your ambulatory clinics and your hospitals—which you really, really need—you should just pick one of us.”

Now, this was a little self-serving on Judy's part because Cerner was not good at ambulatory. It's a calculated, high-risk decision.

However, it's definitely super ballsy to do this. But she knew that Epic had a good product in both. The problem was just that their inpatient product was still new and so didn't have trust yet in the marketplace.

Yeah, but she's still a pipsqueak. Cerner's a $1 billion-revenue business, and Epic is somewhere slightly north of $50 million at this point in time.

So they go a little further in the process, and at one point there’s a technical due diligence meeting where Kaiser asks both companies to come in and present to them about how their systems are going to handle all the volume of concurrent data transactions that Kaiser has. Remember, 8.5 million patients and 11,000 physicians is a high-volume system here.

Yes.

And listeners, this is a great story. This is kind of the 11th hour of the deal. This is, “Hey, we’re pretty close to a decision, but we haven’t made it yet. Can you both come to the same building and sit in different conference rooms? Throughout the day, we’re going to bounce back and forth, spend an hour with each of you, formulate some questions from hearing the other pitch, and then come back and ask you those questions.”

So if you’re on one team or the other, you can kind of learn through the questions what is being pitched in the other room and why you’re suddenly being grilled on this new topic.

The Epic team is doing their planning, getting ready for all the preparation for this really big meeting. The team had decided that the way they were going to handle this question was to do a theoretical presentation about how Epic’s architecture worked—the single system—and theoretically how much load the system could handle all at once.

And the story is that Carl Dvorak, the president, flew into California the night before the pitch, met with the team, saw this plan, and was like, “Guys, no. We need to model out in Excel exactly what Kaiser’s transaction flow is going to be throughout the day in this system and how our system will process it, and prove to them that we have far-in-excess bandwidth capability to handle their system and that it’ll never go down.”

Because I think they had done it at sort of a theoretical formula level, but Carl knew that Epic had the advantage here: They actually could, if they played it all the way out and really built out the spreadsheet, show that, actually, we’re going to be more performant for you.

Yep.

So he and the team pull an all-nighter the night before the presentation. I don’t know if Judy was there as well. I assume she was, too. They come in, and they show the model during these meetings that Kaiser is having, going back and forth between the 2 teams.

As the day is going on, it becomes really clear that Cerner has not done a similar level of modeling and can’t actually prove to Kaiser that their system is going to be able to handle the transaction flow. I think this was the moment when the tide turned, when Epic was like, “Oh, yeah, we’re going to win this thing.”

But it still wasn’t obvious right away. I think they earned big points there, but my understanding is that Kaiser still went to Cerner and said, “We’re interested. Can we do an equity deal on top of this? Can we take part of the company in exchange for basically giving you this big deal?”

I think Cerner said yes.

Yeah.

The story as we heard it, at least, is that at the last minute, right before the decision, Kaiser came to both companies and said, “Hey, we’d really like warrants in your company. We’re the biggest health system in America. This is the biggest contract you’re going to get. We want some equity in your companies for working with us.”

Apparently, Cerner did offer them 10% of the company for this deal. That’s how important this was.

So then they come to Epic and say, “Well, what do you have to say about that?” And Judy’s like, “No, we’re not going to do it. We’re not going to do it for you. We’re not going to do it for anybody. And it’s the wrong thing to do. If we did it for you, we’d have to do it for all our big customers. And then that would turn out poorly for you, too. So absolutely not.”

And they still picked Epic in the end. Through that, through the architecture bake-off, and through the “No, you have to pick one of us,” Epic still won out.

Yeah, and they stuck to their guns all the way through that negotiation, which I think really should just tell you how important the stability and continuity of the system across inpatient, outpatient, and billing is, because Epic is the only one that can offer that.

Yeah. By the way, earlier when I said 24 different companies merged together to create Cerner, I forgot that this chart that I’m looking at predates when Cerner then bought Siemens. So then there are another 12 companies that had merged together to become Siemens that merged also into Cerner.

Yep.

We’re going to come back to the Siemens acquisition in a minute. I think that was 2014 when that happened. But, yeah, this is key. This is why Epic wins.

Yep.

So they win the deal. It basically doubles or triples revenue overnight. The Los Angeles Times writes about the deal when it gets announced:

“Because of its scope, the Kaiser Epic system could become the Model T of its industry. Not the first of its kind, but the first to reach masses of people.”

Once this happens, Epic gets elevated to the new gold standard. If you’re a hospital system, if you’re a CIO or a CEO looking to rebid your EMR, well, Kaiser, the biggest system in the world, just chose Epic, and they chose it over all of these reasons not to. There must be something really good in there. Of course you’re now going to consider Epic, and of course Epic is going to perform really well in these evaluations.

Yeah, it helped a little bit at first, but after the go-live, a few years in, after the whole implementation took place and they didn’t tip over and it did go well, that was really when the floodgates opened, sort of in that 2006–2008 time frame.

Yep. So by 2007, Epic has hit 500 million in revenue. So another 3× what they were doing once they added Kaiser, and about 8× what they were before Kaiser—almost 10× what they were before Kaiser. They’re really starting to transform into a big company.

I spoke with one former employee, and I was asking what the inflection points in the company were. This employee said, “Oh, after we won the Kaiser deal, we had been hiring 10 kids a month out of college, and it now felt like hundreds a month were just flowing in through the doors so we could scale.”

That will bring us in a minute to Epic’s epic, shall we say, Verona, Wisconsin, campus.

Yeah.

And one thing, just for our storytelling narrative, that we kind of skipped over was that there’s another thing that happened in the early 2000s that reads a little bit as a sort of alternate history for what could have happened at Epic.

Imagine you’re a small company and a big company comes to you and says, “Can we co-develop a new product together? You can distribute it to your customers; we’ll distribute it to our customers, under our brand name and everything, but you get some of that revenue, and we’ll sort of build this with you, and it really will charge your business up.”

And if you don’t know any better and you’ve never done it, it kind of sounds appealing. Maybe we should do that. And there are many instances of it working, so that’s extra tempting.

My favorite weird example in the history of this not really working is the HP iPod. Do you remember this?

Oh, yeah. That’s right. I do remember this. I think I might have had one.

No, I have one of the U2 iPods, the red and black one. Listeners, Google it. It’s this really odd thing. It’s an HP brand on the back of an iPod.

That tells you all you need to know about what a weak position Apple was in at the time, that they were willing to let HP put their brand on something entirely created by Apple to get HP’s distribution and get some cut of that revenue. And, of course, it would have bootstrapped their ecosystem since it used iTunes and all that.

But that’s basically what happened here. Philips, the Dutch company, comes to Epic and says, “We want to do something focused on the radiology segment. You do a lot of the development work. We have the customer relationships and distribution.”

Right. Because they’re probably selling the machines.

Yes. Philips will get a license to Epic’s whole IT system. Everything that you already sell, we want to sell also. And we’re going to market that as Philips’ entire enterprise to our customers. They’re going to buy a Philips-branded version of Epic.

Oh, wow. And I imagine Philips probably has a lot of customers not in America, too, like in Europe. So this is a way to go international, right?

So Epic starts hiring people in the Netherlands. They build up this team. They spend multiple years, or at least a year, building it out. I think they even launch it. The whole thing ends up folding, and within a year or 2 of coming out, it was a really expensive detour.

The company develops this intense scar tissue for partnerships. Partnerships in this era to the Epic team means: A, stuff I can’t control outside my organization; B, risk; C, never in our history do we have an example of it working. This is an uncontrollable dependency.

And they sort of internalize this scar tissue as: Stay focused on what we can control. Go directly to the customer. Don’t try to do any fancy partnership integration stuff with other people.

And that’s oversimplifying it, but I think it’s still reasonable to say that some of the DNA of what Epic would become—and they would take tons of arrows for this, this closed, non-interoperable, blah-blah-blah—some of it stems from this failed partnership with Philips.

Yep. Totally. They’re now partnering again, finally, but, yeah, like 15 or 20 years later.

Yes. And thanks to Health API Guy for the tip on that.

5. The Verona Culture Machine

So, at this point in the late 2000s, Epic has its eyes on the prize. We’re building everything ourselves. We can get big customers on our own. We can stand firm and not negotiate, not have to give up pieces of our company. The price is the price, and we know that we’re going to deliver.

And so it’s time to invest in our future. How do they invest in the future? The Verona campus.

So for the probably minority of you listening who know anything about Epic as a company or have been involved with them in the past, if you have, you almost certainly know about their corporate campus.

Yes. So, there are sort of 2 stories of how the campus came to be. First, Judy's son, by this point in time, is actually working at Microsoft in Redmond as a developer. Judy and Carl had always been inspired by Microsoft's way of doing things. So one time, when Judy's visiting her son, she's like, "Hey, can you go give me a tour of the campus in Redmond? We're thinking about expanding. We're outgrowing our space. I want to see what it's like there." And she's really impressed by Redmond, as I think anybody who goes there would be.

Yep. Especially in this era, it was sort of pre-Googleplex. It was kind of the fairy-tale tech campus. It was Google before Google.

Yep. And she's like, "This is like a college campus here." You've got sports fields and tons of buildings, and all of them are pretty short—2, 3, 4 stories—with a lot of outdoor space, walking space, and everyone gets an office. This is perfect. So Judy comes back and basically copies the campus strategy of Microsoft almost exactly.

They had been in that renovated schoolhouse.

If you think about where they came from and what they now have the opportunity to build, they had renovated the schoolhouse to be nice, but—

Yeah. That was well before they were a half-a-billion-dollar-a-year revenue company.

Yes, exactly. So she goes out and buys 1,000 acres of farmland in Verona, Wisconsin, which, just like Redmond, is a kind of bucolic-looking suburb about half an hour outside the city in Madison.

Or I guess maybe Redmond was bucolic before Microsoft got it built up. But just like Microsoft kind of owns Redmond, Epic owns Verona spiritually.

So they build this incredible campus there. What is very different about it than Microsoft is that it is not utilitarian the way that the Microsoft buildings were.

A thing that was already happening at Epic was this fairy-tale, whimsical thing that you can kind of see in their product names. But David, what was the story we heard about the schoolhouse fireplace?

Okay. So this is the second story of the Epic campus. Like you said, Ben, before Verona, the headquarters was in an old school building in Madison that they bought and renovated. As they were renovating it, the designers decided that in one of the main conference rooms, they were going to put a fireplace in there to make it feel more homey.

Yes. Like a Wisconsin lodge.

And I don't even know if the designers did that, but I think Judy and the company were like, "Well, this kind of feels like a Wisconsin lakes lodge. Let's lean into it." So they decorated it like a lodge. They brought snowshoes and furs and an axe that they put on the wall, and it ends up becoming the most popular room in the building.

And anytime customers are coming to Madison, they always want to meet in the lodge conference room.

And so now they're building this new campus, and Judy's like, "Oh, well, let's take the lodge idea and really blow it out. Maximize it."

Yeah.

So they inject this fairy-tale-ness times 10. When they're building out the new campus, they hire the same firm that did the Disneyland California Adventure renovation in 2008.

I think it's even more on the nose than that.

They hired 2 architecture firms. One is that one.

Uh-huh.

And then they also hired the firm that built a lot of the Redmond campus for Microsoft.

Oh, is it really?

Yeah. Wow. And it's crazy. I mean, it's Alice in Wonderland stuff. It's Harry Potter-inspired stuff. It's Wizard of Oz-inspired stuff. Just Google pictures of the Verona campus for Epic. We'll link to some in the show notes, too. It's bananas.

But it is extremely attractive to new hires coming out of college who want to feel like they're still in college and want to go work for a company that seems fun and interesting.

And Epic is this sort of 2-sided culture that seems to play well together somehow: this goofy, whimsical fairy-tale thing and this hard-driving, win-at-all-costs, performance-oriented, fierce competitor.

That goofy, whimsical fairy-tale thing and this hard-driving, win-at-all-costs, performance-oriented, fierce competitor—it just is both. To understand the company, you have to hold in your head that the DNA is both of those things concurrently. And I think it's because that's what Judy is.

Yes, 100%. That is completely spot-on.

I had in my script here the question of all this: Why own the campus? And that is exactly why. We are hiring super-smart, young, hungry new college grads. How are we going to attract them to Verona, Wisconsin? Well, we are going to create a paradise for them.

Yes. So, some interesting stats: It's 1,700 acres. 410 of them are the campus; the rest is the farm. It now covers 89 buildings. There are 4 indoor auditoriums with 18,000 seats total. The big one, Deep Space, is the world's largest subsurface auditorium. There are 11,400 seats.

This is an auditorium. This is 2 Radio City Music Halls full of people smashed together in 1 giant auditorium underground on a corporate campus. The number of seats is much closer to a Chase Center. It's much closer to a basketball arena than it is to any other auditorium that I can think of. It goes down 74 feet beneath the surface.

I mean, the whole logic behind it is, when they built it, they thought, "Oh, we'll never grow to 11,400 people. We can have our all-hands in here. No-brainer. But this can also be the place where we have all of our customers and our whole ecosystem can come here." Of course, now they don't actually fit in there because they've outgrown it. Not even all their employees can come to all-hands there. The logic for building it was that they had a movie theater that they used to do their monthly all-hands meetings in, and they wanted to have this opportunity to do it on their new campus, too. It's wild. When you look at it, you just can't believe the scale of this building.

Yep. All right. I think now is the right time, while we're talking about the campus and Epic culture, to really talk about Epic culture.

Yeah. Because the first thing that you have to understand is Judy refers to it as a software factory. When you keep looking at it and asking, "Why is it so weird?" the biggest takeaway is that, in Judy's mind, since they don't ever go buy any other companies and they don't have any competencies at the company other than making software, what they are is a factory that turns out software. They take in software developers and turn that into software for the medical industry.

That's funny. Apparently, she was just ahead of her time with the AI factories. Was it Nvidia and Dell are calling them AI factories now?

Yeah. Absolutely. And so it starts to click and make more sense when you think, "What would a factory for turning developers into medical code and medical applications look like?" Well, Verona, Wisconsin.

Yep. So, while we're on culture here, one of the most amazing things about the campus is that Epic has a list of 10 commandments—the Epic Ten Commandments, like Moses in the Bible. They have them posted in every bathroom and in every break room across the entire campus. Any visitor who goes there—and it's open to the public—can just go in and see the Ten Commandments in the bathrooms there.

The Ten Commandments are: Number 1, do not go public. Number 2, do not acquire or be acquired. By the way, those first 2 things you don't need to communicate to employees. Only the CEO can do either of those things. So it's sort of funny to put them as commandments. That shows how deeply Judy feels they need to run through all the employees.

And I think the other big motivation for doing this and having it there in the bathrooms is that every customer who comes to visit sees it right there, too. We will never go public. We will never be acquired, and we will never acquire another company. You can trust that this is 1 system forever.

Yep. So, that's 1 and 2. Number 3: Software must work. Number 4: Reality equals expectations. Number 5: Keep commitments, even the unspoken ones. Number 6: Focus on competency. Do not tolerate mediocrity. Number 7: Have standards. Be fair to all. Number 8: Have courage. What you put up with is what you stand for. Number 9: Teach philosophy and culture. And number 10: Be frugal. Do not take on debt for operations.

Zero of those pertain to healthcare.

Yes. When I was looking at them, I kept thinking, "Oh, I'm going to find something here about every life being important or the patient being at the center of everything." No, this is how to run a company. This is my opinion on how to run a company, period.

Yep. And I think, specifically to your point, this is a pretty good way to run a software factory.

So, these hang in the bathrooms. Other interesting things about the campus: They have wedding bells that will play campus-wide when a new client is signed, just showing that that's the type of commitment that this is. It's, "We've now married this client for the rest of our lives."

Yeah. It's like the wedding march, you know? That's their version of ringing the gong.

And I think we haven't really talked that much about what it is like to be an employee there. So this is probably a good time to do that.

This is an insanely awesome training ground if you are a smart, ambitious person out of college.

People accuse them of being cult-like, but there are ways in which that's a good thing.

I mean, they take you fresh out of college and teach you everything. And when I say everything, I mean how to take notes. There is an Epic way to take notes on a yellow legal pad. There is an Epic way to write emails.

And these are hardened practices over the years that they just believe, through iteration, testing, and data—probably going all the way back to Neil and MEDITECH and, yeah, the 3 Days in Boston—this is the best way to do this, period.

And so we're just going to teach everybody the best way to do everything, and everybody is going to be reasonably robotic.

We can sort of trust that once we squeeze you through the Epic system, when you come out the other side, you are able to operate in a way that works really, really, really well in our machine, where people can really trust each other.

So because of that high level of trust, there are very few middle managers. You understand the system that everybody else works within, and you don’t have to corral chaos. Most of the people they’re hiring are right out of school, so they’ve been there for a long time.

They don’t have budgets, which you can only do when you have a high-trust environment like this. There are some financial controls, of course, but Judy has this great story that she used to go see customers and they’d say, “Oh, this is the right thing to do. I just don’t have the budget for it this year, so we’re going to push it to next year.” She’d say, “That’s stupid.”

Or they would tell her, “Can we squeeze this in this year? Because if I don’t spend this money, I’m going to lose it in my budget.” And she would say, “Well, that’s also stupid.” So at my company, we’re not going to have budgets like that.

They’re super light on titles. You might have a business card, whether you’ve been there 20 years or 6 months, that says “Implementation,” which you hand to a customer when you go and do an engagement.

Everyone does immersion trips where the software developers—I mean, everyone—is required to spend time in clinical settings, like operating rooms, to directly observe workflows.

Yep. I think when you start, you have to do 5 of them, and then you do more every year. Wow. Y Combinator preaches, “Go talk to your customers. Spend time with customers.” Epic has been doing this forever.

This is the Epic way that I think the rest of the world sort of woke up to and startups internalized as doctrine. Every person in the company spending time in medical settings, talking to customers, is hugely valuable.

On the developer side, there’s a super-prescriptive software methodology that they use to minimize bugs. You go through this intensive training for 6 months when you join, and then when you start programming, the whole system is designed around minimizing the number of hours between when a line of code is written and when it’s tested.

If a bug is found, you drop everything as the original developer and fix it, so you still have the whole context fresh in your head. You don’t go months and then have the system get tested, allowing bugs to compound into bigger problems this way. They get caught right away.

And I believe the rule is that every developer must fix their own bugs.

That’s my understanding, too. It’s this method of software engineering that places way more importance on a zero-bug environment because lives are on the line than other methods. So you’re not necessarily going to ship software the fastest way. You may not even ship the most innovative, clever, amazing, cutting-edge software. You’re just going to make sure that you’re shipping bug-free software.

Yeah. Well, I think the reasons are twofold. First, it absolutely is true that lives are on the line. If an order for the amount or dose in a prescription gets messed up because of a bug, a lot of people are going to die.

Also, the complexity required for the revenue cycle and billing for your customers is of paramount importance. You cannot have bugs there, either. At best, the hospital is going to lose a lot of revenue opportunity. At worst, they’re going to get sued for federal crimes, for medical fraud.

Yep, great point. So that means you need a highly robust system.

Yes. The work done as an implementation person at the company is insane. It’s like military-level logistics. You’re handling multiple customers, all of which are among the most complex systems on Earth.

Peter Drucker famously referred to hospitals as the most complex form of human organization that we have ever attempted to manage. You have to understand all these dependencies at the customer and the status of a dozen interrelated things on a daily basis.

You really are in this high-adrenaline, high-stakes leadership role as a really young person. You can work 10 or 12 hours a day, but a lot of them love it because you’re winning. You’re doing really big things right out of school. You’re doing it with other really bright people.

So they really try to get high-IQ, high-EQ, often very sweet Midwestern kids to take these customer-facing roles.

Yeah. And I think also because of the flat organization, you’re doing it alongside other senior people and learning from them directly. The story about Carl and Sumit: Sumit is a young programmer working on a project with Carl, the president, who’s also a programmer leading the team. That happens.

If you’re an ambitious, career-focused person, there is nothing more fun than winning in a high-stakes environment with other high performers as a team. That plus the whimsy encapsulates the Epic culture pretty well.

The result of that is it becomes the number one thing in your life. If you talk to a lot of these people who spent time there or still work there, they’re all in, and they’re in the middle of nowhere. This is the other job the campus does: You’re not really getting exposed to other things you could leave and go do. They make it very easy for your whole life to become Epic.

Yeah. You could drive to the big city of Madison.

Yes. I know we keep making the comparison, and it’s a direct one with the campus here, too, but the company that this reminds me the most of is those early days of Microsoft. This is exactly what being at Microsoft in the 1980s and 1990s was like when we talked to people doing that research.

I heard from some people, too, two or three times when I was researching for this. Palantir came up as one of these companies with bright-eyed, bushy-tailed, smart young people where you’re deployed into these really intense environments, but you know your stuff. You’ve been through the training. You’ve been through the process. You’re armed with good tools, and you’re going to go make it happen.

Yep. So the other side of this is that it is up or out. They aggressively trim whatever percentage of performers, and they work you really, really hard. With the vast number of new hires, they’re trying to figure out if you’re going to cut it. It’s more cost-effective to replace you than to keep you as dead weight.

So you have lots of attrition in the first few years, but you know that once someone’s been there for a while, they’re good. You can count on them.

Yep. Their method of hiring is crazy. This is more common now: giving out programming tests as part of the hiring process.

Oh, this story is so good. Can I tell it?

Yes, please.

We mentioned that Judy’s son was a programmer at Microsoft, and that’s how she went to visit him, which was part of the inspiration for the campus. His involvement in the company actually predates him becoming a programmer at Microsoft or even an adult.

In the late 1980s, Judy was hiring software developers and found that interviewing them just wasn’t that predictive of whether they were going to be great software developers. Meanwhile, her son, I think, was in seventh grade or something like that and was doing these programming competitions around the state.

She was talking to him and said, “Well, you do these programming competitions, and they’re these tests. You’re winning some of them, and that’s a pretty good judge of whether you’re a good programmer or not. They’re reasonably predictive. Do you think you could write one of these types of tests for me? I can use it at Epic to test software developers as we’re hiring them and see if they’re any good?”

For 18 years, listeners, that was how they tested to determine whether someone should work at Epic. Many times—I don’t know whether the number is close to 100%, but some large percentage of the time—they just don’t interview you. They believe that their tests are predictive enough that that’s it. You can get a job offer. You come visit campus and all that, but you get a job offer after scoring high on this test.

It’s not just that test now. There are other tests. There’s a Rembrandt test. They’ve got a bunch of tests. It’s all systematized and cataloged, and everybody takes one on the way in for every role in the company.

Yeah. The culinary team, we heard, takes not full software developer tests, but logic tests on the way in.

Yeah. Amazing. But Judy’s teenage son wrote the first test, which is no longer being used. I think the answers have gotten out on the internet, but for a long time it was used. It’s a funny story, but this was in the 1980s. This was way before Google was doing this kind of stuff. These were really out-there hiring practices at the time.

Yep. So we’ve talked a lot about the internal culture. The most important thing that we haven’t really dove into yet is how it touches the outside world. The Epic culture is completely customer-obsessed, and I mean that the way we talked about “customer” before.

There’s a great quote from Jeff Gorkin, the chief information officer for Rush University System for Health. He says, “You get what you pay for 100% of the time,” despite Epic being “not cheap.”

You see that echoed over and over and over again in these customer conversations: It’s reliable. It worked. They didn’t overpromise on something. It’s fully integrated. At this point, no one gets fired for buying Epic, the same way it used to be true about IBM.

Yep. Customers are always number one. They vote on everything. The way that they pick the next things to build is that when all the customers come to campus for their annual conference, they ask for ideas and then they vote. They take customer input as a way of figuring out what they’re going to do next.

Totally. There are basically only 3 roles at the company. There are software developers, project managers, who are the implementation managers doing the new, active implementations for new customers, and then there are technical specialists who do ongoing customer support.

There’s not a sales or marketing department. There are 8 or so salespeople, quote-unquote, in the company who only react to inbound requests, and they all came from either project management or technical specialists.

That’s the whole company. The technical specialists are the biggest group in the company. Every single customer of the 607 or whatever hospital system customers that Epic has has its own technical specialist teams for every single product that they use.

So, if you’re a hospital system, you have your own technical specialist team for your EpicCare EMR, for MyChart, for Resolute, for Cosmos—for you name it. Anything you use, you have your own dedicated team for that. On top of that, every customer has its own dedicated BFF, or “best friend forever”: a single person within Epic whose sole job is to make sure that you, as a customer, are successful with its products.

They do things like grade you as a customer every year, benchmarked relative to what other customers are doing and how you’re doing with the Epic tools. They’ll send separate report cards every year to your CEO, CIO, and CFO, and they’ll grade you from 1 to 5 on a bunch of dimensions. Then they’ll show you benchmark data against other customers in your peer set of relatively similar-size hospital systems, so you can see how you’re doing.

Wow, that’s crazy.

Yes, I’ve never heard of any other company that does stuff like this.

Well, they have a lot of leverage in the customer relationship. I think at this point in history—in 2025—when the customer wants to do something a certain way and Epic wants to do something a certain way, ultimately, Epic is customer-focused. They’ll do whatever the customer wants, but they’re going to lay out very compelling, convincing arguments for why their way is the correct way.

What this leads to is things like a standard package. A lot of times, when people are setting up Cerner, every implementation looks completely different. Epic is highly opinionated: please use as much standard stuff as you can so that we can easily push out updates, easily add new modules for you, and make sure interoperability all works exactly the way that we’re thinking it should.

They have a strong negotiating position with customers when they’re saying, “I think you should do it this way.” At some point, they may choose to just say, “You know what? I don’t think you’re ready to be a customer yet. I think we’re going to focus elsewhere this year because we’re going to pick up only 10 to 20, maybe 30, new customers. We’re happy to wait until you’re ready to work with us.” They actually have the leverage to pull that off.

Now, to your point about the standard implementations, I believe this is the only way that you can get discounts on pricing with Epic: by doing either a fully or mostly standard implementation. I think there probably are some tiers based on how standard your implementation is, and the more you deviate from it, the more you have to pay.

I could see that. I know if you stay up to date on things like database maintenance and versioning and all that, then they give you discounts.

Yep. So, all of this sounds like Epic accruing power for Epic, which it is. There is also this almost altruistic part of the company and its customer relationships, and I think this comes from Judy Faulkner and who she is as a person. “Altruistic capitalism” does feel like ultra-competitive, ultra-value-maximizing, altruistic capitalism. That’s kind of how I would describe the company.

Yeah, I think that’s how you sum up Epic.

And this altruistic piece is that, despite certainly being expensive, they basically never raise prices once you’re a customer.

They do, but their average yearly increase is about 2% across the board, so below inflation. Compare that to lots of other software companies out there, and they’re not raising prices 2% every year.

They also do things like never having, in their entire history, changed the price of MyChart. We got told this multiple times.

I think that’s a red herring. We don’t know what the bottom-line price looks like. We got told this piecemeal. It’s not like we have a full contract in hand showing what it looks like to be a customer and adding up all the subcomponents.

Good point. Two points to make: Yes, they’re customer-obsessed. Yes, they’re listening. But yes, of course, it’s to do what’s in Epic’s long-term interest.

I mean, the core functionality, if you’re a customer, the reason you pick Epic is to turn an interaction with a patient into as many dollars as possible for the health system, without risking downside, with the lowest risk possible. That is why the customer is picking them.

And what is Epic trying to do? Epic is trying to win deals, stay in forever, and achieve world domination. So, they do these very interesting things that feel very customer-focused and are, but are also very valuable for Epic.

Some new piece of software comes out. It’s 2020, and suddenly telehealth is really important. “Here’s a HIPAA-compliant way to do Zoom. By the way, we’re going to start working on our own telehealth thing. It’ll be out soon. You should feel free to use Zoom right now. By the way, when our new module comes out, it’s just going to be free to you. You’re already a customer. It’s just going to be free to you.”

You’re certainly not going to go around shopping for some new thing when you know that it’s just going to come free to you. You’re not going to take it right away because it’s going to be too bare-bones, but at some point, it’ll get good enough where you can say, “Oh, yeah. I’ll just adopt the Epic version now. It’s part of my enterprise-wide agreement.”

It’s this kind of amazing bundling strategy that certainly reminds me of the Microsoft episode.

Yep, you’re totally right. I think that’s the right way to look at it. They’re customer-obsessed because, in the long run, that is the right thing to do for Epic also.

There is a Jeff Bezos quote, and I just thought of it. I don’t have it in my notes, but it’s something about how, in the long run, there is no difference between what the customer wants and what Amazon wants. I distinctly remember him saying this in 1999 or something, in some really old video.

Epic is absolutely customer-obsessed—again, with the CEO and CIO of hospital systems—because if you deliver for them, you deliver for Epic in the long run.

Well, I mean, that’s the goal of running a company—the theoretical underpinning of capitalism, right?

Right. So, all that to say, it is a very fascinating corporate culture and organization.

Yes. All right, listeners, if you know anything about this industry, you know that we haven’t gotten to the important part yet. We’ve gotten to the Kaiser deal—that was a big deal—and we’ve gotten to the move to Verona. That’s an important part for understanding the culture, and it’s this cutesy thing that most journalists who write about the company latch onto: “Oh, cool campus. I should go take some pictures and write a cool story about campus.”

There’s a whole big, crazy thing that happens as a part of the Great Recession and some legislation that gets passed. It’s transformative for the industry and causes a whole bunch of good and a whole bunch of bad to happen.

Okay, so we’re coming through the 2000s here. We just got through the Verona campus. You know everything about Epic’s culture now, or at least everything that we could discern from the outside, anyway. We’re in this pretty interesting era in 2006. We’re in the Bush administration here in America, and Bush says in a State of the Union:

“For all Americans, we must confront the rising cost of care, strengthen the doctor-patient relationship, and help people afford the insurance coverage they need.”

“We will make wider use of electronic records and other health information technology to help control costs and reduce dangerous medical errors.” Wide applause.

This is one of the craziest things to me in doing the research. The narrative out there—and certainly, insofar as I knew or paid attention to any of this—was that Obamacare and the Obama administration were the ones who really pushed EMR adoption and meaningful use and all this stuff that we're really going to get into. Totally bipartisan, totally started in the Bush administration.

Yeah.

And just to give you a little look into how much the window changes in terms of which side represents which party, the next paragraph, George W. Bush also says, “We will do more to make this coverage portable so workers can switch jobs without having to worry about their health insurance.”

I mean, it's crazy. If you're a Republican during the Obama era talking about having coverage that's portable across employers, that's a scary thing to be talking about just 4 years later.

Yep. So funny.

So you're starting to get these political winds of, “Hey, everyone wants a good system for electronic medical records. We think, in the abstract, there will be a lot of good that comes from it.”

Well, I think there's an even deeper motivation than that. In 2006 already, I think everybody knows in America the current health care system sucks. Yes, there are great things about it, but overall, this is cost disease run rampant: a huge portion of GDP, massively inefficient. We want to try and fix it. How to fix it, we don't all agree on. How to fix it, we don't know.

But this promise of digitizing it and incentivizing EMR adoption is held forth as a promise that can deliver us from this problem, right? It feels like a step in the right direction.

At the time, only 13% of health care facilities in America had an EHR system at all. This was a massively undigitized industry in the mid-2000s, which I think there's some pushback on. People would say, “Well, the definition of EHR just changed, so actually there were a lot more than 13% that had it.” But that's the best data that we have.

Is that fair enough?

Okay. Now, before we get to the actual legislation and what happened, there's one other concept to have in your mind, which is interoperability, because this word comes up every time Epic comes up, or any EHR comes up, and it's worth knowing the different buckets.

There's the first and easiest interoperability: Epic to Epic at a different hospital. I want to transfer my records from one hospital to another, and they also have Epic. It's the same technology, so that should transfer easily.

Two is Epic to another EMR that's at a different hospital: Epic to Cerner, Epic to MEDITECH, Epic to Allscripts, Epic to some homegrown system. You can imagine the technical reasons why that would be harder. They have different architectures. All the other ones, arguably, except MEDITECH, don't have a single database, et cetera. And standards have not been quite as standard in this industry as they are in SaaS software, in easier B2B software worlds.

Before we get to the third category, let's just think about some of the incentives that are probably at play here. Epic transferring to itself? Fine, easy. A hospital transferring to a different hospital—you can imagine these places are competitors. They're businesses. They may not want to support that unless it's really in the patient's interest, but they're at least going to be sensitive to it.

Epic to another EMR at another hospital? Well, not only might the hospitals not want that, maybe Epic doesn't love that either, right? Interests are sort of aligned against this happening. Epic doesn't want that, and the hospital doesn't want that. They talk about a lot of good reasons why, “Hey, actually, we do want that,” but let's just call a spade a spade and say there's no business reason why Epic would love for that to happen other than, of course, it's in the patient's interest. Everybody really should do what's in the patient's interest because we all get care from multiple places.

Then there's this third category: Epic to a third-party application that wants to use data from Epic, sit on top of Epic, or interoperate somehow with Epic. As a company, they have been very, very careful about this one over the years, and for good reason. You don't want patient data to leak. Epic has publicly referenced the fact that they have never had a Cambridge Analytica situation because they're very, very careful about sharing data with other application developers.

But the result of this is that it's been much harder to integrate with Epic as an application developer than you would be used to in any other software category.

Yep. Again, there are very justifiable reasons why that should be the case. We're dealing with HIPAA data here.

But it's also massively in Epic's interest for that to be the case. It's very convenient as a strategy.

Yes.

So, as for that first scenario, Epic to Epic, it's actually awesome. They have this thing called Care Everywhere. Today, there are 20 million patient records exchanged daily. I used it as a part of prepping for this episode to join all my MyChart accounts across the 3 health systems I have in Seattle. It just seems to work pretty well.

There's a great story of how it actually happened and how it came to be. Judy made the call personally that Care Everywhere wasn't a thing—or shouldn't be a thing—where their customers could pick and choose. If you were enabling it, it worked across all Epic customers, even if they were your local competitor.

One of the first customers to accept that new software that came with Care Everywhere unknowingly agreed to the interoperability feature. He later admitted that he would have declined it if he realized that it was included in the contract he was signing.

This was the CEO of that health system. He later admitted that.

Yeah, of that hospital. And Judy described it as pure luck that Epic was actually able to move ahead. After that win, they then made it mandatory for all of their customers, and they retrofitted old versions of the software to support Care Everywhere.

So every Epic customer in the U.S. has it and can share with every other Epic customer in the U.S. This is kind of their talking point: “Hey, actually, we have lots of interoperability. We launched this thing called Care Everywhere. It shares millions and millions of records all the time.”

They talk about how they are the biggest sharer of medical record data of anybody, right?

So, coming into 2008, that's the historical baggage that you should know about in Epic land: they believe they're doing a lot of interoperability in the way that they like to do it. Almost no one is great at interoperability. I saw some transcripts of conversations with hospital administrators who were like, “Interoperability in this industry is just laughable, period. No one's incentivized to share with each other unless they absolutely have to.”

And it's not easy. These are incredibly complicated systems, and there's lots of risk. One CIO put it to me in an interesting way when I was chatting about this with him. Because it's patient data and health data, there's this sense that interoperability should be a thing. You should be able to share your data everywhere, and, sure, of course you should. I don't want to discount that whatsoever.

However, he was like, “Imagine this were a different industry. Imagine this were the airline industry. Does United share their customer data with Delta even though it's customer-friendly to do so? Even though it's the customer's data? No, of course not. They never would. So do I feel great about sharing my patients' data with my competitor down the street? No, of course not.”

This is where you really start to feel the friction. I'm sure you're listening to this and emotionally getting charged up at this point because you're feeling your inner capitalist think, “Well, it's a business. They should do the things that make sense for them as a business.” And you're feeling your inner human being, who is part of a society and has your own health needs, thinking, “Well, maybe—but this thing shouldn't be subject to the bad parts of capitalism. Maybe this thing should function differently.” It feels yucky to me that business interests are governing the way that things get done.

You should feel that tension because this is one of those bothies.

Well, that kind of describes the whole health care industry, right? Make no mistake, this is definitely a for-profit industry.

Yes.

And it's highly regulated, and there are all these issues.

Yep.

So we're into the meat. In 2008, the Great Financial Crisis happens. At some point, we should just do a whole episode on the Great Financial Crisis and the collapse and the government bailouts, and how we managed to figure out how not to collapse as a nation and globally. We pulled out of it, and it was amazing. It took some time, but we pulled out of it.

But one element of pulling out of it, just like during COVID, is the government said, “All right, we need stimulus, baby. Stuff. We need stimulus. We need programs that we can throw money at the economy.” Interest rates go to zero. Money out all over the place. Turn on the printers. Figure out how to get people doing productive stuff and reward them for it.

There was the monetary policy stuff of interest rates to zero, but there was also the fiscal policy stuff of, “We need to helicopter money into the economy.” So the question is, when you're going to helicopter money into the economy, you could mail everyone checks. That happened during COVID.

Yep.

Not an ideal way to do it. A more productive thing to do, which calls all the way back to the Civilian Conservation Corps in the FDR, sort of New Deal era, is that we want to look for shovel-ready projects. What are things that we just kind of all agree are good ideas for the country, that should happen? Let's just pass some legislation that rewards the crap out of people with money—with free money—for doing the things that we think are a good idea anyway.

The hope is that you get this double whammy. You get the money distributed, which is just a goal. You're trying to create jobs, trying to create economic motion, and you get something accomplished that is widely agreed upon to be a good idea.

Oh my gosh, a shovel-ready project right here: electronic medical records. Let's do it. Let's just figure out how to finally make that a thing. So the goals are: 1) fiscal stimulus to stimulate the economy after the Great Recession; 2) promote the adoption of electronic health records; 3) promote their use, which is different—the actual use, the meaningful use, of electronic health records. Not only do we want people to become users of your system who never log in, we want them logging in all the time and using the stuff; and 4) promote the adoption of interoperable standards so that providers could share patient data nationwide. These are the goals of the HITECH Act, the 2009 Health Information Technology for Economic and Clinical Health Act.

I love how they name these things.

I know. The HITECH Act gets folded into the almost $1 trillion American Recovery and Reinvestment Act.

So the mechanics in 2009, when the bill is passed, are that there is $27 billion available in direct incentive payments to hospitals that implemented electronic medical records. If you include broader incentives for other health IT projects, data exchanges, and training, the total actually comes to $36 billion.

This is amazing. Imagine if you're a software startup and the government just starts mailing checks to your customers—not only for signing up, becoming a customer, and paying you money, but then paying them even more for actually using your software. It's like the best thing that could happen if you're a software vendor.

It's the government going to your customers and saying, “Hey, I'm going to give you money, but it's like a dedicated-use credit card. The only thing you can use this money for, and you must use this money for, is to buy this software startup's products. And then you must also use them. If you don't buy it and you don't use it, not only are you not going to get the money, I'm going to penalize you, and you're going to have to pay me money.”

Yes. Carrot and stick. So specifically, it amounted to $44,000 to $64,000 per physician in incentive payments over a few years for adopting it. That actually went to the hospital, but it was basically: You get paid for how many physicians are doing this.

Yeah. So imagine a huge system like Kaiser with 11,000 physicians. It's a lot of money.

Yes.

HITECH contained the phrase “meaningful use,” which is this crazy phrase where, if you say “meaningful use,” or you even say “MU” to anyone, it is a triggering term in this industry. First, meaningful use was a carrot and then a stick. So after the stimulus ran out, David, as you were saying, after about 5 years, health systems would face meaningful financial penalties for not meaningfully using health IT software like EHRs.

There's a KFF article entitled “Death by a Thousand Clicks” that has a quote on this. “The EHR vendor community, then a scrappy $2 billion industry, griped at the litany of requirements, but stood to gain so much from the government's $36 billion injection that it jumped in line.”

As Rusty Frantz, CEO of EHR vendor NextGen Healthcare, put it—and NextGen is one of the top 10 Epic competitors—the industry was like, “I've got this check dangling in front of me, and I have to check these boxes to get there. And so, yeah, I'm going to do that.”

For everyone that wants to think this was the most amazing thing ever for Epic, yes, it was obviously net very good. They had to do a bunch of hoop-jumping to make sure that it was exactly the thing that the bill was going to reward. Everybody in this industry had to make sure that they spent a bunch of development time and reprioritization in the company to make sure that it was exactly the thing that HITECH was saying that it needed to be.

Yep. It's more nuanced than “this is really good for Epic.” Of course, it was really good for Epic, but really, this was good for the entire EMR industry. The competitive dynamics within the industry are almost like a separate kind of question.

Well, okay. So who's it going to benefit the most?

Probably the most reliable one. If I wasn't going to do this before, and now I just have a check dangling in front of me as a hospital to implement an EHR, I want the one with the integration that's 100% going to work. I'm taking no risk. Also, I think someone told me before that Epic was an expensive one.

Well, I guess it doesn't matter that it's expensive anymore because it's free. So I'm just going to buy the good one.

Yeah, right. It's effectively free.

Epic, by being the high-price, high-value vendor and the one that you could count on working with the lowest risk, was going to win in this situation.

Yeah, that's a great point. It's like you're getting a stimulus to buy a handbag, and you could get the Target one or you could get the Birkin bag, and they're all free. It's not exactly like that, but it rounds to that.

It also is interesting because it basically rewards the most reliable software, not necessarily the most innovative software. You end up not taking any risk in situations like this because you're not going to get rewarded unless you actually stand the system up and then start getting meaningful use. You end up with the one that's just going to work.

So what did HITECH actually do? We'll go through each of the goals. Did it accelerate usage?

Absolutely. Market penetration of electronic health records went from 9% of hospitals in 2009 to 95% by 2014.

That's insane. David, what other markets have you ever heard of that saw usage go from almost nobody to almost everybody in 5 years?

Certain categories of software during the pandemic.

Yeah, that's a good point. I would imagine maybe Zoom during the pandemic is the closest thing. I mean, to be an Epic employee at this moment, where suddenly your category is just free for all your potential customers, it's crazy.

Digitization was great for patients. Even though a lot of doctors will complain about it, it's totally amazing to be able to message your doctor about something and pull up your own records electronically. As we talked about with MyChart, you can schedule your own appointments. You have all of the above. You can manage your family's care.

Yes, absolutely.

Okay, so then the next goal: cost reduction. There are stats both ways on whether EHRs generally reduce costs. We found one stat that claimed that costs in hospitals went down by 10% with the introduction of EHRs. There's also an argument that the adoption of EHRs may actually have accelerated the ordering of unnecessary care, as well as increased billing codes—either more codes or higher-dollar codes for the same procedures. So that's the counterargument to whether it decreased costs in the system.

The critics of EHRs say that basically more stuff is getting ordered from exam rooms for the exact same procedures than was happening before. And frankly, that's the goal of a hospital buying an EHR. Part of it is revenue maximization, right? Again, these are commercial entities. Even though many of them may be nonprofit organizations or part of a university system or what have you, at the end of the day, this is revenue being generated by a hospital with a management team and with its primary staff being doctors and nurses who like to make money.

The reason you go to medical school to be a doctor or nurse is to get a good job and make a good living, and for the prestige of it. You want to help people, but it's a high-paying job.

Yep.

There's a great read called “An Epic Dystopia,” and I will say this article presents a one-sided view of things. This is the American Prospect article.

But there's a quote. One doctor, for example, stated that her supervisor regularly contacted her and said something like, “That appointment was a 2. Don't you think it might be a 3?” Obviously, in this case, a 3 could bill more to insurance than a 2.

Someone else gave me a quote: “Hey, if you did X, almost certainly you did Y in that appointment. If you use those codes, you could charge more.” If you have a system—a computer system—versus not having one, it's more likely that you are billing for more stuff. That would be the counterargument to this mass implementation of EHRs saving costs for the system.

Yep.

Data interoperability: not really. The legislation was prescriptive about meaningful use. It really wasn't prescriptive about the data standards that people talked about. That's a goal of the legislation, but it's not like there were incentives for data interoperability the way there were for meaningful use. People follow incentives.

The data interoperability did not really happen, which was very convenient for Epic, which had a whole system that it made itself and didn't really need to integrate with anyone to make the software suite useful. So again, as the leading player in the industry and the one that could kind of do everything itself, Epic didn't need interoperability, and it wasn't explicitly rewarded. Customers didn't prioritize it either.

There is a fact pattern that opens Epic up to criticism here, which is that Judy Faulkner was on Obama's health IT council through all of this, advising the Obama administration on the HITECH Act. Other competitors had people on other committees, too. It's not like she was the only one with a voice in government.

Right. I was going to say that is a convenient fact that gets left out by a lot of people making that argument, or the fact that all the competitors—Cerner, Allscripts, et cetera—were all on those councils, too.

Yep. But let's just look at the data interoperability thing. It didn't happen, but it's happening more now, 15 years later. At least we have the foundation of digital records, so that as pushes and shoves happen to make it more interoperable—

Without operability, you can't have interoperability. And I think for me, that's kind of my takeaway of the HITECH Act and meaningful use here, at least when it comes to the industry of EHRs.

Yeah. Anytime you've got government coming in and distorting a market, you're going to get weird stuff happening, especially anytime you have government coming in and regulating how products are to be developed and used. In a market, you're going to get really messy stuff happening.

Okay, so there were 2 really bad side effects here. One, by literally defining what is meaningful use, Congress not only accidentally designed the software and specified the features, but they also accidentally told doctors how they needed to do their jobs. And so doctors are running around now clicking 9 things that they never needed to click before, some of which are poor implementation by their hospital system, and some of which are because the software is really complex. But a lot of it is just making sure that this is what the legislation requires.

Yes, it complies with meaningful use so that the hospital doesn't then come and get hit with a financial penalty.

Yeah. This is sort of the most unfortunate aspect of meaningful use. I think—I think it's the second most. I think the biggest one is that they increased the regulatory burden of practicing medicine, period.

So now, since hospitals can face big financial penalties, there are all these mandatory fields and workflows everywhere, forever, even after the stimulus runs out. And so it just generally increased the operational overhead of the whole industry. The thing to optimize for now is hitting the MU definitions legally rather than the spirit, which is to help doctors and patients get more out of the system.

And you end up with things like health systems merging with other health systems because the cost of doing business is higher. That's a preexisting trend, but it's certainly an accelerant. There's more overhead, more burdensome regulation, just more crap in the system, more waste.

So all of that is unquestionably bad. At the same time, what's the alternative? Are we going to be in the paper-based system? That's not good.

At least we have adoption here. A lot of people have chalked this up to: It's definitely better to be digital. Is this the best form of digitization that could have happened? No. Is digitization overall good? Yes.

And I think if you zoom way, way, way out, you give the most credit to the government and the legislators who created all this. Was the goal actually a jobs program and an economic stimulus for the country? If that was the goal, A+. A lot of jobs got created because of this. 18% of our GDP's worth of jobs are in this field.

So it's now a huge part of our society: people who work in medicine, for better or for worse, and related fields to medicine, like IT administrators and Epic administrators at hospitals. These are a lot of jobs, and these are a lot of good jobs. And if you play it one step forward, these are all domestic American jobs, not dependent on imports or manufacturing, that have created viable, great career paths for a lot of people in this country from a policy standpoint. Maybe not bad.

And I hate it. David, I'm not saying I love the idea of jobs programs. I'm just saying, if you zoom way out and ask what the goal of the government was in doing this, did they succeed? If that was the goal, they succeeded.

Yeah, it was good to create fiscal stimulus for that period of time when we really needed it. Should the government be in the business, on a durable basis, of using taxpayer funds to prop up industries and create and maintain a bunch of jobs in those industries? No, absolutely not.

And so it's a very low bar, what you're saying. Did it stimulate the economy? Mailing checks would have stimulated the economy. I look at this as, yes, it stimulated the economy, and it had the free stapled-on coupon of getting some amount of electronic medical records. Even though they're not interoperable, and even though it's not the best system we possibly could hope for as a country, it's a better one than we would have had otherwise. And it kind of came for free with the stimulus checks.

Yep. So, back to the Epic story here.

Yeah. So, what did meaningful use do for them?

Well, interestingly, if you just take a step back and say, what is the byproduct of the government paying off a whole industry to adopt something new? What you're basically doing is pulling forward the future and saying, this was all going to happen eventually. We want to make it all happen right now.

And to be clear, I think especially after the Kaiser win, Epic was going to become the dominant player anyway, absent meaningful use. They were starting to run the table on competition.

So that's the question: They were starting to.

But what you definitely do when you throw $36 billion at the problem is say, “We're going to close the door on anyone that might become a dominant player after this.” The current competitive set is now what we're dealing with, and we think the current competitive set is good enough that this is shovel-ready enough that we want the best ones to get implemented everywhere.

These are stuck in there for multiple decades.

Yeah, the switching cost here is enormous.

Yes. And that's a fair trade-off. I don't even think it was that intentional of a trade-off that they made by passing this legislation, but it is definitely true that a new EHR that may have been more theoretically innovative in some way, that could have come after HITECH, wasn't going to happen, at least for a long time.

Yep, because you pulled forward so many RFP processes to a single point in time.

Yes, and the majority of them chose Epic, which they probably would have anyway. But a lot of those processes would have happened in the future, at which point other competitors may have emerged, right? There's no greenfield bidding for those theoretical future innovative competitors. Everyone's bidding against your current system.

I buy that, but that's kind of nitpicky, right? It's a little bit of, “For whom are we holding the door open?”

So if you're looking for a smoking gun on regulatory capture, I think the right way to characterize this is as regulatory tailwind rather than regulatory capture. I don't think the folks at Epic loved it either. My impression is they felt like they were going to take this market, and now they had to do it in sort of this Frankenstein-style way.

Well, it definitely made the product worse.

Yeah, it made the product worse. It increased burdens for everyone. It made it more sure that they were going to win the market, because if something's going to happen 5 years from now or it could happen now, you'd rather it happen now and be certain. But they certainly don't love it that doctors have to check 57 boxes on every patient visit, right?

So all this compliance burden and the software being, frankly, hard to use once it's legislated.

Yeah, product development by legislation, which, to be clear, isn't just Epic; it's all of them. And also just the complexity of software of this magnitude, with this many different screens and participants and all that.

A big criticism is that it takes time away from patients. There was a 2016 study that showed that entering data into EHRs consumes about 2 hours of doctor time for every 1 hour spent providing hands-on patient care. There's all this other research around doctors now working 11–12 hours a day. They're constantly responding to messages, which again is a hospital configuration thing. There's a way to triage messages.

Another downside to the HITECH Act and meaningful use is that before digitization, and before moving all this activity from hospitals into the EHR, there were a lot of regulations and guidelines that maybe weren't always followed, and that was sort of for the best in the hospital.

So things like: It's supposed to be doctors themselves who put in orders for pharmacy, for meds, or for scans or stuff like that. But in the old world, before everything moved into the EHR, doctors were supposed to do it, but they would have their medical assistants do it. If you're seeing 20 patients a day and a whole bunch of them need scans or meds or whatever, you can just be like, “Hey, medical assistant, do all this.”

Yeah, I'll sign them all later.

Exactly. But now, once everything's digitized and in the EHR, that means that the rules have to be followed because they're tracked, right? And so that means that all of a sudden, doctors don't have as much help and slack in the system as they used to.

Yeah, it's interesting. So this is the sort of thing that ends up contributing to doctor burnout.

Yep. But you do have to compare it to how it used to be. There's a great old study from 1970 that found that communications activities, such as managing physical records, accounted for 35–39% of total hospital operating costs. That's not the doctor's time, but especially in the paper world, it was always a huge amount of cost and time from the hospital, right? You're shifting around the burden of this a little bit and creating more burden through all the meaningful use regulations, but this is not a new problem.

You know who else doesn't love it? In 2017, Obama himself told Vox that he felt that the HITECH legislation did not live up to what he wanted. And here's his quote:

“The fact that there are still just mountains of paperwork, and the doctors still have to input stuff, and the nurses are spending all their time on this administrative work. We put a big slug of money into trying to encourage everyone to digitize, to catch up with the rest of the world, and that's been harder than we expected.”

The quote: I actually don't think it was at all about catching up with the rest of the world on digitization. I think we were a leader there. But if the hope was that we could somehow downgrade our costs to be in line with the rest of the world, that would have been great.

And that did not happen.

Yep.

One of the CIOs I talked to had a great quote on this. He said that meaningful use and the HITECH Act wildly succeeded at digitization of the industry. It did absolutely nothing on digital transformation of the industry.

Hmm.

We digitized, but we didn't transform in the way that I think people were optimistically hoping for.

That's interesting.

And the good news is that can still come now that all the records are digital.

Yes. Now that we've digitized.

Yes. You actually can do interesting things with the data that you couldn't have done before. And we'll talk about, when we get to the end of the story in a couple of minutes, the promise—at least right now—of AI and ambient AI in making a lot of this onerous process just disappear for humans.

Yes.

6. Epic Outlasts The Competition

So, on the back of all of this, Epic continues to win epically. In 2011, they hit $1 billion in revenue. They keep winning all the big systems that come up for RFP: Johns Hopkins, Cedars-Sinai, UCSF—the list goes on and on and on.

For Epic's competitors, merging and consolidation had always been happening. But now, with meaningful use in place, consolidation really picks up, which again is just going to accrue to Epic's advantage. Consolidation in this space means the products get more complex, they don't work as well together as a suite, and Epic's advantage just becomes all the more pronounced.

In 2008, Allscripts merges with Misys. In 2010, that new Allscripts entity then merges again with Eclipsys. They keep acquiring more smaller players. Today, that company is called Veradigm. In 2011, MEDITECH acquires LSS Data Systems.

Then, in August 2014, one of the big ones that we've already alluded to happens: Cerner buys Siemens Health Services for $1.3 billion, which starts the downward spiral for Cerner.

That's my understanding.

Yes. I heard in the research that even today, in 2025, customers will still refer to themselves as either Siemens customers or Cerner customers. It's still not fully integrated. Meanwhile, I think post-Oracle acquisition, Oracle's trying to rewrite it all anyway, right?

And that's not going well, which we'll come to later.

Yep. So, this leads us right into the Department of Defense contract—the mother lode of all opportunities if you are an EMR vendor.

Or if you are an American taxpayer, the mother lode of everything that's wrong with governmental waste in America.

Okay, David, the DoD debacle.

The 2015 DoD debacle.

So, the Department of Defense decides that they're going to bid out a global EHR contract for all of their hospitals and health care across all branches of the military. At the beginning, this is just active-duty U.S. military, separate from the VA system, which comes later—the Veterans Administration, covering all the veterans of the U.S. military.

This deal in 2015 is, up until that moment in time, the biggest health care IT contract of all time. It ends up being a $4.3 billion deal, even bigger than the Kaiser deal. It remains the biggest until 2 years later, when the VA contract does go out. That is a $10 billion deal.

Now, Epic, of course, bids on both of these contracts. As you would expect, just like with Kaiser, it comes down to Epic and Cerner—but not directly. Just to add to the insanity of this system, it's basically impossible, unless you bid on government contracts all the time, to bid on a government contract directly. So, you need to partner with one of the companies that subcontract off of someone who makes their bread and butter by contracting with the U.S. government.

Yes.

So, they each choose their partners, and there's all sorts of drama around this. Basically, the knives come out on all sides. As we shall see with Cerner, if you win one or both of these contracts, this will keep your company going for years and years and years.

Drama aside, Cerner ends up winning both of these contracts—the DoD and the VA—as a subcontractor. Sadly, in what will perhaps not shock you at all, both of these projects, despite being huge at the bidding phase, go massively over time and budget. It's really, really bad.

The DoD system, the active-duty system, only just fully went live late last year, in 2024. This contract got bid out in 2015, so that is a 9-year implementation process. The VA system is way worse. It is nowhere near fully live today.

The VA system was bid out in 2017. The latest announcement from the VA about this project is that it will be live at all VA sites by—this is a direct quote—“as early as 2031.”

What?

So even the most optimistic scenario that they're saying, which seems very unlikely, is that this is a 14-year project. The latest I had seen is that it's still not live. They actually gave an “as early as 2031.” Now, it is live at a few sites. It's not like nothing has happened over the last decade, but, oh my God.

I mean, if as an American citizen or watcher of America you were disgusted by the whole meaningful-use thing, if you're an American taxpayer, this is just beyond disgusting.

So, what took so long, or what made it so expensive?

I don't really know. I think it's a combination of a lot of things.

Or I guess, what are the incentives?

It's really unfortunate, but there's an old saying that you make more money on failed government contracts than successful ones. I think that's probably the incentive that's applying here.

Oh, because as long as the contract doesn't end and you're still implementing, you can keep finding more costs. You can keep getting paid.

Now, I think it would be unfair to blame Cerner for this. Maybe some of the blame lives with them, but they are a subcontractor to big government primes that are the general contractor here. We talked about this on the Lockheed episode. This is just how the military-industrial complex works now.

You've got the compounding layers of bureaucracy of the government, the military, and EMRs generally. As we've talked about, this is just a rat's nest of awfulness and bureaucracy. You've got these now-multidecade projects, with many billions of dollars of taxpayer money going into them, and they're still not live. Just freaking brutal.

Do you think Epic is glad they didn't win?

Well, here's the thing. Epic, of course, was part of this process, bid on the RFPs, and I'm sure would have loved to have won them. Even the $10 billion initial price tag, before the overruns, is twice Epic's revenue today. That is transformative for your business in the best case.

Yes, kind of like Kaiser was transformative for Epic back in the day.

Yeah, that's it. Talking to Epic customers and CIOs in the research, they are down on their hands and knees thankful that Epic did not win this deal, because Cerner just got dragged so far into the process, so far into the muck here.

And there were a lot of other compounding factors happening here. Cerner, as we've talked about already, was now, post-Siemens—what do you say, 36 companies that had been acquired to get pulled together to make Cerner? Neil Patterson, the founder, CEO, and longtime leader, got cancer right around this time, and then he passed away in 2017.

So, your founder and leader is passing away in the midst of this very complex process. After that, Cerner cycles through a whole bunch of different leaders over the next few years.

Meanwhile, Epic was unburdened by this DoD and VA shitshow, for lack of a better word. They just kept winning deal after deal among the large system providers. In its own way, this is a testament to how Judy thinks a company should be run: “We’re going to stay extremely focused. We have a very particular playbook. We listen to customers. We meet our promises. We’re hardcore about our software engineering and our product architecture.”

When customers find new use cases, Epic develops that software, too. She runs a company in an unconventional way, and that unconventional way is proving to be the right way to win this market.

Yep. So after the DoD deal, Epic wins Partners HealthCare in Boston, which is the new entity of Harvard and Mass General—MGH—and the original big research institution. They win all of the Mayo Clinic, all of its hospitals. That’s one of the biggest contracts, again, outside the DoD craziness. They win Cambridge in the U.K. They win Intermountain Health. Recently, they won CommonSpirit Health. The list goes on and on and on.

In 2018, Epic hits $2.7 billion in revenue. They announce that they have all 20 of the top academic hospitals in the U.S. News & World Report rankings, which is critical for them because winning the top academic hospitals means that all the new doctors and nurses coming through, getting minted by these institutions and trained, are all trained on Epic.

I think they sell them a separate educational license, too. I think they treat the university hospital and the academic classroom as separate. I mean, I’m sure it helps to be in the hospital, but I think you also buy it as a school.

I’m sure that’s right. But also, a huge part of medical education is practicing in the hospital.

And I think now 90% of med school students train on Epic.

Yes, that is the current statistic.

In 2019, Epic hits $3.2 billion in revenue. And then finally, in December 2021, as we’ve been alluding to, Cerner gets bought by Oracle for $28 billion.

Cerner, as we talked about, was and still is a big company. They’ve acquired all these companies over the years. There’s a lot of revenue there, they’re big internationally, and they have the DoD and VA contracts ongoing, which was a lot of money coming in. So they’re doing, maybe call it, $5.5 billion a year in revenue—maybe a little more—but that’s been flat to declining since 2018.

Oracle acquired them in 2021. Oracle now no longer reports Cerner’s financials as a separate segment within the company. They talk about Cerner and Oracle Health being a “headwind” to overall growth and profitability for the company. There have been a lot of layoffs within Oracle Health. Maybe it’s too early to tell, but I think this is probably one of Oracle’s worst acquisitions of all time.

You’d be hard-pressed to find an analyst who would say that this was a great acquisition by Oracle, shall we say.

Yes, to put it lightly.

Meanwhile, this is all just great for Epic and totally reinforces Judy’s whole thesis of company-building in this space, and her whole story to customers of one single, integrated platform that is hardcore about software development, 100% customer-focused, and will never acquire another company, never go public, and never get acquired. You can see why it’s just this warm-hug embrace to their customers out there.

Yeah, it becomes more and more different from everyone else in the market as time goes on.

And the lesson here, which is really interesting, is that they started slow. They started really slow, and they did things in a way where they almost built up momentum for the future, in a way that would pay off 20 or 30 years down the road, rather than inorganically trying to take shortcuts and pull the future too far forward into today.

The first 20 years kind of look like an unimpressive company because they were just building strength, building muscle, and growing the way that you need to grow in order to be as bulletproof as they are today.

That’s such a good point. And I was trying to square in my mind earlier, when we were telling the meaningful-use story, how, on the one hand, it really was obviously good for Epic and an accelerant, and it further cemented the market dynamics where they were already emerging as the leader. On the other hand, when you talk to people within Epic, they’re very mixed on meaningful use.

Yeah, and I could never figure out if that was lip service—“Oh, it didn’t help us that much.” But when you look at the graphs of customer counts and revenue, it was going great.

Yes, it did pull forward some growth, but the trend lines were going to get here anyway. It’s not like it was some big, massive step function in 2 years. That is not how the graphs look. As we told the story, they had the better product, they had the better customer offering, and they were on the way to winning anyway.

I think, to your point, they’ve always been philosophically allergic to artificial growth. I wonder if that’s part of why, to the extent they are serious about really being ambivalent about meaningful use, that’s the reason why.

Yeah, they want to win slowly.

They absolutely want to win, and they’re going to win, but they want to do it slowly because it will help them win harder in the long run if they win slowly.

Yes. Yes.

So meanwhile, for them, I think things have never looked brighter, basically. There are some risks that we’ll talk about, but they’re now starting to sell internationally and build that as a real market. It’s not huge, but it’s, you know, 10% to 15% of the business now.

In 2023, they went live at London’s Guys and St. Thomas’ NHS Trust, which was, I think, their biggest-ever single implementation and was happening in the U.K. The U.K. is becoming a real market for them.

Interesting. You know, there’s another big tailwind that’s happening for them. Like you’ve probably been hearing from me over the course of this whole episode, there are tailwinds that don’t make you feel good, but they’re still tailwinds for Epic’s business: the acceleration of the trend where hospital systems are merging and rolling up all the smaller local practices.

Oh, totally. It’s just an excuse to rebid.

Yes, there are a bunch of reasons this is happening, but one of them is the Affordable Care Act, which also increased the compliance burden of practicing medicine. That burden was already really heavy from all the regulations over the years, including HITECH, which added big penalties for HIPAA violations.

That’s another thing we didn’t talk about in HITECH. One of the things that came stapled onto it was even stricter oversight of HIPAA. So you now require scale as a local medical practitioner to be able to afford the compliance burden. And what that means, if you play it out, is that all health care—or most health care—gets provided by these megagiant hospital systems rather than community clinics.

That really benefits Epic because their strategy from 30 years ago of going after the biggest and most complex health systems now looks totally genius, because those are the only customers really left standing.

Yep. Totally.

I actually think this is one of the most messed-up things in the U.S. health care system. Aside from the really big, hard-to-negotiate-with private health insurers that have gotten a lot of power over the ecosystem, you now also have really big local pseudo-monopoly hospital systems. And so this consolidation is not good. This consolidation will not lead to prices coming down. Let’s just put it that way.

Yep. But you see why it’s been happening. You and I entered doing this episode expecting to be like, “Ah, the hospitals are really the bad guys. They’re charging so much money.”

I don’t think the hospitals make very much money. They don’t make much money at all. They’re barely surviving, which is why they’re merging, right? And any money that the very successful ones are making, they tend to plow back into growth because, like we’ve been talking about, hospitals kind of need scale in this day and age.

And so it’s new buildings and acquiring other practices locally. Everyone is just chasing scale because they constantly are having to do business with other scale players. So they need to scale up to do that.

Yep.

So that brings us to the other big, I think, potential growth vector for Epic going forward that they’re absolutely pursuing. They have enough scale on the provider side now that they can start going to the other big players in the system—the payers, and then also the biotech and pharma side—and say, “We actually have ways that we can work with you guys now and offer compelling products to you.”

So they call this the system of connectedness, or the “grid of care,” that you might sometimes hear from Epic. But this is building products for and working with payers, the insurers, and pharma companies, as well as other things like home health companies or post-acute rehab. But really, the big opportunities are payers and pharma.

One example of this is prior authorizations.

Yep. Prior authorizations are a huge problem in the American health care system. This is when a patient needs, or a doctor thinks that a patient needs, either a medicine or a procedure that the payer—the insurance company—says, “Well, you can’t just go do that. It’s expensive enough that you need to ask us first, on a case-by-case basis, if we’ll cover it for this person.”

Yeah, exactly. We need you to prior authorize this.

Epic now has enough data and enough scale on the hospital side that they can go to the payers and say, “Hey, prior auth—our customers don’t like it, you don’t like it, and we have all the data. We can just automate this for you.”

I would argue this is an area where you are seeing one of Epic’s favorite business strategies, which is: We sell the main thing to a hospital system. We see what other vendors they’re using. We figure out if they should also buy that from us, if we have some competitive advantage other than just being a single vendor to be able to do it. Sometimes, just being a single vendor with the same offering is enough.

But then they try to figure out if they can also do that and include it in their enterprise agreement. They want to sell everything to the hospital and be the single IT vendor. The hospitals do still buy other IT services.

So the Epic playbook is: go from selling a hospital 1 thing to figuring out how to then sell the hospital everything. And then, once you've exhausted that, figure out who else you can sell stuff to. That's why they're looking into payers, and that's why they're looking into pharma companies. It's how can we leverage the asset that we have to then, now that we've saturated this market, sell to other customers too.

Fair point. All right. So that brings us to today. If you're in this ecosystem, you're going to say there are 50 things that Epic does that you didn't talk about. That is true. There's no chance that we could talk about all of that on this episode, but there is 1 really cool, interesting thing that they are doing right now that we do want to talk about as we catch up to today. Then we'll give you the stats on the business today, and we'll talk about the future, and then we'll do our analysis.

7. Cosmos Changes The Data Game

They launched something a few years ago called Cosmos.

Yeah, this is pretty cool. This is awesome. So they've realized, okay, there is data on patients that's owned by patients and hospitals, not by us. That lives on a bunch of different servers. Some are on-premises at hospitals, some are in our cloud, some are on AWS, and some are on Azure.

But either way, there's all this interesting structured patient data stored in Epic instances. Can we anonymize that and get it all in an Epic-hosted instance of something very queryable and very useful for a bunch of activities? They did it. It's called Cosmos. And Cosmos has 295 million patients' worth of data in it. It's all anonymized.

Yep. And I think that across those 295 million patients, there's data from 15 billion individual patient encounters.

Yes. So structured data from a doctor encounter 15 billion times with patients. It's crazy.

So we were asking the question earlier: do EMRs or EHRs really benefit patients? And there are some arguments that, aside from how awesome it is to just access your records at home, there's a bunch of stats around, yes, it benefits patients. It's patient safety; it's patient ease. There are studies that say 45% of patients reported improved quality of care with EHRs, while only 6% noticed a decline.

Epic has a stat: in 2023, its system prevented 66 million potential adverse drug interactions and 250,000 potential surgical errors. All of this seems very plausible. If you're tracking all this stuff digitally, you can do intelligent things that improve care. But on top of that, more interestingly, what if all of the data was actually in 1 place and you could do stuff with it? What is the value of having something digital versus paper? This is sort of like the utopian dream that people have always had and that the Bush administration and the Obama administration had. Epic is finally doing it.

Yes. And 1 very clear, interesting, illustrative example is: imagine there's something wrong in the water supply, but all the records are on paper. So unless you're going and pulling individual files and then looking at some specific test result across a whole bunch of people—which you probably need a research grant to do, to actually go and sit down and pull all these individual files—you're not going to find it.

But if you have a whole database full of information and you can quickly and easily look horizontally across a whole bunch of records, this is Flint, Michigan. This is how they figured out there was a water problem in Flint, Michigan. And you would not have found it in a paper-based world. Cosmos is that on steroids.

What can you figure out about the whole world if you have 295 million patients' worth of data in an anonymized, accessible database, and it's accessible for free to any institution that contributes data to it? So, of all the Epic customers, anyone who opts in gets to access and query it for research or cool stuff like, “I have a really unique patient who has some crazy condition I've never heard of. Has this ever happened to anyone else in the world? And what was the result?”

Yep, this is what's really cool. They've productized this and made it free, so any of their customers can just turn it on.

So, Ben, exactly like you're talking about. I'm a doctor. I have a patient. There's something going on. Cosmos can surface to me: “Hey, here are other examples of similar cases, anonymized.” They call them look-alikes, I think.

Yeah, look-alike patients throughout the history of everybody that's ever been on Epic.

And here are the doctors who treated those patients. So this is already happening quite a bit, I think, with rare diseases and crazy cases.

I think so, too. It'll be useful for clinical trials. It'll be useful for research. I do think it's free right now, which is pretty interesting. We'll have to see how they price and package it in the coming years, but, yeah, you're right. This is like the utopian dream of, “Wouldn't it be cool if we had all this data in 1 place?”

Yep. And then that really presages something we're going to talk a lot about in analysis, which is: wow, imagine what you can do with AI with this data.

Yep. But we will get to that in a minute.

All right. So the business today, let's just make sure we're all on the same page on where things stand. They have 607 customers with 3,200 hospitals between them. As you know, they've never lost a single one, which is insane. They add 10 to 25 new health systems as customers each year. Their customers represent 590,000 physicians and 495,000 staffed beds. And that's across 325 million patients worldwide.

Yep. 280 million of which are in the U.S.

So, almost all of America gets seen in some part of their healthcare by an Epic system.

Yes, that is correct.

In revenue last year, in 2024, they did $5.7 billion, which grew 13% year-over-year over the last 5 years, and 16% last year, from $4.9 billion to $5.7 billion. David, I wanted to bring something up with you. They actually do less revenue than I would have figured.

Yes. And for people that are like, “$5.7 billion is a lot. What are you talking about?” If you divide it by the number of customers, it comes out around $10 million a year a customer. If you live in the enterprise software world, that's not crazy. This is the central nervous system of these giant health systems. This is what everything else relies on. They run their entire business on this.

It costs a lot of money because there are all these people administering it. I talked to someone at a mid-sized health system who said there are 100 full-time employees administering Epic at her hospital system, but Epic actually only captures about $10 million a year of the value created. So there's this interesting dichotomy of, yes, it's a lot of spend, but the minority of it actually goes to Epic.

Yeah, it is this element of—I know I said altruistic earlier when talking about the company. You're right. It's not altruistic, but I think it's really, really important to understand. It's a very long-term orientation. They could charge their customers a lot more money.

They definitely could. They didn't used to be able to. I think they used to saturate willingness to pay. Now that they're this dominant, they probably could.

There are so many costs that get paid out to the consultants—Accenture, Deloitte, IBM, Nordic—the post-implementation people, the optimization people, the headcount internally, the lost revenue from doing the switch and taking that downtime. I mean, this costs health systems a lot. Epic actually just doesn't capture that much of it.

I'm not as convinced that they have room to take price as much as I'm surprised they don't capture more of the overall value of a central nervous system for a hospital system. They could probably capture some more. And as we get further and further in the future and they get more and more dominant, they certainly will gain more pricing power.

I think they could. Again, to your point, on average, $10 million a year is the cost for a piece of software that is so absolutely vital to your business. I think they could capture more. And I think part of the reason they don't is they never want to give a customer any kind of incentive to rebid or consider another solution.

It's kind of like the Jeff Bezos quote about AWS in the early days, that he wanted it to be an irrational and irresponsible decision not to use AWS.

Oh.

To pick someone else.

Yeah, I think they want it to be an irrational and irresponsible decision for a healthcare system not to use Epic.

Yeah, I think that's right. And just 1 other number to throw out to analyze this on a different vector: when I said $5.7 billion, top line isn't that big of a number in healthcare. UnitedHealth Group, which owns an insurance company and a hospital system, does $400 billion a year in revenue.

That's wildly apples to oranges because they're actually administering care in the case of their hospital system. So it's a different thing. But that company does $35 billion in EBITDA. UnitedHealth Group's profit dollars are 6 times Epic's entire revenue. So Epic has become this incredibly important and powerful linchpin in the system despite not actually having that big of a profit pool in the industry.

Yep. Or, even maybe better put, it doesn't have that big of a revenue pool in the industry. I came into this episode expecting to be like, “My God, this company just mints money and is such a small company, given how much people talk about it as a part of the healthcare system. I just expected it to have more cash moving through it.”

All right, other quick stats. 14,000 employees today. The market share is only 42% of hospitals, but those are the largest and most successful hospitals. So they actually have a bigger share of all care performed. 58% of ambulatory physicians now use Epic. 79% of the U.S. ends up using Epic in 1 way or another. In terms of profit, we heard a few different estimates.

There are EBITDA margins somewhere between 30% and 35%. So, just to be conservative, we'll take that at the low end and say about $1.7 billion in EBITDA if those estimates are correct. So then that kind of brings us to: What is Epic worth? Which is sort of a silly question, because if you own shares, it would be very difficult to ever sell shares to anyone other than the company itself. So what's the point of valuing something that you can't own? And what's the point of owning something that'll never provide a financial return?

Yeah, there's never going to be a liquidity event for your shares, right? But it's worth doing. So we started this episode talking about Judy being one of the most successful, if not the most successful, female founders of all time. Forbes has a list of the wealthiest self-made women. I think their estimate is silly low.

Yeah, wildly wrong.

In 2021, Forbes estimates her net worth to be $7.6 billion, implying that the company is worth, what, $15 billion.

That is a ludicrous valuation for Epic.

If you do the very conservative thing of looking at Cerner's EBITDA multiple of 30 times when Oracle bought them, and you assume, conservatively again, a 30% EBITDA margin at Epic, that would give you a number of $51 billion, valuing Epic at $51 billion.

But come on, Epic's revenue streams are way more durable than Cerner's. And Epic is growing while Cerner was going through a hard time.

You could get to a similar number of that $50-ish billion if you just slap a 9× revenue multiple on it, which is a reasonable public-market software comp. But this company has insane durability.

Yeah, it's missing the point that I think this is maybe the most durable software company in history.

Yes. My guess is, if this is public—which it never will be—investors would value it somewhere in the $100 billion neighborhood, giving Judy's shares a value of about $50 billion and making her the wealthiest self-made woman in the world and one of the most successful entrepreneurs, period. And that would put Epic among the 150 or so most valuable companies in the world, right along with Shopify, Arm, Lockheed Martin, and Starbucks.

Yep.

Another thought exercise to kind of come at this question is: If, in some alternate universe, Microsoft or Google or Amazon—or, I don't know, maybe even Berkshire Hathaway, maybe Apple—could acquire this company, how much would they be willing to pay for it? I think $100 billion is on the very, very low end of that spectrum. That's 4% of Apple or Microsoft. Of course, they'd be willing to pay 4% of themselves for this. There's the market, there's the durability of the revenue and the profits.

Oh, I see. You're arguing that on a financial basis alone.

Yes. And also the strategic value on top of that. The strategic value is both in terms of becoming the most important partner to this huge industry—to the biggest players in this huge industry in healthcare—but then also just the data and the AI world that we're in today. I mean, my God, Cosmos: 15 billion patient interactions. You think Google or Microsoft or whomever wouldn't pay a lot of money to own that? Like, of course they would.

Part of me is like, “Ooh, thank God Judy's putting it in a trust.”

All that to say, this is never going to happen. So, the final chapter of the story here is that Judy is 81 years old. What is going to happen at some point in the future? Even if she's the next Warren Buffett here, we're still talking about the next couple of decades. There will have to be a transition. Like you said, Ben, she owns about half the company economically and 100% of the voting shares. She has been transferring her nonvoting shares into her foundation called Broots and Wings. So she's signed the Giving Pledge and is transferring her wealth into the foundation. They then sell those shares back to the company at a price that changes every year, and that funds the foundation.

Judy has announced that upon her death, all of her voting shares—so, the 100% voting control that she has of the company—will transfer into a quote-unquote purpose trust that will be administered by 3 constituencies: 1, her family—her husband, if he's still alive, Gordon, and their 3 children; 2, a set of 5 longtime senior managers from Epic; and 3, I believe, 3 customer CEOs, CEOs of big customers.

Mhm.

That group will manage the trust. And in the trust bylaws are several things. One, the company can never be sold or taken public. Ironclad—that can never happen. Two, the next CEO of Epic must meet 2 criteria.

Oh, be a software developer, right?

Yes. One must be a longtime Epic employee, and two must be a software developer.

Ah, I love the idea of the CEO being someone who came from the core skill of the company. It's very Costco. You think about how all the Costco CEOs have been people who were core to Costco's operations over the years and had worked there for 30-plus years. Nike's heyday was when a guy who started in sneaker design ended up running the company, Mark Parker. There's sort of a beauty to when that happens.

There is a playbook to this.

Yes.

One last thing to know about Epic today, looking forward. We talked a lot about interoperability and their notorious lack of partnership or ability to access their data or integrate with them in the past. A lot has changed, but there would be a whole other 4-hour podcast on specifically how there are many programs, each with different methods of integration, different names, and different standards. They've all evolved over the years. They come with things like revenue shares back to Epic. And if it is the rare true partnership, they can also come with warrants to own a chunk of your company. But the bottom line is, you actually can do a lot more building on top of and integrating with Epic than you could have 10 or 20 years ago.

All right, that's our Epic story. Should we move into analysis?

Let's do it.

8. Epic's Enduring Power

So the first part of analysis is a section called “Power,” which we have shamelessly ripped from Hamilton Helmer's book 7 Powers. The question is: What is it about any given business that allows them to achieve persistent differential returns, or to be more profitable than their closest competitor on a sustainable basis? And there are 7 of these. There's counter-positioning, scale economies, switching costs, network economies, process power, branding, and cornered resource.

As usual, we have to separate the takeoff phase from the phase that they're in today. David, I'm curious: Do you have one for the takeoff phase? Do you feel like you can analyze why, in the takeoff phase, they gained power?

Well, it's funny. They didn't take off for quite a while. Right? And I might argue they didn't actually have any power during the takeoff phase.

Yeah, I don't think so. I mean, ultimately, their biggest point of differentiation, being the single platform, didn't matter that much in that phase because everybody else was starting up new, and they hadn't built that much of it out. There wasn't that much computing adoption among hospitals generally, so it didn't really matter.

I think what was happening is they had this kind of niche market in the proto-EHR world where they were able to sell something small enough at enough of a profit to be self-sustaining while they bided their time and built out the bigger suite, which they were able to basically use time as a resource that most people don't flex.

Most entrepreneurs want to do something in a short period of time, usually because their capital structure demands it. And so you launch a platform and try to get other people to build the applications because you're in a really, really dynamic, fast-changing market. Or you launch an application on a platform that somebody else built, similarly because you're in a dynamic, changing market.

And Epic's market was developing slowly over a long period of time. They were willing to run a small business for a long period of time and then sort of grow into building all the functionality for the eventual huge market that wanted all the functionality tied together.

Yep. Okay. So then what powers do they have today as the scale incumbent?

Yeah, a lot. A lot is the answer. A lot. A lot. I mean, the most obvious one is switching costs. This is the biggest switching-cost piece of software ever.

Ooh, that's interesting. Is that true? Do you think it's bigger than people's ERP instance?

For sure. This is like an ERP on steroids. This is the most important nervous system for your entire business.

Oh, yeah. And by the way, if things go wrong or it goes down, people die.

That's a very good point. So people spend, including all the opportunity costs and everything, hundreds of millions or billions to put this in. Are there any ERP or CRM implementations that cost $1 billion, $5 billion, or $10 billion? Maybe, but I don't think any are so core to everything about your customer's business.

You're right. Even if it was really expensive to put it in, you're still more likely to swap out one of those systems than you are your EMR at this point.

Right. It's so much more than a medical record.

Yes. Yes.

So that's definitely true. Scale economies is definitely true to the thing we were talking about a minute ago. You can't make the investments in product breadth without having the scale that you have today.

Yep.

You also can't do their clever bundling in a bunch of free stuff that you might want to use in the future unless you have the scale economies. Also, they just develop a lot of software that they either throw in for free or offer to you in the future. And the only way you can develop that much software is to have a huge number of customers to amortize it across.

Yep. If you're trying to start a new EMR today, you have to write a lot of software to be competitive.

Yep.

Network economies.

Yep. You bet. You betcha. They hit this interesting tipping point where they could win just on “We're the best, most reliable platform” for a long time, and then at some point it hit a tipping point where there was a second value proposition, which is that most other hospitals are also on Epic. So your patients and your doctors will all appreciate it if you adopt Epic to make it easier to interoperate with the records at all those other hospitals.

It’s like icing on the cake.

Oh, you’re talking about interoperability. I mean, there’s interoperability, but there’s also—do you think you will need to hire more doctors in the future, either to grow or as your existing doctors retire?

Right, pick the standard.

Yes. Well, what do they know how to use, and what are they trained on?

Epic.

Great.

Yeah, that’s true. It’s a second network economy in addition to the Care Everywhere concept of being able to share my records across institutions. There’s 1 institution in Seattle that I go to for my son’s food allergies. They’re on Cerner. I have 3 institutions that I’ve gone to over the years for various other stuff that are all on Epic. I don’t know why that 1 hospital is still on Cerner.

Yeah, it’s really annoying to you, isn’t it? I’m sure everyone else in Seattle is like, “Why don’t you just do the thing that all the other big hospitals do so that it’s all the same platform?”

That’s network economies.

Yeah, it’s kind of amazing that they have built a network economy business. Who would have thought: medical software? There is value to every participant in the system, from the other hospitals to me as a patient to doctors. If that 1 holdout hospital that, for some reason, is on Cerner just switches to Epic, it’s the definition of a network economy.

I think they also have branding. They’re the most trusted by all the most prestigious institutions. At this point, if they had no other powers and they just came in at identical bids on the table, 100% of the time people would say Epic because it’s going to work. “I know your brand. Don’t get fired for buying IBM.”

Yep, exactly.

Yep. I didn’t think about that. I was like, “Oh, no. They don’t have branding. It’s not really a consumer brand.” But no, no, you’re totally right. They totally have branding.

That is in part why they built that massive auditorium to bring the whole universe of Epic together every few months.

Yep. And then I think you could argue that they do have process power. I think that’s the last one I would say: in their development process and the language that they use, it’s kind of unique to Epic. This is more of a defensive one. I think it’s about the training of young employees, the way that they sort of make them the Epic way.

Well, interesting. There’s the Epic way. I buy that generally, but I think specifically you can point to their software developers, who they’re hiring out of college, mostly building on Caché and MUMPS, and they don’t develop transferable skills. It’s not like you can easily then go work at Cerner and port that over, right?

I mean, sure, if you’re a great software developer, you can learn other languages and other frameworks. But if you spent your whole career developing in this sort of proprietary Epic system—and I should say “proprietary” is the wrong word. They don’t make MUMPS or Caché; a separate company called InterSystems makes them, and they license them from InterSystems. But most of the rest of the industry does not operate on that. It is a bit of a process power.

Yep, I think that’s fair. All right, that does it for power. Great. Do you have any playbook themes?

Yes, I do. The one that jumped right out at me was the heavy spend on research and development. This one is from friend of the show Arvin Navaratnam at Worldly Partners, in the always excellent write-up that he does associated with these episodes, which we’ll link to in the show notes.

Epic, of course, builds, as we’ve talked about a million times, their whole systems from scratch, and they don’t buy companies. It’s all in-house development. The result of that is that 35% of their operating expenses are spent on R&D. If you compare this to athenahealth, a competitor, that’s 10%, or Oracle all-up is 23%. If you look outside this industry, it’s actually equivalent to Apple at 36% of their operating expenses on R&D. Amazon’s 28%. So 35% is quite meaningful. It’s quite high.

Yeah. The only one that beats them is Google at 45%. Think about all the money that Google is just dumping into R&D for all these future-looking projects, and you know they’re a core technology company.

On the one hand, duh, because Epic doesn’t really buy other companies or spend any money on sales and marketing, and they keep their G&A low. What are their other expenses? Of course it’s going to be high on R&D.

Yes. Unlike most companies, they spend zero on sales and marketing, right? If they classify those 8 people that they employ as salespeople, maybe they spend—I don’t know—not much.

Yes, small. But I think to get to this point that we’ve been hitting in a bunch of different ways, as long as you can win deals in the short term, in the early days, without sales and marketing spend, and you can just spend all of your costs, or as much as possible, on R&D, that R&D does compound over time to give you a big competitive advantage later in life that other forms of spend do not.

Yep, especially in software.

Yes. My next one is about growth. I was reading an article that said Faulner once described her approach as climbing a mountain—not trying to see the entire mountain at once, but by focusing on the next hill in front of her—and it reminded me of something I’ve read before. Does this remind you of anything?

I’ve drawn a blank. Go for it.

Okay. In Paul Graham’s legendary 2012 essay “Startup = Growth,” he’s talking about how the whole thing you need to do as a startup is focus on next week’s growth. He then says:

“In theory, this sort of hill climbing could get a startup into trouble. They could end up at a local maximum. But in practice, that

And so, in most industries, when you're a startup looking down the end of the barrel, you're like, “All right, I am not doing this top-down enterprise sales thing. I'm going to do product-led growth, get some bottoms-up adoption, and play the Slack or the Figma game, or whatever you want to call it.” But HIPAA makes product-led growth much harder.

So, from the great Substack health API guy: under HIPAA regulations, a business associate agreement typically needs to be signed by someone with the proper authority to legally bind the organization—not individual providers or employees. So, David, this is something I kept thinking: Why doesn't one doctor or one practice adopt something, then get bottoms-up adoption, and that's how we'll get the next EHR in the future?

No, these are all CIO sales because they all need to be, right? You can't have the typical, “Oh, a small team within a big company is going to buy this on a credit card.” Nope, not going to happen. There are some exceptions. I know there's some PLG that happens in healthcare, but it's way, way harder to figure out how to do that than normal B2B SaaS.

Yep. And my last one: Judy is a wacky, wacky founder. And so is every truly great entrepreneur that we study. She is truly one of one. I mean, she has stated that it is more likely that she'll die than retire, just like Warren Buffett always jokes.

She is singularly focused on making Epic great in her life, just like Ingvar Kamprad and IKEA. She's totally obsessed with rapidly incorporating feedback from customers, just like Jeff Bezos. And she believes, correctly, that the nerds will prevail in the long run—at least correctly for her market, her point of view, and her industry. She is completely unique in the way that all the great founders we study are.

Yep. You foreshadowed in the intro that Judy is almost certainly the most successful, by any measure, female founder in history.

I think that is actually true. I looked up to try and see who else might even come close. You could maybe make a case for Estée Lauder. However, despite Estée Lauder reaching about a $100 billion market cap during the pandemic, it's only worth about $20 billion today. I don't know what happened there.

Taylor Swift and Oprah, which we've talked about—and we did our Taylor Swift update at the Chase Center show last year—we think the enterprise of Taylor Inc. is only worth about $11 billion. I mean, only, quote-unquote.

I love that you and I have an ongoing Taylor Inc. calculator.

Yeah. Oprah's less than that. Forbes currently thinks that Diane Hris is the wealthiest female founder in the world, or at least in America. She's the founder of ABC Supply, which is one of the largest roofing suppliers in the country.

You know what's crazy? This means that the 2 most successful female founders in America are both based in Wisconsin.

I know. Isn't that amazing? I was going to say that, too. It is a big and great company, for sure. However, ABC Supply does about $20 billion a year in revenue. I assume not at anywhere near Epic's margins.

Yep. So, I think Epic is worth a minimum of $100 billion.

You're now up to a minimum of $100 billion.

Well, I'm going by my rubric of, if somebody could acquire this company, what would they be willing to pay?

Yes. That's kind of funny. If someone offered you shares in Epic right now at $100 billion, would you pay it?

Oh, good question. If I could get profit distributions from Epic, yes, I would love to hold Epic shares for the rest of my life. I would for sure pay that.

Yeah, I totally agree. If I were banking on a liquidity event, then no.

Yeah, but me personally, yeah, I would love to hold shares.

Yeah. When people are speculating on the value of something, the interesting way to turn it around is always, “Okay, are you a buyer at that price? Open your wallet.” It makes it feel much more real. I would for sure buy shares of Epic at $100 billion. There's no way I will ever be allowed to, but I would love to.

That sounds like a standing offer, and the show is not investment advice in this category.

Man, it is crazy how powerful founder continuity can be. I was thinking about this: How many other companies have had a single founder-leader for 47 years? I mean, Jensen Huang is 31 years into NVIDIA. Zuckerberg is 21 years into Facebook. Obviously, there are a lot running smaller companies. There's Berkshire Hathaway, right? That's actually the exception. Buffett's been running Berkshire for over 50 years, but in Judy's case, having 47 years to imbue the founder's personality into the organization as it's built all the way to this scale is really rare and really powerful.

Yep. And as you've alluded to, I mean, she really is like Jensen. She's still running the company. She's still highly engaged.

Yep. Absolutely. All right, bear cases.

Yep. Let's do it.

All right. A few different things are contributing to the bear case, if I were to make one. U.S. customers had very big dollars they could pay. The market is big because, one, the U.S. economy and population are large, and two, unfortunately, healthcare is a large percentage of that. So, if the future is coming from international, that could be less fruitful: much lower willingness to pay in most countries for healthcare administration software. So, that's one sort of bear-case element.

The other would be if this particular suit ends up actually getting picked up and there's a material event that happens for Epic around antitrust and a violation of the Sherman Act. That's a huge, huge problem, and that is a company-changing thing if that comes to bear. So, we'll have to watch the news on that.

The third is that there is legislation from the last few years about information blocking. There's an act that passed called the Cures Act, which essentially says that an EHR or health system cannot block access to information. So, even though Epic makes the very reasonable argument around data privacy and security and, in the long run, patient safety by locking down data, the Cures Act makes things like screen scraping, Chrome extensions, or RPA legal as a means to extract data from Epic, and there's nothing they can do about it. That is required and allowed.

Interesting. So, at this point, I don't know how that could really displace them. What also sort of begs the question is, yeah, what would you do? You scrape Epic data, but you're not going to build Epic, right? Epic has always been hyper-sensitive about locking that stuff down for good reasons for their customers, but also for their own durability reasons.

The fear is that someone deploys to, like, 70% of your customer base. They're all using some Chrome extension. They're all feeding the data into some nice UI. That nice UI is the thing everyone prefers to use, and then that UI vendor is like, “Oh, instead of using Epic on the backside, you could use my own homegrown thing on the backside.”

But the probability of that happening is near zero. Yeah, not going to happen. So, I'm not that worried about that.

The other thing is maybe, if interoperability comes to pass in a bigger way, then you could kind of see best-of-breed applications becoming a more dominant paradigm versus the all-integrated paradigm. Again, this feels pretty hand-wavy to describe as the bear case.

Yep. Well, and should that happen, they still have the Microsoft playbook, which they absolutely run. Okay, let's even say there are better best-of-breed point applications out there. Great, Mr. or Ms. Hospital CIO: You could go pay money for those, or you could just keep buying your Epic enterprise license, where you get access to all of these point applications that you need for 1 price.

Yep. And then the last one is paradigm shift. This is always the bear case for any dominant company. Is AI going to be such a dominant paradigm shift that it changes the needs of health systems? Or is value-based care going to be such a paradigm shift that it changes what a health system is? Maybe you don't go to the doctor most of the time. I have a hard time even imagining what it could be that would mean you wouldn't need a system like Epic.

Maybe new care-delivery models will make the old type of EMRs obsolete. This requires some imagination, and again, it's very hand-wavy. Epic also has its eye on all of this, so it's hard to imagine any of those things being the—

Yeah, right. My question to you is going to be: How much do you want to talk about AI here, or how much do you want to talk about it in the bull case?

Yeah, it's probably more a bull case.

All right. Well, let's do bull case. Great. All right. The bull case is they've successfully expanded from EMR plus billing to all these other specialties and modules. They basically served all those same customers with even more products. Those customers, A, will need more products in the future; B, there continue to be more health systems they can sell to; and now, C, they are expanding the customer base from just health systems to other types of companies—pharma, payers, and researchers—from this new dominant position that they're in.

So, they're sort of expanding to a whole new set of potential customers who could pay them because of their dominant position.

I think that's extremely credible. They could have an even larger business. I mean, prior auth seems like the first and most obvious business opportunity for them here. In some ways, hospitals are kind of crappy customers because they don't generate much profit. Insurance companies are probably much better customers if you can figure out something really compelling for them.

Yep. Unfortunately, again, I feel like every time I come up with some profit pool in healthcare, I'm like, “Boy, I really wish that wasn't there.” I have to sort of think philosophically about why I keep feeling that way. But I'm definitely critical of anything that becomes too profitable in the healthcare system.

Well, behind that, though, is that you as a consumer—we as consumers—feel like some of these entities are not in any way providing value to us.

Right. I feel like I’m getting ripped off.

Yeah. I saw a study that said 30% of spending in healthcare—in the 18% of GDP—goes to waste. Truly waste. I saw another one that quantified it and said $800 billion of waste is in the system. The equivalent of Switzerland’s entire GDP in waste exists in our healthcare system.

Yep, sounds about right.

Another interesting thing that someone brought up is they asked me how much I pay in premiums a year for health insurance for the family. I was like, “I don’t know. I think somewhere in the neighborhood of $25,000 to $30,000.” Their response was, “In what world are you ever going to use $25,000 or $30,000?”

Don’t imagine what’s on the bill, because it’s all made up. All the numbers are made up. The top line, the negotiated rate, what I’m covered for—they’re all made up. Just imagine the dollars coming out of your pocket for healthcare. And don’t just pick this year. Pick a 10-year period, because stuff might come up. You’ll have surgeries in some years but not others.

What are you actually willing to pay—what dollars are you willing to give a doctor out of your pocket—for everything you need over the next 10 years? Compare that against everything you’re paying into the system if you were paying the employer side, the employee side, and all the out-of-pocket stuff yourself. We’re not getting a good deal. We’re just not. We can all feel it.

Yep. All right, continuing the bull case. There’s a whole category called ambient listening. One of the biggest complaints that people have with EHRs is that doctors spend too much time in front of computers. They’re typing in notes. They’re clicky-clicky on 9 different dropdown boxes and alerts.

Sometimes that gets fixed by having someone in the room typing while the doctor is doctoring with you. But either way, there’s a scribe, and the doctor is also at a computer because they need to read the medical records. The doctor ends up in front of a screen. They’re wasting a bunch of time typing, or they’re hiring someone to type.

Oh my God, AI would be amazing for this. If only there were an ambient-listening AI scribe that could write down, categorize, and structure all the data that comes out of an interaction with a physician. Well, good news: it’s happening. It is happening, and it works.

So, David, there are a few companies that are partners of Epic, right?

Yep, that’s right. One is Microsoft and Nuance. Microsoft bought a company called Nuance a few years ago. They used to do Dragon Dictation, so they’ve always been in the voice space for healthcare.

They now have an ambient AI product. Then there are 2 big startups: one called Abridge, based out of Pittsburgh, and another called Suki. They have really good ambient AI products that plug into Epic and are AI scribes. Physicians who use them love them. You can just focus on the patient, and it scribes everything.

Yep. This would be a great solution to some of the doctor burnout and fatigue and too much screen time that a lot of physicians are experiencing.

Yep, that’s the bull case for this. These products are all partners of Epic, and in some cases Epic may have relationships or warrants with some of these companies, or just revenue. You could also imagine Epic might want to build its own product someday. It’s a big opportunity.

Epic is kind of the choke point of the industry at this point, deciding what software innovations reach hospitals. I don’t think it’s overstating it to say that if you’re developing new breakthrough software to be used by physicians in hospital systems, Epic is the one that gets to decide whether you’re going to reach that customer or not.

Yep, exactly. I think the basic bull case on ambient voice, ambient AI, is that this is a great new product revenue opportunity. The mega-bull case, to borrow from our IPL episode, is something one CIO was talking to me about. He said, “Look, I don’t have high confidence that this is going to happen, but as ambient AI becomes better and better, is there a version of the future where the EHR, quote-unquote, itself just fades into the background and this all becomes an AI operating system?”

What is the EHR primarily doing? It’s capturing data about what happened. It’s recording it both for clinical uses and, importantly, for billing purposes. Then what needs to happen with that for billing purposes? That data needs to get shipped over to the payer, whether that’s the government or an insurance company, and then it needs to get judged and adjudicated. Payouts need to happen.

Is there any real reason why this can’t all be done by AI in the future? Why do we actually need user-facing software here, or a lot of it? There may be some future—and this is a utopian-type future—where a lot of the administrative bloat and costs in the system actually get taken care of by ambient AI, sort of just sucking it all up.

I’ll believe it when I see it, but it sounds nice.

Exactly. But a real CIO of a real big hospital system made this case to me. It’s just so hard to believe that bloat ever goes away. Getting administrative costs down is so hard. What are you going to do? Put a whole bunch of people who have good administrative jobs at hospitals out of work? That is going to be hard for our whole society.

For sure. This is the question about AI, right? But you’re right. I think it’s a bull case if you’re a shareholder of Epic, though, right? If Epic becomes that system.

Yep. All that to say, like every industry, AI has big potential here.

Yep, I think that’s fair.

And that’s on top of everything we’ve already talked about. No one is switching off of them. Revenue from their install base will just continue to grow. They’ve become the standard that everyone will switch to. Once they get a customer, they don’t leave.

Honestly, at this point, regulation is probably also a bull case. This is sort of what happens once you become an incumbent. Regulation tends to shut out new entrants and entrench the older companies, because the older companies are the ones that have the resources to comply.

So, they’ve got their ear to the ground and a ton of resources. When new data interoperability standards come along—there’s one called TEFCA that’s in the works—Epic can be the first to implement them well. That can further entrench them as they comply correctly with new regulations and new requirements.

Yep. As we talked about, there’s probably pricing headroom at this point, which may or may not be great for their customers, but it’s great for Epic. They probably can start extracting more value if they wanted to pull some of the future into today.

I would be really, really surprised if they do that.

I would too. I don’t think they will. Then the last one is becoming a platform. They really haven’t done this to date. They’ve built it all themselves, but I’m curious if they start platformizing a little bit more.

They’ve always had little things here and there, like App Orchard, but I’d be curious if they at some point make a real play to be a robust platform layer upon which other applications build. To your point about vertical software, maybe that’s just not the way this ends up playing out. But they have the opportunity if they want it and think it would be more valuable than their current path.

All right, should we do quintessence?

The biggest question I have is: Why did it uniquely work in this field to build the entire platform and all the applications yourself? Any other time you get that pitch, you pass. “I’m not investing in that. That’s a stupid strategy.” Why did it work in this industry?

So, I sort of phrased this as a different question to myself, but I think it’s the same thing. I asked, “Why did Epic win?” But I think it’s the same question here. It just smacked me in the head: This is vertical-market software. This is the correct playbook to win in vertical-market software, and it is very different from building horizontal-market software.

In vertical-market software, you’re only serving 1 category of user. The deeper and deeper you can go into their business operations and solve their problems, the better. Whereas if you’re building Slack or Salesforce—let’s pick a horizontal, complex piece of software, like Microsoft Office—you need to be really, really careful about going too far into 1 customer segment.

You hear all the time from really great product people and engineers at the big horizontal technology companies that the surest way to design bad products is to listen to your customers. You internalize the feedback, and then you decide what the best feature would be based on their own experience or on collecting all the different experiences. I think that’s generally right for building horizontal software and products.

However, when you’re building vertical-market software and products, you absolutely want to listen to your customers. That is what Epic is really good at. You want to build exactly what they want, and you want to understand who your customer is. Your customer is the CIO and the CFO of the hospital system.

I now agree with you. My answer probably just layers on top of why it’s healthcare-specific, not just vertical-specific. It is all the carrots and sticks we were talking about earlier—all the sticks in this industry.

If your software is bad, they die. So, the right thing is 1 big integrated approach rather than any risk at all of discontinuity between different applications. That’s 1.

Two is all the compliance stuff. Again, if you have any data leak, whoa, it’s catastrophically bad because of HIPAA. If you don’t function correctly, your Medicare and Medicaid subsidies get crushed for violating Meaningful Use requirements. There are just all these sticks for any mistakes.

And so, in a situation where you can't make any mistakes, the single-vendor playbook is the right playbook, even if it's going to take 47 years to build the dominant company.

Yep. You have to write a lot of software, and it all has to work together really well.

Yep. And one other thing I was thinking about with this idea of Epic as a vertical-market software company: I don't think we've ever really covered any other vertical-market software companies on the show before.

No, I don't think we have either.

We sort of have. And I was thinking about why is that? I'm like, oh, well, that makes sense. Generally, vertical-market software companies are not going to get that big. You're limited by your vertical market. This is a rare case where, because the market they operate in is so big—the American healthcare market—and it's something we all use, and it's something that's so important, you actually can get a really valuable company built serving just one market.

Epic has got to be the most valuable vertical-market software company in the entire world. What else could possibly be up there? I started racking my brain. Bloomberg was the only one I could come up with that could plausibly maybe be as valuable.

We've got to do a Bloomberg episode someday, but that'll be our second vertical-market software company that we cover.

But then I was Googling and thinking about it, and I was like, people talk about Viva Systems.

Yeah, Veeva, which is also healthcare. They're probably the largest public company, but I was like, yep, they're definitely not worth as much as Epic.

And they are also operating in healthcare. They're not worth as much as Epic.

No, it's about a $20 billion market-cap company.

Oh, wow. Yeah, I don't think so. You know, the only other company I could come up with in the vertical-market software industry that's also in this kind of big league is Constellation Software.

Of course. Fan-favorite friend of ours, but they bundle together a bunch of different verticals.

Exactly. They're a roll-up. And I think some of their biggest companies are vertical-market healthcare and healthcare IT companies within Constellation.

Yep. Mark Leonard, I'm sure, is a big Epic fan.

Or maybe not a fan, because he probably operates some competitors.

Yeah, I'm sure he has respect for Judy.

I am sure. I mean, she's got a gold medal in the capitalism Olympics, so you got to respect it.

All right, that winds it down. Should we do some carve-outs?

Let's do carve-outs. Let's do something fun. I've got 2. I've got a regular carve-out, a today carve-out, and then I have a preemptive carve-out.

Great.

My today carve-out: a friend of mine turned me on to a very, very popular OG YouTube video that I had never seen. Ken Block, the rally-car driver, did a YouTube video back in 2012 in San Francisco. They shut down the streets in San Francisco, and he did this incredible, 10-minute-long rally-car video through the city.

It's like watching the old movie Bullitt with Steve McQueen and the iconic car chase, but with a 600-horsepower rally car driven by one of the best in the world. I can't believe I'd never seen this before. I'm like, “Oh, I know those blocks. I live there.” Such a cool video.

He sadly died recently in an accident, which is why he was back in the news. But I did some more research into him, and I knew nothing about him before. He was one of the 2 co-founders of DC Shoes, the skateboarding-shoe company. After they sold it to Quiksilver, he got into professional rally-car driving, became one of the best professional rally-car drivers, and filmed this whole series of amazing videos in cities. Super awesome.

I don't know anything you just said, so I have to go look it up, because none of that made any sense to me.

Right. It's a short, very impressive video. Very cool to watch. That's my main carve-out. And then my preemptive carve-out is that I am so excited for the Nintendo Switch 2.

Ah, if it ships—if they start taking preorders again.

Yeah, maybe by the time this comes out. I'm not worried about it hitting the ship date. I think they've got to sort out the U.S. price and everything, but it looks incredible.

Basically, everything from our Nintendo episodes that we hoped Nintendo would do with the successor system to the Switch, they're doing: full backward compatibility, maintaining the online accounts, and an incredible system. It's shipping with a new Mario Kart. It's going to be awesome in and of itself. I can't wait to have one for me.

Also, my older daughter—I have been waiting since the day she was born to play video games with her. By the time it comes out, I'm thinking she might finally be ready for Mario Kart, and I'm so excited.

Sweet. Oh, that'll be so fun.

All right, mine.

Go for it.

I just rewatched the movie Knives Out.

Oh, yeah. Good one.

Daniel Craig. Extremely fun. So good. It's just pure popcorn. It's visually stunning. Everyone loves a whodunit. Daniel Craig's awesome. The whole cast is great. If you haven't seen that, you'll really enjoy Movie Night.

My second one is—I must be the last person to discover this album—I've been listening to Brat by Charli XCX. It is awesome. I was watching Charli XCX at the Grammys, and ever since, I've just had the album on repeat. It's amazing. It'll transport you to another, much more fun world than the one you're currently sitting in front of your computer.

My last one is that ODESZA just released an album called Music to Refine To. It's remixes from Severance. It's good. Leave it on lo-fi in the background while you're doing work. I love ODESZA. I love Severance. It's my worlds colliding.

You're definitely not the last person, because don't worry, David, I will always be behind you in discovering music.

I'm grateful that all of us have you, too. Keep us up to date on music here. My music is frozen in amber at 2006. If you want to feel like you're at a much cooler party than you've been invited to in 20 years, listen to Brat.