Ep 33 - Max Marchione: Superpower, Hims, and the future of AI-powered medicine

Jonathan, Patrick, excited to be here.

So I think one of the most important questions in healthcare for us, for Hims, for many companies, is assuming healthcare were built today from the ground up, knowing what we know about the latest technology, diagnostics, therapeutics, peptides, frontier compounds, and knowing what we know about AI, what would healthcare look like? And I think fundamentally healthcare actually looks quite different, and it's structured quite differently. And I think the implication of advanced diagnostics, advanced therapeutics, and AI means that there are a few implications. I think one is there's a new consumer health front door that is likely an AI, and that people go to the algorithm before they actually go to their primary care doctor or Google or ChatGPT. So there's a new consumer health front door that is AI-driven and knows everything about you and knows everything about medicine. Like, we're interested in that.

And I'm sure Hims thinks a lot about that. Aura thinks about it, Apple thinks about it, Whoop thinks about it. That's one way healthcare fundamentally changes. One of the implications of the algorithm is that consumers disintermediate doctors. Again, HIMS is an early harbinger of that trend. Superpower is another example of that trend, and other companies showing that as well. In a world where we believe that the AI knows far more than any human can possibly know, we have to trust it. And as a result, we trust it almost to the exclusion of other things. And therefore, consumers who have access to the technology are empowered to disintermediate doctors. That started with ED drugs with HIMS, then it moved into slightly more complex compounds like GLP-1s, and I think we'll continue to see a larger and deeper part of the healthcare stack done in a way that starts with the consumer and what they want and the information they are learning online, rather than starting with doctors.

People laugh at me for saying this, but we'll see where we get to. I think another implication of healthcare in this paradigm is that you basically move to a world where therapeutics — the future of healthcare looks almost more like the magic pill than like better exercise, diet, and sleep. And we know this is true because 100 years from now, if I said magic pill or diet, exercise, sleep, which one is it? People would say magic pill. Now the implication is somewhere between now and 100 years from now that magic pill reality has to exist, and I contend it has to exist and will exist in the next 5 years, and is almost increasingly here. So I think one thing that any company has to think about — Hims thinks about it, we think about it — is how to actually educate someone on the therapeutics that are soon going to be available that can drive really powerful outcomes that look more like sci-fi than many of the things we've seen today and in this world. So I think that's one other thing that's interesting about where we're going from a future of health standpoint.

Now to come full circle, basically 3 years ago I was sitting there thinking about what would healthcare look like, and I had many of these beliefs. And 3 years ago — it was August 2022 — people laughed. They're like, 'AI? The chat? What do you mean? What about the art of medicine? These complex therapeutics? No, GLP-1s, they hardly work. They're just coming onto the scene. These other peptides are illegal. Like, come on. People are going to be harmed.' And now, two to three years later, we're starting to see some of these beliefs pan out. I had a health problem the other day and I went to an AI rather than a doctor. I was speaking at an event in New York and I asked, 'How many of you go to Google or ChatGPT when you have a health question first?' and like 90% of the crowd puts their hands up. So what we already implicitly know is we go to the algorithm before the human, that we disintermediate doctors, and we're seeing frontier compounds start to drive really large outcomes. One in eight Americans have used a GLP-1. So that was a fundamental belief in the early days of Superpower.

If I look at how we changed over time, we actually moved a little bit away from that belief. I think that everyone disagreed with me. Everyone said, 'No, this can't be true.' And I listened to them a little bit too much. So we started thinking more about how do we start to replicate high-end concierge medicine. We said there's a model of healthcare that costs $100,000. Let's start by making that 10,000, then 5,000, then 500, then 200. And that didn't really work, because the second you start with a $10,000 membership, you end up in service-led care. And you can always say we're going to automate it afterwards, but it's very hard. You don't have the forcing function to automate it. So one of the big changes we've made is actually to say let's start very well priced, $499 today, hopefully towards zero over time, and that acts as a forcing function to actually make this technology-led paradigm, this technology-led model of healthcare, reality. So that's one of the ways we've changed over time.

We will have humans in the loop for the foreseeable future, partly for safety, trust, and validating the output of the AI. We will also have humans in the loop when we go into more complex cases, and when we go into actual therapies, right? Your surgery is not going to be performed by AI anytime in the next 5 years across most surgeries. But I do think we get to a world where the AI knows so much about the latest healthcare and so much about human biology and your individual biology that it far outperforms anything that is possible for a human. And the point of the AI isn't to replicate the best of doctors. It's actually to do something that no human could ever do, which is connect dots in a way that's highly personalized and highly complex, and to make recommendations based on the way those dots are connected. And we get to a world where people aren't thinking 'I can't trust the AI,' they're thinking 'I can't trust the human.' And we've seen this in other industries when airplanes came along. If someone said to you, 'The AI is driving the airplane, no more humans,' everyone would have freaked out. 'No, no, we have to trust the humans.' But today, if you're midway on a flight and then you heard over the loudspeaker, 'The technology and the AI is no longer working. This plane is just going to be equipped by humans the rest of the trip,' you'd be freaking out. You'd be like, 'No, no, I want that AI. I really, really want that AI back.' And I think we'll see a similar thing in healthcare.

Well, just quickly on that point.

My problem with all of these arguments is they reason from the past. And the key way we need to think about this is that we have seen nothing yet in AI. And if we actually have full context and full information, it's inevitable that the AI is better. So that's kind of the challenge I have there. And the great irony is people go to the AI or Google before their doctor, and then they go to the doctor and they might get the doctor's opinion, and then they go back to Google and the AI because they don't fully trust the doctor, and during the consult the doctor is using an AI now.

Yeah, I think people don't care about AI. The airline doesn't say we're flown by AI to get you from New York to San Francisco. It's just like, no, I just want to get from San Francisco to New York. And what we see today is that people want to, one, understand everything going on inside their body. They want to make sure they don't have something going wrong. They want to make sure they don't have diseases. They want to get ahead of diseases. They want to optimize their health. So the first thing we sell is diagnostic testing: blood testing, gut microbiome, toxins, all sorts of other diagnostic modalities. People love that experience because they were never really able to get that through primary care. The blood tests part of the core membership is around five times more comprehensive. And people say, 'Oh, wow, this is the first time I've seen CRP for inflammation, or APOB or LPA for cardiovascular health, or this is the first time I've actually had my hormones tested and I didn't realize my hormones were completely out of balance.' That's the first thing.

The second thing we sell is once people see what's going wrong, the response is always, 'help me,' right? When you see a level that's really out of range, particularly if you've never seen it or tested it before, you want help. Where our job is to also help people and drive outcomes, and I think patients are quite good at sensing when they feel that quality is high and the reasoning of the recommendations are really strong, and they're good at understanding whether they see outcomes as well — you feel it, you feel more energetic, you sleep better. So the second thing we're selling is actual outcomes and helping someone fix many of the problems that they're seeing at the level of diagnostics.

The cancer at 40 can start at 20. The heart disease at 50 can start at 20. The Alzheimer's at 70 or 60 can start at 20. Many of these things start very early. Now, the health system today will tell you, no, you should start taking steps when the cancer cell starts growing or when your degeneration starts happening. I contend that it starts happening when you're a little baby, right? So we should be taking steps very early and very proactively. And part of our job is to help people understand how the things that are happening today when you're young could maybe potentially result in something downstream, and that there's research that supports how dysregulation today at the level of biology can result in things downstream. Now, the buying decision is a little bit different. We find that people who are older are thinking more in terms of, 'I really want to make sure nothing's wrong, I want to prevent disease, I'm thinking more in terms of the...'

Well, Hims is getting into prevention in 2024, and we're getting into prevention in 2024. In other words, markets move, different things are in vogue, different things matter, different modalities of diagnostics and therapeutics become available at different points in time. I contend that today the right wedge into building a new healthcare system is not starting with an ED pill, it's starting with something else, and it's my belief that that something else is diagnostics. If you look at what's required to deliver really high-quality care, it's understanding the person, and if you look at what's required to deliver care in an AI paradigm, it's having a lot of data on a person, because whilst a doctor spends 5 to 10 minutes with you and is not going to read through thousands of pages of data, the AI will, it can do that pretty quickly. So what we think about a lot is how do you create a new healthcare system? And that's the far-fetched version of things. And my belief is the starting point for that is understanding someone's biology, because when you do that, every subsequent recommendation is more accurate.

Well, this is the nature of opinions, right? We will have someone who looks at our site and says, 'Well, this is sci-fi. This is weird.' And we'll have another person who looks at our site and says, 'This is the coolest brand I've seen since like Apple or OG Nike.' We have someone who looks at Hims' site and says, 'This is really warm and friendly.' And someone who says, 'This feels kind of stale. It feels like a lot of other things. It feels like it's for my parents. The colors are kind of dull. There's no real emotion and brand and aspiration here.' So the first thing I'd say is there are opinions, right? And the nature of good brand is that they should be polarizing. Apple, the early Apple was for creatives. The early Nike was for athletes. They were intentionally polarizing. And I think one thing that's really interesting as a founder is that you'll get a million opinions and they're all conflicting and they're all different, and if you listen to all of them, you build the whole healthcare system and you build nothing brand. It doesn't really mean anything to anyone.

I think one thing we think about a lot is what does it mean to build a brand that is part of culture. There's no one really walking around wearing United T-shirts or HIMS hoodies. They're not really brands that are part of — there's no Nike of healthcare or Apple of healthcare. There's not much emotion, and I think that's because the first era of healthcare on the internet was very much about, well, we're scared of healthcare on the internet, let's be clean, warm, simple, let's build trust. I think the next era of health is people are comfortable with health on the internet, and they want emotion, they want like a culture or something to follow. We shipped what we called the world's healthiest hoodie. healthiesthoodie.com was the site, and it was like a cult sensation. They sold out in 10 minutes. People relisted them on Grailed and eBay for $2,000. I get every week someone traveling around the world and seeing one and sending me a photo. They're like, 'I just saw a Superpower hoodie.' I have friends who wear them to events and have people approach them. Like, that's a lifestyle brand, right?

Now, could we lean more into humanity and less away from technology? 100%, and we should. But the other thing that's true is if you look at any brand in year 1 and 2, and then you look at it in year 5 and 10, they also look different. So there are a few of my thoughts here.

I'll just throw in, Google is a great data-driven brand, and Apple's a fabulous non-data-driven brand. So I'm being facetious, but I do agree. I do think that understanding the way people perceive things is really important, and then it's a matter of filtering that signal through the noise. But sorry, Patrick, go on.

Yeah, I think good branding is polarizing. I think HIMS did a very good job of this in the early days. I think good branding is opinionated, and there's always the tension between knowing whether the opinion is correct, because the opinion can also be wrong, which is the nature of opinionation.

I really believe that the main way people are going to access care is via an AI doctor within a super app on their phone, and that's going to be the first place they turn to access care. And that doctor will know everything about them — all their omic data, their clinical data, their personal history, their wearable data. It'll know everything about medicine, not just the medicine on WebMD and Healthline and in papers, but also the latent knowledge trapped in the brains of doctors. That's where I think a lot of the best medical knowledge still is trapped. And that AI doctor will actually be able to take action, with humans in the loop, with validation, but it'll be able to take action. It will know how to order diagnostic tests, prescribe, refer out, in a way that is far more personalized and can drive outcomes far more effectively. Partially because the recommendations are precise and personalized, and partially because all the recommendations will be able to be fulfilled within an ecosystem, right? And that reduces friction, and through bundling also reduces cost. And we haven't seen that in healthcare. We certainly haven't built all of that yet. But it's my belief that that's where we'll get to in healthcare.

I don't think about acquired or public as goals. I think about goals as: do 150 million Americans have a Superpower membership, the same way as 150 million Americans have an Amazon Prime membership? Are people coming to us first, or they go to Google or ChatGPT first? Are we able to drive outcomes that look more like sci-fi than today's reality? Or are we still constrained to the level of outcomes that we see today? Those are the things that we think about. I'm indifferent to whether we're public or private. I hope we're not acquired. That's probably a failure case for us. Been 10 years. So we want to be building this for at least 30, 40, 50 years, because I think nothing great comes over 10-year time horizons.

We have no plans to be acquired. If we sold the company for a billion dollars, we've failed. So no, no plans for that. I have a lot of respect for Hims. As I think about what they do a lot, I think that...

They are making bets that people don't quite understand and that are quite opinionated. I do think Andrew has a sense or an intuition for where the world is going, and I think he simultaneously balances that with understanding where customers are at today and the thing that can be monetized today. No, no plans for us to be acquired by them.

I was speaking about two things. One is the learnings from how HIMS executed on the way up, which was a story of being very focused. They started with ED and they didn't move into a million things right away. That's challenging. Founders' egos often get in the way of that. It's embarrassing for everyone to be saying to Andrew, 'You're just a rapper, you're a pill mill, you're just selling dick pills.' That's embarrassing, right? But they stayed very focused. They continued betting on the core before moving to the incremental. The second thing was the bets they're making going forward, largely around vertical integration and around personalized compounds, particularly peptides, which could be some of the more promising compounds that emerge. My sense is we see two things happen in healthcare. One is that real-world endpoints — so the things you need to actually change your health: diagnostics, supplements, pharmaceuticals — get increasingly commoditized. And in a world where they're increasingly commoditized, you want to do one of two things. You want to either be vertically integrated so you can win on pricing, like Costco kind of style, and we kind of see that with HIMS; they're increasingly vertically integrated. Or the other thing you want to do in a world where these real-world endpoints get commoditized is push towards personalization or some sort of uniqueness that you can get through data, and HIMS is also doing that. So there's a world where three, four, five years from now every single person is getting personalized compounds through HIMS, and I think that's part of where the world goes: personalized therapeutics for everyone. So that's kind of the first category of things — vertical integration because the industry will become commoditized and you need to compete on price, with personalization to stand out.

What's true at the same time is the move towards peptides, which I believe are likely going to be the specific compounds that are the most effective. And not many companies move towards them because today they're category two or category three; they're banned, they're not allowed, people think they're pseudoscience. But I think HIMS is making a bet in that direction because they've seen the anecdotal evidence, and before there's hardcore clinical data, there's anecdote. And some people dismiss anecdote. I say that's the first step in the scientific method.

I think there are a few parts to the AI story. One is, talking about AI is a great way to raise money and keep your investors happy for HIMS. Two is, AI will likely replace low-acuity primary care and act as a front door, and there's totally a world where HIMS is trying to do that. It tried to move into primary care several years ago, didn't really work. Managing a big clinician network is a pain. Doing a lot of that with AI, particularly where you can initiate the conversation with AI but then be able to actually have actions taken — you can actually have something prescribed to you — is interesting. So the second thing is, I can see them moving towards an AI chat with agency or action on the end of it. The third thing that's true of AI is it's good for just automating and improving everything. I don't think we should call this AI; I think we should just call this better technology, that people like to put the AI label on it. The fourth thing that is more far-fetched, and I don't think it's quite the zone of genius for HIMS, but we'll see if they can make anything of it, is the more far-fetched use cases of AI, either in terms of personalized therapeutics or connecting the dots across diagnostics and then potentially the mapping between the two. HIMS doesn't have a lot of useful diagnostic data, but with their acquisition of the lab, there's a world where they start to collect multiomic data. They certainly have some longitudinal clinical data. They might also integrate continuous wearable data. When you have those three data sets and a lot of customers, AI presumably can make new connections and discover things and realize things that we haven't yet realized in a human-led paradigm. So that's maybe the fourth, more far-fetched application of AI.

I suspect they want to just scale their own business, their current core, into new markets, and they don't want to set that up from scratch. They have a reasonably low cost of capital now as a public company. They're very good at operations and brand and marketing. So my bet would be: look at the largest HIMS copycats in every market and pick them. The Australian/UK one that's in several other markets is Eucalyptus. So maybe Eucalyptus. My understanding is they're probably the largest outside of the US and China, but there might be other large ones as well.

I presume the ones that are large, that have somewhat similar healthcare systems. So the UK is an obvious contender, Australia is a contender, Canada is a contender. I think the Asian markets are a little bit more complex; they're large but likely more dissimilar to the US, and the buying patterns are different. China is always tricky because it's China. India is a little bit less tricky and has a lot of people. Brazil is obviously large, and we see a lot of 'this company exists in the US, copycat in Brazil,' so I could see them expanding to somewhere like Brazil as well.

Yeah, sure. So Germany could be on the cards. I know Eucalyptus operates there. I don't have a sense of the market in France, Spain, Italy. If they're in the UK and Germany, I can see those markets being interesting. Yeah, my sense is HIMS is going to want to acquire companies that are very similar to HIMS — so quite vertically integrated, similar SKU offering, very large number of members. I think the inorganic growth makes a lot of sense when you're HIMS. They likely have stronger retention curves than many of the new companies in the space, which means that they can pay more per member, right? LTV per customer is higher. And I think that it's less likely that acquisition is for some sort of IP acquisition or expanding from the core. I suspect it's likely just doubling down on the core and expediting the path to new markets and growing topline revenue.