Can You Reverse Decades of Drinking Damage? Perhaps. - EP 75 Jacob Kimmel Is Back

I come this week with terrific news. It is possible that the hangover has been cured and that, uh, liver damage from years of drinking might be cured. So if you are an imbiber or no one or an alcoholic, rejoice. Possibly. Our guest is New Limits co-founder and president Jacob Kimmel. He is a repeat guest. This episode really is about the cutting edge of longevity technology. NuLimit is one of the most fascinating companies working in this area. Jacob has been on the podcast before. He's our very rare repeat guest, and we have shot a YouTube video at NuLimit's laboratory, which not enough of you have watched and you should watch. They— NuLimit's doing just some of the most pioneering, groundbreaking, groundbreaking work in the longevity field. And, and Jacob is eloquent. He's a genius. He's brilliant. This is why we've had him on again. What the company has seen is that they can get mice drunk and then give them their therapy. And there are old drunk mice that are waking up from their big night out, and then they're running around like young mice who are healthy. And it's incredible. We've got some video to show you in this podcast of these old drunk mice. Being unfazed by booze. And more importantly, it appears that this therapy is reversing severe liver damage that would have accumulated over time. We must pump the brakes, of course, because this is in mice and not in humans. Although Noolimit is going to head toward a clinical trial very shortly, which we talk about in the episode with this new therapy. It's incredibly encouraging. Remarkable, really. And, and Jacob is fascinating to listen to. Please leave us a review or a like wherever you listen to this podcast so more people can know about the end of the hangover and, uh, and reversing liver damage. I am Ashlee Vance. This is Core Memory. You're gonna love this episode.

Off we go.

Jacob, you are— you're a rarity. You're a repeat guest.

Privilege to do so.

It's— it's— it is our fortune. Um, I don't know, this will air before our Max Hodak episode airs, I think, but Max is our only other—

if I'm second only to Max, all is right with the universe.

So we love talking to you. Uh, we love watching what New Limit is doing, which is the purpose of this particular podcast in some ways, is there's a newsy angle to some of this that we're going to get to. You guys have— well, I'm going to tease it. I'm going to tease it. You guys have broken ground on the hangover and on liver function. So many people, you're going to have to stay tuned to figure out exactly what has happened.

But I feel like you're giving away the whole game, man. Well, I don't know. It's a tease.

It's a tease. It's a tease. You know, we'll have to find out what you've done with The Hangover. And so, but for people, if you checked out our— you can check out our previous episode, we really get into the, the heart of New Limit, the history of New Limit, what you guys are doing. But let's assume not everybody's going to do that. And, and this is a field People know the longevity field, but they don't know the specifics of what you guys are doing. Yeah, you want to take a crack at it instead of me? Just the, the basic overview.

I'll try my best. So we're working on developing medicines for aging, and the technology we're working with is called epigenetic reprogramming, which is way too many syllables for a pretty simple concept. Um, the basic premise is that all the cells in your body have the same DNA code, and each cell actually uses a different set of genes from that DNA code. This is how your eyeball has a different function than your kidney. Despite the fact that you have the same genome everywhere. Now, this is controlled by a set of chemical marks on top of your DNA and the proteins your DNA is wrapped around like string upon a spool. And those chemical marks are called epigenetic marks, and collectively they're known as the epigenome. So basically you can think of these kind of like the control flow in software. When a user uses an application, not every line of code runs at the same time in the same way that every cell doesn't express every gene. Rather, it's controlled by these epigenetic marks. Unfortunately, as we get older, these marks get degraded and cells stop using the right genes at the right time. They aren't able to keep you healthy and you become way more susceptible to disease. So what we're working on doing is trying to make medicines that activate very special groups of genes called transcription factors that can actually rewrite these marks back to the way they were when you were young. So the hope is if we find the right group of these transcription factors, we can take any old cell in your body and actually reprogram it so that it acts just like its young self did before, helps keep you healthy longer or treat diseases that have emerged as you've gotten older.

It's this fascinating area of science. People who listen to our show a lot, you'll have to forgive me. We're going to do a little history on this because I just want everyone to be up to speed. So the transcription factors you're talking about, there was a Japanese scientist, Yamanaka, who, who found these 4 particular transcription factors, these proteins that were very effective at taking— doing exactly what you said. You take an old skin cell off of a mouse, hit it with these 4 proteins, take it back to like this stem cell blank state, um, blank slate type state. And then, and then we found, wow, okay, maybe you don't have to go all the way back. We can take you a certain way back so that your skin cell is still a skin cell instead of this blank slate, but it's a, it's a youthful one. And, and so he found these 4 proteins. Then there's been this idea, okay, well, he sort of stumbled upon these. Maybe there's other ones, there's other combinations. And you guys, part of what you've been doing is this big hunt to find, are there other transcription factors? And let's just go test it out in this massive AI plus wet lab experimentation to see.

Yeah, absolutely. I'm going to have to bring you on as a branding consultant because you're explaining it all so wonderfully. Exactly as you said, one way I like to think about this is that Shinya Yamanaka, who won the Nobel in 2012 Ashlee Vance for this work demonstrated you could turn on just 4 of these genes and reverse both the age and the type of any old cell at the same time. So this is an amazing scientific revelation because prior to this, no one thought it was actually possible to reverse the age of a cell. But it turns out you could take any old cell, reprogram it back to a stem cell. Then when you make new adult cells from that stem cell, they act like young ones regardless of how old the cell was to start with. In the most extreme form, you can actually take that new young stem cell and give rise to a whole animal that has a normal lifespan, and you can run that process on a loop. So there's actually a group in Japan now who's done this 56 times. They published a new paper last time you and I chatted, Ashlee Vance. The latest result was 13 times, and they kept going. It just keeps working.

Why are they keeping— why do they keep going?

Why do they keep going? Because it's amazing. Because you want to demonstrate that there's not as an upper limit based on the epigenome in terms of how many times you can repeat this process.

So they're— I mean, you're talking about literally you could take a skin cell, take it all the way back, and then grow a whole new animal.

Exactly. Take any, any adult cell, reprogram it back to a young stem cell, and give rise to a whole young animal. And so one thing you can infer from this is that while many things change with age— mutations can accumulate in the DNA, you can get changes in nuclear pore proteins that don't turn over— I by no means would claim the epigenome is the only thing that changes as we age, but does tell you that epireprogramming is sufficient to actually restore youthful function in old cells. So this is a really amazing result we've known about for about a decade now. The challenge is that what Yamanaka discovered again was you could change cell age at the same time you change cell type. And it turns out therapeutically, that's often not really what you're going for. If I actually turn any of your old cells in your body back into young stem cells, they would form tumors. That would be bad. When you do this in animals, you can actually unfortunately cause mortality within 5 days. It can be really severe negative consequences. What you really want to do is just half of that equation. You just want to make the old cells young again. You don't want to change their type. I don't want your liver to turn into a bag of stem cells. I don't want your skin to start getting new stem cell growths. I just want your liver cells to be young liver cells, your skin cells to be young skin cells. And so when we started the company about 4 years ago, it was unknown whether or not you could actually decouple or disentangle these two phenomena.

Like we knew how to do it, but it's this fine control that you're talking about. The— there's a line between we can do this, but then it spins out of control and creates these teratomas, these cancerous cells, and, and like, how, how much can we really hold on here and guide this?

Yeah, exactly. So there were groups of people who had found, hey, we know these Yamanaka factor proteins— he rightfully has an eponymous name upon the factors— um, and if we dose them just right, we give just enough, we can go back to a state where we reprogram cell age and type, but then type kind of recovers. The cell remembers what type it is over time, but the changes in age might be a bit more static. And so folks figured you could do this in a research lab, but it turns out that the doses that one needs to dial in are so narrow, it's very unlikely this would be a therapeutic. To give you a sense of just how narrow these dose windows might be, the difference between a dose that was beneficial and helped make this animal have young tissues that were more functional and the dose that killed the animal in 5 days was less than 2.

Okay.

So imagine now you have a drug like Tylenol on the shelf and it's like, take one pill, you feel slightly better. Take two, you die in 5 days. It's like, it's not a very attractive sales pitch. So our hypothesis going in was we could find other groups of these factors, just as you had laid out, that would allow you to actually reprogram a cell's age while leaving its type alone so that dosing wasn't so much of a concern here. Ideally, we're only moving along that cell age axis. And when we started, this was really just that hypothesis. We didn't have a lot of data that this was possible. And so we built, as you alluded to, technology that's allowed us to test thousands of fold more combinations of these transcription factor genes than, as far as we know, all the other groups in the world combined. And then we've also built AI models that are trained on top of that data that help predict for us what will a new combination of these factors that we've never tested in the real world do so that every time we're spending budget experimentally, every time we're going to the lab and doing hard experiments, we're guided by all of the data that we've collected before in a very intelligent and quantitative way. Just a few years back, this wouldn't really have been possible, both because running those experiments wasn't actually as scalable as it is now, but two, because I'm not nearly as clever as our AI systems are today, and I'm much worse at walking up to a whiteboard and trying to write down what the right experiments are to do in which order, and these systems are far superior. So we think those two technologies and the interplay between these high-throughput experiments and these modern AI systems have allowed us to find really compelling, what we call payloads, groups of transcription transcription factors that can actually reverse the age of hepatocytes, T cells. And now we have a few that we think might even be working in endothelial cells in a way that no one had demonstrated previously. And so that's really the basis of our therapeutic program.

Yeah, it's incredible. And, and, well, maybe I'm fast forwarding, but even when you guys started, I mean, what year was it?

It was, uh, it was 2022. It took us until '23 to actually like have a lab, be doing meaningful experiments. Yes.

Yeah. But then even in just— so you were already making this bet on like AI was here and it was doing interesting things and data mining and all that was already there. But I mean, even just in these last 3 years, we see the rate of progress. So that must have just— I guess you were banking on that happening.

But it—

but I would assume it seems like AI has really hit coding and biotech first the hardest. I mean, so this must have been like quite beneficial to you guys.

Yeah, absolutely. Again, I don't think we would have been as successful in discovering the prototype medicines, the very early-stage therapeutics we have today without these systems. And something that's been really amazing to watch is that our models are built on top of some of the intelligence that's encoded in frontier systems, both in natural language and in the languages of life, really. Models trained on protein sequences and DNA sequences. When these models improve in performance, we get sort of a free boost. Our models start from a smarter place because the model they're learning on top of actually is more intelligent. That means that we're able to achieve higher performance as a result. So we get to sort of draft in these nice tailwinds. Every time our friends at the frontier labs make some innovations, we get some free benefits that just emerge for us alongside.

Is your model called Ambrosia or your software?

Yeah, we have a model we call Ambrosia. One of our scientists, Lukas Suning, loves Greek mythology, and so he names a lot of our software after various Greek mythological figures, and ambrosia is the food of the gods. So, you know, we enjoy that one.

And then it has multiple foundation models at its core, or you have a favorite and you're slotting in and out, or?

Yeah, we found there's actually benefit to using embeddings from multiple systems. So the way this model works is that we basically want to show the model a group of transcription factors and a starting place from an old cell. Hey, here's an old hepatocyte. Here's the genes this old hepatocyte might use. Now, if I turn these transcription factors on the way I might with a drug, What will happen to this old cell? Will it look more like a young one? Will it act like a young one? Will it change its cell type in a way that's dangerous that I don't even want to bother with? And in order for that task to be achieved, we need to first represent those transcription factors in a way the model can understand, which means we basically need to turn them into some list of numbers, what in machine learning we would call an embedding, to be fanciful about it. And so we derive these embeddings not only from natural language. We can do things like take a long block of text that's a summary of lots of papers that humans have written about this gene and use a language language model to summarize that into a vector. But we also can look at the actual underlying languages of life, the protein sequence of that transcription factor, which wonderfully is known. We've sequenced the human genome. We can then even look at things like the DNA sequences that this transcription factor might bind, because other researchers have figured out where in the genome do these transcription factors tend to scan and eventually lock on, maybe which genes do we think they might be talking to in terms of activating and deactivating. So our model then goes from each of those embeddings. And that's just a starting place. It kind of gives our model a notion of what is a transcription factor to begin with, which are different, which are similar to one another. And then only with the data we've collected in the lab, which is the data where we've actually gone and we've turned on tens of thousands of groups of these transcription factor genes, we've measured what happens in various old cells from different types. We ask our model to go from those embeddings to predict the outcomes we ever actually measured in the lab. And then we can do the inverse. Once we've trained that model and it's now incredibly performant, it explains more than half of the variation in the experiments we run, we can turn it around and say, here's the cell state I want. Here's what a young hepatocyte looks like. What transcription factor sets should I put in to try and get to my goal? And so one way of framing this formally is that we're sampling from the posterior of a model once we've actually trained it. So it's the system we use today. And each of those foundation models that we use as inputs really just comes in at that very first stage. How should my model even consider which transcription factors are similar and which are very different?

Okay. All right, what do we do at Core Memory? We cover innovative, fast-moving, forward-thinking companies, which is why Core Memory is sponsored by Brex, because Brex is the intelligent finance platform for many of these companies. 30,000 companies from startups to the world's largest corporations rely on Brex's technology for their finances. They've got smart corporate cards, high-yield business banking and expense automation tools that are fantastic. I hate doing my expenses, and Brex's AI software runs right through those expenses, figures out where we're spending money, and takes care of so much stuff for you so you don't have to waste your time on it yourself. Go to brex.com/corememory to learn more and just, you know, get with the program. Let's get going. Let's get out of this Archaic finance software and move toward the future. Core Memory and Brex. You would also make this video if you want to see how this process works. We went, we went to the New Limit Labs. We got the, we hung out with the mice and people could actually get a very vivid visual representation of this. For some reason on YouTube, every time we do biotech videos, they don't do nearly as well. It hurts my soul too, man. It's like the. I just— I don't know if we— the algorithm, it loves us when we're doing manufacturing and hardware. And then every time we go into biotech land where I think we're doing some of the most interesting stories that frankly other people are missing, the algorithm is rejecting us for the moment.

But we'll have to talk to our YouTube friends. You know who you can blame for the lack of views? Do you remember I was actually your gaffer when you came by? Yeah, yeah, yeah. It's my fault.

Okay, so there's a million directions we could go. One thing I was going to call out is that the mainstream media had ignored a lot of this. This interest in transcription factors is enormous. You guys are doing work here. Altos Labs, which allegedly has $3 billion or so in funding, is supposed to be doing work in somewhere in this, this realm, although they're a bit secretive. We don't know Maybe, you know, but I don't know all the ins and outs of what they're approaching. And then, and then RetroBio, which we've written about and filmed with, as they've been in the space and then some others. But lots of interest here in Japan where, where Yamanaka is from, you know, obviously kind of home, home field there. They're deeply interested in this area. And then there's, there's myriad startups dabbling in all kinds of very specific areas here. And then finally, last week, I think— well, it depends when this comes out— this month, The New York Times did a, did a like a giant story that they seem to have gotten some access to Altos and, and, and for the first time at least, we're hitting on some of these, these topics. Yeah. So this is, it's bubbling up.

It is. It's finally finding its way to the frontier. You know, one way I often think about this is my career has kind of straddled research in machine learning and research in molecular biology. I remember in 2012 through roughly 2014, people talk about the AlexNet moment in ML when it was very clear that deep learning systems were going to outperform certain types of classical machine learning systems. When I say it was very clear, it was very clear for the nerds hanging out at NeurIPS, and it did used to only be nerds. The parties were not that good. But it still hadn't actually percolated out into the broader ecosystem. I don't think many people in the general population generation would have seen the very inklings of modern AI systems in some of these early cat classifiers that people were training at the time. And I think epigenetic reprogramming in terms of its legibility, how easy it is for someone to understand who's outside biotech, outside the field, is kind of at that AlexNet stage. It's clear to people inside the field who spend time on the science that it's one of the most interesting opportunities we have in biology and in healthcare more broadly. But it's still pretty technical, and I think we're still doing the work of telling our own story in the right way that can make it accessible to a broader group of people. But my genuine belief is that if we zoom forward into the future, once we have products that are actually out in the world, the same way that ChatGPT brought deep learning systems to the broader population, once there's a medicine that provides benefit for most people over a certain age in their daily lives, you'll have the same amount of attention going to the science that you currently do to artificial intelligence systems.

Yeah, Well, and yes, I mean, the fundamentals of this are miraculous in already what we've seen and then the potential is extraordinary. But I think you're exactly right, which is there's also skepticism even among my friends in the longevity field who say, look, you can do these interesting things, but it's always going to be too hard to control this. It's going to be— and maybe we'll address this somewhat when we get to this, the newsier bit of the show, but the, this idea that you start reprogramming things and then it has this weird effect on your, your body as well, and exactly how all these systems work together. I, you know, I, I, um, I get so into this topic that it seems like maybe the most interesting thing, and then, and then I do have friends who are skeptical and they always Kind of gives me just a little bit of caution. Yeah, because they're smart people.

Yeah, totally. I think especially in therapeutics, we all need to be honest with ourselves that most medicines we put into patients don't work. And so you could be correct 90% of the time if you guess for any arbitrary drug that it won't work, even if you don't look at it first. And so I do think the skepticism is warranted in this domain. And I think hopefully what we can contribute to this field is that I think currently a lot of the skepticism is based on some of these early research results where what folks are trying to do is take this really foundational discovery Yamanaka had made in a pretty different context, trying to create stem cells on a plastic dish with these four factors, and translate that into a medicine where there are very real limitations. And I think we're on the frontier, as far as we know, of actually demonstrating that this technology is broader than just that one combination of genes that was discovered in that one lab in that one experiment. That actually this more general idea that you could probably reprogram the epigenome in any given cell type back to a more youthful state in a way that would be beneficial is possible, and that you can actually find those groups of transcription factors. And our hope would be that once we've demonstrated enough of these discoveries and we've gotten further along in the therapeutic process, when we can show folks results that are not just in our lab but in humans out in the real world, that we'll have the data we need to start to dissolve some of that skepticism in a real way.

No, I'm sure. Okay, well, let's— so let's, let's go into the news then just a little bit, and then we'll come back. Back to some big picture questions around all this, but maybe this will help people have something to hang on to. One part of this is money. You guys had— you raised— you did a funding round last year and then some other people gave you some more money after that because they were excited. And then, and then now, I mean, you— yeah. Do you want to tell us the— what's going on?

Sure.

More than happy to.

I love that you get right to the meat of things.

Tell me about your money.

Tell me about our money. We are very grateful grateful to be partnering with a number of wonderful investment partners. In 2025, we had the privilege to do a Series B round. Actually, it was probably shortly after the last time you and I had a chance to chat. That was led by Kleiner Perkins, as well as a number of other really amazing folks. And following that, we raised a convertible note. In total, that was about $175 million in capital. And so with that capital, we actually were in a position to carry forward One Medicine into clinical trials. And so we would have been in a position to demonstrate what we think of as the first real, tractable, scalable reprogramming medicine in humans. It's a really privileged position to be in. Things were going so well, actually, that initially when we raised that, that tranche of capital last year, we thought it would still be several years until we were able to get into the clinic. And the results just came much more quickly than we expected. And so we ended up moving up that clinical trial timeline. So our hope is now to actually dose a patient next year. So in 2027. And as a human patient, human patient, not mouse patient. We dose a lot of mouse patients, but no, these would be human patients. And to our mind, this would be one of the first times we're actually testing this hypothesis of can epigenetic reprogramming provide benefit to very broad groups of people? And alongside some of those new pieces of data, we had the chance then to raise another round of financing. So this is from some really wonderful folks like Founders Fund and Thrive Capital. And that capital actually will allow us not to just go forward with one medicine, which always carries with it some risk, but actually to diversify and carry a portfolio of these products into the clinic so that we can ask the question in a really general way, can this technology provide the benefit for humans that we're hypothesizing? And that won't just be for our first program, which we haven't mentioned, but is really focused on restoring function in the liver. So making old hepatocytes, Greek for liver cells, function the same way young ones do, being resilient to damage, regenerating once the damage has emerged. But also enabling some of our follow-on programs to move forward as well. We have a new program where we're focused on restoring function in the endothelium. So these are cells that line your veins and your arteries. They're really in every single tissue in your body. So the sort of big idea here is if you could restore youthful function in your blood vessels, it might provide benefit in so many places you don't even expect. Everything from restoring function in your kidney to actually preventing cognitive decline and retaining mental acuity later in So we'll be able to carry that program forward as well as another program we have focused on T cells in the immune system to dampen inflammatory disease. And alongside that, start others. So I think we feel really grateful to be in a position where we can test these hypotheses at a scale that's sufficient to give us really definitive answers. And based on the data we've seen thus far, we feel very optimistic those answers will be positive.

And so, and so this, this next batch of money, how much was it again?

So it's $435 million on top of what we raised last year.

$435 million on top of what you raised. Yeah. Okay. All right. You guys are not fucking around.

We don't try to. We try to only do things that are serious.

All right. There's a lot to get into. Let's do the liver program for a minute. So since I've been coming by your office, well, not the very first time, but I think the more recent times you've been—

The second time you came by, we were doing livers. The first time it was just blood.

Right. So this one was jumping out as is— is for some reason this hit first, or this just grabbed your attention first, that the liver was working?

Yeah, absolutely. So the program that we've worked on in hepatocytes moved incredibly quickly, part of the reason being that there are maybe 3 core challenges you have to solve whenever you develop any drug. The first is you have to figure out what the heck you're trying to do in biology. So often we talk about this as discovering targets, figuring out which genes you want to turn on or which genes you want to turn Classically, for most drugs, you're just turning one gene on or off at a time, which is kind of funny that any of this works at all because the body is incredibly complicated. You have 20,000 genes. How the heck does just turning one gene on or off actually, actually do much? So it's a miracle we have medicines. In our case, when we talk about targets, we talk about combinations of these genes. So we first have to figure out what's the group of those that we want to modify. Then you have to build a molecule that actually performs that action in the body. So let's say we figure out which transcription factors we want to turn on. We now have to figure out how do we actually turn those on in the right cells cells inside an animal. The final thing you need to do is then develop a clinical study that actually allows you to show this works in humans, this provides benefit. So hepatocytes have this really wonderful property, which is that the second step of that process, which is actually making a medicine, getting the drug into the right cells, has been demonstrated by many other firms. So the way our drugs work is we actually have RNAs which encode each of these transcription factor proteins, and then those are wrapped in a bubble of fat called a lipid nanoparticle, and that carries these RNAs into the cells, kind of like a Trojan horse. The cells take them up thinking they're energy, then the RNAs get out into the cytosol, and you're off to the races. Your transcription factors get translated, and they go to the nucleus and change which genes are active. In the context of hepatocytes, all of the companies that have developed gene editing tools where they actually deliver gene editors that similarly go into hepatocytes and then translocate to the nucleus and then try and make an edit, they use the exact same delivery technology. So there's already tools on the shelf that have been in humans now for more than a decade that we know are safe, we know are effective, that allow you to go really quickly from a list of transcription factors on a sheet of paper, or more aptly, in a terminal window on your computer screen, and then turn that into a drug that you can put into an animal. And so that's allowed us to test these hypotheses with actual drug-like molecules really quickly. And so one of those hypotheses that we actually discovered in just one of our first screens ended up yielding what is now our most advanced asset that we found is able to restore youthful resilience in hepatocytes, make them more resilient to damage from things like like high-fat diets or alcohol, and also allow them to regenerate that damage once it's happened. There's also some ancillary benefits we've discovered where, for instance, you can actually modify pretty broad behaviors that are related to liver health. In young animals, when you expose them to a damaging diet like alcohol, they're generally fine. They behave normally like young mice do. They run around, they look for their next adventure.

Party on, man.

Yeah, party on. In old mice, they actually become terribly sedated. They flip on their backs, they sleep for about 8 to 12 hours at a time.

Sounds right. Yeah, this is all checking out.

Yeah, the human data replicated in a few animal studies. We have not heard anyone contest this thus far. And so internally, people joke this is a mouse hangover. And we actually find even that this medicine is able to prevent that hangover behavior in these animals for weeks after these animals have first been dosed.

Okay. So they— the older mice were drunk, but they're not getting a hangover.

They got alcohol, but they don't get the hangover.

And so they were inebriated. But there's no ill effects. They're moving around the next—

or they move around in a way I can't quite ask them. You know, they haven't spoken English yet. That's another medicine for another day. But behaviorally, it is totally binary. You know, they're either passed out asleep or they're running around looking for an adventure. And after we reprogram the liver to function like a young one, they're just as resilient as the young mice are. Meaning that just like you or I when we were 18, we don't suffer negative effects the morning after you have a couple of glasses of wine.

Yeah. So we don't know, like, if there's there is a little bit of a headache back there.

We can't— we can't quite ask, but they seem okay. What I'll say is that the contrast between the ones that are treated and the ones that aren't treated, very obvious. You definitely want to be the guy who got the drug.

And even in the— yeah, this is in the old ones. And then I think we could probably run it. You showed me a video. I'm sure we can run some of this.

Yeah, it's easier to look at it than it is for me to describe.

I mean, it was very obvious. The suffering, sad, hungover mice were quite sedate. It's very funny. Yeah.

To our knowledge, we're one of the first groups that's actually discovered that this phenotype even exists. Giving alcohol to animals is actually pretty still, it's still a pretty rare field of study. The disease we're trying to model in this case is alcohol-related liver disease, which is the number one cause of death from liver disease in the US. It's 30,000 deaths a year. About 2% of the US population has some version of this disease. It's a really terrible burden. There aren't any treatments. So when people get diagnosed and they're severe, it's often a death sentence. So it's as bad as having certain types of cancers. And despite this, not many people actually study it. If you actually go and look at the literature, the weight of inquiry around what causes this disease and how we might treat it is far smaller than in many even subdomains of oncology. And so one of the model systems people use here is, you know, thankfully quite similar to how this disease manifests in humans. You can give animals alcohol. But as far as we can tell, no one's really done this in old mice before because old mice are expensive. So why would you have done that? And it's kind of a weird thing to try. And so actually, one of our research associates in the lab was the first person to tell me that this even happened. You know, he came up from the mouse room and he told me like, "Ah, the old ones are asleep and the young ones are awake." And I was like, "What are you talking about? It's like not how this works. We go down to the mouse room. It's totally binary." And so as far as we know, we're one of the first groups to actually demonstrate that not only can we rescue this phenotype, but that it even exists, that hangovers actually persist in mice.

But to your researcher, it was just blindingly obvious because because their behaviors were so stark.

It's so different. We didn't need to measure anything. It's probably the only time in my career where I've had a phenotype that is so obvious you don't need a machine to measure it and you don't need to even do any statistics. It is just a binary. They're either passed out or they're not.

So the big picture is not really to cure the hangover, but it's to restore— I mean, what you're saying is if you've had severe liver damage, you can walk it back.

That's exactly the case. And one interesting thing about this medicine—

This is insane, right? I mean, this is a big deal.

We would like to think so. So I'm glad. I'm glad you're as stoked as I am about this. When I tell people about fixing livers, they're often like, what does my liver do again? So I'm glad you appreciate it.

Well, no, I mean, so you talk— I mean, obviously alcoholism is a big problem. And then you're talking about somebody— I haven't heard you say the word cirrhosis yet. I mean, essentially that's what you're talking about, right?

So when patients get very severe, then they have cirrhosis. But actually a much broader population than you would think has liver disease that hasn't yet progressed. To cirrhosis, but nonetheless has negative impacts on their well-being, how they feel, makes them more susceptible to other types of diseases, even cancer. Having liver diseases, even in mild forms, increases your susceptibility to cancer. So cirrhosis is the most severe patients. But actually, when you think about the broader spectrum of disease, again, it's about 2% of the US population just from alcohol has some liver disease. That's not even accounting for, for high fat. So there's a lot of folks. It's not, not even just severe alcoholism that causes visit.

All right, pod friends, I am here to tell you about SendCutSend. They are a manufacturing phenomenon. Whether you're working on your own car or you're a giant automotive company, a rocket maker, if you need metal cut, machine, or bent, you're going to SendCutSend. They'll do it for you right, and they will do it for you fast. Go to sendcutsend.com/corememory for a 15% discount on whatever you're trying to make. We love Senketsen. Go visit them today. So if you try to— if you try to measure this lasting effect of alcohol, so you would have to give the mice alcohol for quite a while, right? I mean, to replicate an old mouse that's been drinking for— yeah, like not just for one day, not just like binging for one moment.

The model system we use, we've adapted from the NIH. They have a center actually on alcohol abuse, and so this model known as the NIAAA model for that reason. And what they've discovered is that exactly as you allude to, it's not like you just give them one big drink of alcohol and suddenly you have a disease that's reflective of what's happening in humans. The mice are actually incredibly resilient to alcohol. So I'll tell you now what that model's like and why it has to be so severe. So we actually give them alcohol for 11 days in total, and for that period, alcohol is their only calorie source. And so imagine that, you know, you're a collegiate version of yourself and the only source of calories you have is something something a little bit stronger than beer for about a week and a half. That's roughly the paradigm we run these animals through. And at the end of that, then they get a binge, and that actually introduces the more severe disease that reflects what happens in human patients. So it leads to a lot of damage within the liver, and the hypothesis we have is that a lot of that damage that develops is what leads to some of these sedative behaviors. They're experiencing a lot of tissue trauma in that moment in the older animals. What we find is that the young animals, not only do they not get hangovers, but their livers are really resilient to this alcohol model. You actually can barely manage— measure damage in some of them, um, they're so resilient. And it's only with age really that the hepatocytes lose their ability to be resilient to this sort of injury, and you start to develop much more dramatic trauma as a result of this alcohol exposure. So we can actually take that model and you can then run it on a repeat. So after they've run for 10 days plus that binge, you then just repeat that over and over. And what you find in that setting is that most old animals will eventually succumb to mortality mortality. And even with just single doses of our prototype medicines, we're able to substantially improve their survival in that setting over the course of multiple months. So it's pretty wild when you think about it. They're getting just a single dose of this medicine, and then months later they're able to survive substantially better than an animal that was untreated or just got out of control.

Okay. And so, and so you've seen such good results. This is going to be the first that you take to human trial next year. And would this be in healthy patients, or are you going to look for people— I, I, I never quite know exactly how these trials work. Or do you find a group that is, is suffering and, and see if it works there?

Yeah, so typically what you would do for a Phase 1 is you want to start in a patient population that's at low risk, because by the Hippocratic Oath, the first thing you want to do is no harm.

Yeah.

And so there are cases where you could imagine going straight into the sickest patients and trying to help them, In our case, we think probably the more pragmatic path is to find patients who have some degree of liver disease, but not a very severe form. So you can actually find patients who have just what's called steatotic liver disease. That's a very fancy way of saying they have some fat in their liver. Turns out about 40% of the population has some fat in their liver. So when we talk about this disease, the people who might be eligible for such a trial actually could be, you know, 1 in 3 people that you pick out on the street. It's not a super acute population.

Are you gonna start in Australia? Australia.

We probably will do the first trial.

We hired a 1 in 3, buddy.

Okay, as the Aussie in this group, or at least the Texan turned Aussie, you're allowed to say this.

I'm half Aussie, everyone relax.

That's allowed.

Point of pride.

So, so we'll be doing it in Australia. And so the first patients, they'll have some degree of liver disease, um, but they won't be very severe. And the reason that population I think is optimal is not only do we have a chance to provide some benefit to those patients, but because they've got a bit of liver disease, in addition to showing that the medicine is safe, you also get a little bit of signal as to whether or not it's actually efficacious. And so that's the current path forward. Of course, these things might change. You know, nothing is written in stone at this point, but we think that's the modal expectation of how we'll start to get some of the first data in humans, not only about how this drug works, but to our mind, how reprogramming might be able to provide benefit in humans in general.

Okay. And then I know you're always judicious about what you're claiming, and then I know you have to be when you're starting into a clinical trial extra, but I'm just want to make sure I'm calibrating what you're saying. I mean, based on what you have seen in mice, is your expectation— yeah, I mean, so what are you thinking? I mean, say you get into human, things are going well. I'm trying to calibrate whether you're talking about like a little— you're dialing things back on a damaged liver a bit, or are we talking about, okay, you know, all of this damage is erased and I'm back back at like 18-year-old that I am, I haven't been drinking liver.

Certainly our hope would be the latter. Yeah. What I, as you alluded to, I try to be judicious. I want to make sure I'm not overstating. In the animals, the results we see are really compelling. And so I think we are very excited about the product. We wouldn't be moving it forward if we weren't. So that seems kind of obvious. And in addition to just the mice, we've even, we actually discovered this particular group of transcription factors in mice that have human livers. So they're humanized liver animals.

Oh, that's right.

It's about us that we did that. Yeah, we chatted about this last time, but it's about as close as you can probably get to testing this in a human liver without actually going into a human. So I don't want to claim it's perfect, but it's probably about as close as you can get. So we have a lot of reasons to be optimistic. That said, I want to be moderated about what we claim. The hope would be that we have really dramatic effects. I think even if we're able to reset the liver back to, let's say, an 18-year-old liver. It probably doesn't immediately reverse all pathology. Even 18-year-olds sometimes have sick livers if they eat really high-fat diets, as an example. But what we do know is that if you develop any sort of liver disease, even just metabolic syndrome, which happens to about half of all people over 60, which is an increase in rates of obesity, diabetes, cardiovascular disease, you're way better off with a young liver than an old one. And so while it may not be a panacea that fixes everything just as one medicine, it could be incredibly broadly applicable if it's efficacious. Yeah, okay.

In the humanized livers. Oh, before— oh, let's do— okay, fine. We show this, we show this in the video. I mean, this is wild. You do have this like this human-ish liver in a mouse.

Yeah, it's pretty weird. So we actually— we've got a guy on our scientific advisory board, Marcus Grumpy, who invented these things. So the mice basically are— one, they have no immune systems, so you can transplant human cells into them. Which is wicked cool that it works at all. And then two, they've got a genetic defect in their own livers. And what this means is that if they're untreated, their livers just start dying. And so you can actually treat this genetic defect with a drug you put in their water. So most of the time you keep these animals healthy, you give them the drug in their water, but then you can actually just inject them with some human liver cells and you take the drug away. And because their liver is getting sick, the human cells will land in their liver and they'll start regenerating regenerating. They see damage all around them, just like they would in a human liver, and they say, time to start repairing. They try and regenerate and repair this, this tissue. So what we do is actually, before we, we build these animals and run through that whole process of regenerating their liver over a period of about 3 months, we are able to also edit some of those human liver cells so they harbor groups of transcription factors. We can then ask, in this setting, which resembles like a real human liver, do any of these groups of TFs make an make an old liver cell look young? It was a human liver cell. And more importantly, do they make it act young? Does it make this cell regenerate better like a young cell does? And so that was actually some of the first signal that led us to this first product. In one of just our very early screens, we saw, oh, this combination looks really exciting. So we went and we followed up and we did a bunch of follow-on studies. And it turned out that each of the follow-on studies we did only increased our conviction. You know, it's kind of funny to think that we discovered this based on the expression of genes and regeneration in a mouse walking around with no immune system with human liver cells in it. And then we test it in a mouse model. These are just normal mice, and we give them alcohol, and it prevents hangovers.

Yeah.

And I don't think anyone would have drawn the line and suggested that system would be quite as predictive as it has been, but it's been very fruitful for us.

And in this first human trial, somebody will get an mRNA injection.

So what they'll get is a dose of these LNP RNAs. Yeah, exactly. And you get this by an IV. IV. So it's actually a pretty quick infusion. It's what they call a bolus infusion. So rather than being stuck in a chair for like 8 hours, which can happen with some medicines, it'll be about 20 minutes. So you'd be in a physician's office, you'd wait about 20 minutes, it would go into your arm, and then you'd go on your merry way.

And again, I mean, the, the greater cause here is, is reversing this, this long-term damage. But so if this all works, do you sell like a, uh, like a pen for, for the hangover? Is there a different version for emergency?

This, this does get to like Actually, very interesting piece of science. So I'm gonna like, I'm gonna like take your question, which I can't quite answer as a fiduciary, and like pivot it to something cool. So right now the drug is actually being administered through IVs, and that's how everyone's done these LNP RNA drugs. But obviously IVs, while they're very tolerable for patients who have liver diseases, they're not the ideal form. You know, you could still commercialize a medicine. Lots of medicines do this. But all of us would rather that the eventual form of the medicine is kind of like Ozempic today. You've got a pen in your fridge and like you do it at home. It's like not a big deal. And so there's some really exciting science going on that's actually helping admin— helping make lipid nanoparticles with RNAs administerable through sub-q dosing. Sub-q just means subcutaneous. So that's exactly how Ozempic and the other GLP-1s are administered. And so now I think because of those medicines, we found that as long as you're providing real tangible benefit in people's lives, they can feel it, they can see it in the mirror, whatever it may be, that's actually a very widely consumable format. Many patients are comfortable with it. And so our hope would be in the long arc we're able to actually move our medicines to that same format too. That science is early today, but I think in a few years you're going to start to see more and more medicines like this type of LNP RNA go to that sub-q format. So eventually, the short answer is yes, it'll be a pen, but the first version will be more an IV.

Well, and I was being a little flip, but I, I did mean this quite seriously. I mean, if the— for the— to reverse the, the longer-term damage, you would have to have this therapy what like every so often?

So we don't know the exact timing today. That's part of what clinical trials will tell you. Based on the data we have in animals, we think this would be on something like a monthly or every couple months schedule. So hopefully convenient enough, but still you'd rather not have to go in if you couldn't, or make it last longer if you could.

Yeah, and I'm not trying to put you on the spot, I'm just very curious, but would there be just like a, I've had a rough night out version of this in some form?

It's a longer, longer-term speculation that I can submit to at the moment. What I, what I will say is I think we've been really inspired by some of the movements by the incretin mimetic companies. So these are companies making GLP-1s like Eli Lilly and Novo Nordisk, where they've shown now that you can actually sell medicines like this that provide real benefit to patients direct to consumer. So, you know, Eli Lilly, now you can go on their website to LillyDirect and you can buy tirzepatide, which is their current best generation weight loss drug straight off the website the same way you would a pair of Nike shoes. That has never been how our therapeutics industry has worked. It's been horrible. Imagine if to get your iPhone, you first had to go to like an iPhone approval agency and they like, you know, made you wait in an office. It smelled kind of weird and you read like a 2-year-old National Geographic and then you talk to some guy for 30 seconds who doesn't want to see you. And then eventually he gives you this piece of paper which says you buy an iPhone, then you go to like another store and you wait in another line. You finally get an iPhone. At the end. And yet people do this because medicines are such amazing technologies. They provide some of the most dramatic benefits in our lives, I'd argue, far and away above some of the consumer technologies we use. And I think we're really inspired by this notion that we might get to a world where medicines like ours could be available to consumers directly, where because consumers see that benefit, they can move around that traditional system and actually get access to the product in a way that's a much more direct relationship.

Yeah, no, we've been covering this a a bit on the podcast and in our writings. We're at a very interesting time with all this stuff, both encouraging and there's a little freewheeling out there on the peptide gray market.

There are certainly some things from Temu that you can buy that I like personally wouldn't inject, at least mass spec at first, is my suspicion.

Yeah. Yes. Okay. And this liver therapy, how many transcription factors, how many proteins is it?

So we haven't disclosed that yet. What I can say is that we certainly I haven't disclosed what the names are. Well, I can say is that all of the payloads that we've disclosed publicly are somewhere between 1 and 10. So I should think it's a countable number. You know, you could list off the number of genes here. It's not 100 different genes at the same time.

I didn't realize this. I thought everyone was thinking 4 might be the max, max number, and then it would be 4 and under. But so you guys have seen stuff in your universe where you were getting a above 4 and interesting things are happening.

You can definitely find interesting effects from larger groups of transcription factors. And there's actually decent evidence for this even in the public domain. So when we think about development, how does each cell type in your body decide what cell type to be? Often it's pretty large groups of TFs. It's 5, 6, 7 that might be required to specify one type of cell versus another. So we sort of have pretty good evidence at the scale of hundreds of cell types. You might need bigger those. And then likewise, we found that, for instance, if you take Yamanaka's original 4, you can add on a couple more transcription factors and even get better efficiency. So I think we know that you can have really dramatic effects with as few as 1, and you can start to get more and more interesting effects as you get larger and larger combinations over time.

And will you— will you guys reveal this at some point in like a patent? Would somebody be able to reverse engineer what you've done at some point?

So I think when we go into the clinic, you know, we will have IP that becomes public, and individuals will know which genes we are drugging. What they will not be able to do, because of how we've filed some of that intellectual property, is try and really quickly reverse engineer that and make a copycat. Um, so it turns out in transcription factors, these are— they have very important regions of the genes that are necessary to bind the right pieces of DNA. And we've demonstrated and built the various versions of these sequences that are necessary in order to get really strong IP moats around these different groups of transcription factors. So that gives you what's called a composition of matter. And you also get a method of use. You know, if you're the first person to discover that drugging Gene X helps solve Disease Y, I think one of the beautiful things about American capitalism is our patent office will say, that's a really important thing for humanity to know, and you can get a patent on it. And so we'll have intellectual property that makes it difficult for anybody to just copy. But by the time it goes into humans, yes, those patients will know what's going into them. They're not going to be asked to sign a mystery bag of who knows what that goes into their arm.

I'm possibly getting out of my depth here, but I know when I talk to my friends working on mRNA therapies that the liver is usually the easiest place to go because it wants to suck all this stuff up, and getting outside of the liver— outside of the liver, sorry— is, is the great challenge, um, where things usually start to fail. Does that mean I mean, does that lessen how encouraged we should be by your result? Like, if you hit the lowest hanging fruit and then when you start moving out of the liver you run into the same problems?

So you have some smart friends. They're definitely right. The liver is the easiest place to go with these lipid nanoparticles. And the reason for that's pretty easy to understand. The liver is where a lot of fats go in your body to be processed as energy. So it's kind of like the whole organ's built to suck fat out of your blood. So to the degree you're using that as your delivery vehicle, that's where where it'll go. Of course, I am obligated to say that no, we should not be less stoked about making livers young and curing hangovers because, because it might be harder, harder later on. Still a win. Thanks, Ashlee Vance. Give me that one. It's still a win. What I will say is that we've been really impressed by how the field has progressed on delivery to other cell types. So you're totally right. If you just pick a random cell type in the human body today, it's going to be pretty tricky to deliver because there are hundreds of cell types and only a half dozen or so where we can deliver. Deliver well. That said, those cell types where we can deliver well still control really important aspects of biology. So there's now something like a dozen clinical trials going on where people direct lipid nanoparticles to T cells in your immune system. So that's one of the reasons— that's another program we have active. This has been demonstrated in humans in the real world walking around. It's not just a science fair experiment anymore. We also have this new program in the endothelium that I mentioned earlier. One of the nice things about the endothelium is that its whole job is to line the vessel. So these cells are right there. When you get an IV and you put these new lipids into the blood, one of the first things they're gonna see is endothelial cells. And so we've engineered some tricks that I can't quite disclose all the nature of, but there are ways you can build these particles so they avoid the liver and then go to one of these other target cell types like an endothelial cell.

Okay.

And something we'll announce publicly soon is that we actually have developed a lipid nanoparticle that allows us to get very meaningful amounts of RNA into endothelial cells in kidney, where we think we're going to be able to help treat one of the most prevalent diseases, which is chronic kidney disease, creates huge burden on not just the healthcare system but the US economy. And then in addition to those products that we're already working on now, I think we send around news every month about another cell type that an academic group has managed to crack using this lipid nanoparticle technology. So as an example, there have been some really exciting results just this month in skeletal muscle, where it looks like you'll be able to get mRNAs into most of the skeletal muscles in your body just with a systemic IV administration. Administration. And so I think that train will continue. And as of right now, the delivery folks are still well ahead of us on the protein reprogramming, and they've outpaced us in terms of how many cell types they unlocked. So if we get to a place where we're so successful that we've made successful medicines for all the cell types we can get to, I think we'll be in a happy enough place, in a happy position, where we can start to develop the delivery tools to unlock the next sets.

This is not at all to undermine what you've discussed. I mean, liver, vasculature, kidney, T cells, all enormous, important things. Is it slightly discouraging that I don't hear you saying, we've done all this initial work and we're bringing the heart to a more youthful state, we're bringing your lungs, like your massive organs, to— that this isn't the first things that are lighting up as you guys do your research? Is that Is that discouraging at all?

Yeah, I do think one should be discouraged if your hope was we're going to have reprogramming medicines next year in the clinic for every cell type in the body, because unfortunately the delivery tech's just not there yet. Candidly, no one knows what payloads to deliver next year yet. So I think if one wants to take a pessimistic view, that's the way. But I think if we look at this on the scale of a decade, there's actually a lot of reason for optimism. And I feel like anyone listening— I actually didn't plan it this way, but the cell types you just mentioned are actually some of the others where there's pretty encouraging data. So in the lung, it looks possible. You can actually get into the lung through nebulization. So you can actually use a nebulizer and breathe in lipid nanoparticles and get access to a lot of the epithelia there. And then in the heart, this was actually one of the very first places mRNA was delivered, and it can just be delivered directly by injection. Um, Moderna did a trial here back in— oh gosh, 2013, 2014— showing that this worked. Um, and so I think there's a lot of reason for optimism there. And in each one of these tissues, one other reason to be optimistic is that the endothelium is a really important part of that tissue. And so if we can get to the endothelium, you know, we might start in kidney. That's where we're going to go first because it's such a big problem. But if you find payloads that work in the endothelium and you're pretty good at getting into one tissue in the endothelium, then it's going to be easier to get into the endothelial cells of another tissue than it would be some totally de novo cell type. So our ambition would be something like, if you're able to restore function in the endothelium in the kidney, then potentially could you take that same payload and actually send it to the blood-brain barrier and actually help prevent cognitive decline, preserve mental acuity later in life? That might not be what people's first guess is at how you preserve cognitive function, because people might immediately assume you need delivery to neurons. But I think people underrate the fact that there are some of these cell types which are common across tissues, where if you can unlock them, potentially you see benefits in a lot of places you wouldn't expect. Yeah.

Okay. When you guys— I hadn't thought about this before. So when you guys first started out and you're doing this grand survey and you want to go to places where you think you can be effective first. So you were, in addition to trying to find these transcription factor combinations that were new, you were already looking out to delivery mechanisms and, and using that as something of a guide of, well, well, we know we can get to the liver, let's go see if there's something interesting there. Or do you just set the— in my head I pictured you set the AI to work and, and you want it to go, you're just doing the whole body, and it's like, oh, if the heart lights up first and the AI is kind of hitting on something interesting there, that's where we go. And it just now, as you're explaining things, I was running through my head of, oh, okay, well, you'd be limited even if you found something. So maybe you start? Where, like, have you guys already seen that there's promising transcription factor combinations for these other parts of our body, but you're just not chasing that because you don't think we can get there yet?

We have not done tons of experiments in lots of other cell types in, in human, so I can't say like, oh, we have the combo for cell type X, we just don't have delivery for it. But you're alluding to exactly the process we walked through when we chose which cell types to work on, which was, you know, if I can trivialize it down to just 3 things. One, does delivery work? If I discover the right combination, can I actually make a drug? If the answer is no, then it's just kind of a science experiment at that point. Maybe you'd be able to engineer delivery down the line, but you don't want to have to bet on that. You want to start in a place where you feel really confident. Step 2, and this will sound shocking, but it's actually quite difficult to establish, is if I made that old cell young, just that one type of old cell in a human body, would it actually provide a lot of benefit? This might sound like the answer should be obvious, like, of course it would provide benefit, every cell type should be younger. But you never know. If I have a broken chain and I just fix one link, does that actually make the patient better off? So the second thing we look for is, do we have really good evidence that if we could just make that one cell type younger, patients would actually benefit a lot? To give you one example of what that looks like, in the liver, we actually have this natural experiment that's occurred where humans get liver transplants all time. And sometimes you get a transplant from a young donor, sometimes you get one from an older donor. And it turns out that getting a transplant from a younger donor means you're way better off than if you got one from a younger donor. Your prob— probability of survival is higher. And in some studies, actually, your probability of a number of other diseases that might manifest, like diabetes downstream, is lower as well if you got a young liver. So we've got this kind of nice experiment where these patients are older and just the liver is younger, and that seems to do great stuff. And so that can give us conviction. Prediction. And then the third piece is exactly what you alluded to, which is really our core technology. If we've decided we want to go after cell type X, can we actually find the right transcription factors? Do we have a way of running the screens? Do the AI systems— are they able to reason through the reprogramming space? And so all three of those things have to be true in order for us to pursue a cell type. And that's kind of how we ended up on liver, endothelium, and then T cells in the immune system where we started. And it's the same process we run through as we think about what should be fourth, what should be fifth.

Okay. And like, in my notes— and now they're fuzzy because it's been a while since we chatted— but I mean, you've, you've talked in the past that you've, you've found 36, I think, potential things that could restore function. Am I making this up, or—

number's bigger now. Yeah.

Like, so what— okay, what, like, what does that mean? That means, um, because we've talked about kind of like 4 more specific cases. So, but you have seen other hopeful stuff, and it's Is it just a matter of testing that out or, yeah, I mean, why not? I mean, I know it costs money and time, but why not?

Why not make all the drugs?

Why not chase them all? Yeah.

Yeah, it's great. So we are gonna chase many. You know, we're gonna underwrite a number of programs into the clinic, but each of those payloads that we're describing, so we now have dozens where we've shown we can make an old cell act like a young one after reprogramming, take some function that gets worse as these cells get older. For a liver cell, that might be it's less regenerative. For a T-cell, might be secretes too many inflammatory cytokines that, you know, are aberrant and give you inflammatory diseases. For an endothelial cell, it might be that it's less resilient to damage in the body. And we can actually fix some of those, uh, those functions that have been lost just by reprogramming. So again, we have dozens of payloads that, that pass that goal. But then you need to actually take those payloads and turn those into a prototype medicine. It's one thing to show it using what we'd call a tool molecule, some way of turning these genes on that gives us a hint at what might happen if we made drug. Making the drug itself can be more complicated. In endothelial cells and T cells, this is a little trickier. And in hepatocytes, it's pretty straightforward. And then you actually need to go and run all of the preclinical experiments, so put it in all the animal models. And if the payload really makes an old cell younger, what we'd expect is that all of the functions that decline with age should actually improve. If we're actually making a cell younger, then everything that gets worse should get better. That seems straightforward, forward. And so running that full battery of tests actually takes quite a long time, costs a decent amount of money, and so we can't do it for every single promising payload we have. And so what you end up with is what in drug discovery is very canonically called a funnel. You have some number of things you're excited about, and at every stage you have to make some decisions about how many of those do you pursue at the next more rigorous tranche of experimentation.

Okay, okay. Based on what you've seen with the mice and your liver therapy, would you Would you take it if you were allowed to?

I may have been quoted as saying, "If I had to take the medicine for us to progress, I would do it today." Yeah, yeah, okay, so like meaning— I haven't, just to be clear, I have not.

You feel it's safe?

Yeah, it seems in the toxicology studies we've run thus far, it seems very safe.

Yeah, I think I, how do I get in the trial, man?

Do you have an Aussie passport?

I do, I do. Can I just like use that but stay here? I think the FDA would have a thing or two to say about that. I mean, so does this— does it mean— not that we would, you know, actively encourage a life of binge drinking, but I mean, is that— is— are we— are we heading to a world where that— it sounds like that might not be—

is there a moral hazard here that we have to worry about?

Well, no, no, no, no. I just mean, I mean, it sounds pretty good for people who might have been— might have been enjoying themselves for, um, 2 or 3 decades and worrying about the consequences of that.

That would certainly be our hope, you know. And I think one of the one of the, the things we're really excited about is that because not only does it prevent new damage from happening, but it helps the liver fix damage that's already occurred, this could be something that benefits both people like us who are already healthy, stay healthy longer, but also help patients who have maybe already been through some of that disease journey and help dial them back to a healthier state. And when we think about sort of the moral hazard question, it does come up, and I do think we have to take it seriously. The way I've thought about this is very similar to, again, being inspired by the, the incretinimetics, Olympic, a lot of questions came up around that time as well of, are we encouraging people to eat poorly? Are we making it easy for people to sort of navigate around their own poor decisions? And I always felt that those questions were really motivated from kind of a foolish place. I think if we take on an abundance mindset, we want to create technologies that give people as many options as possible and the opportunity to recover from as many decisions they may or may not have made along the way. As possible. And so our hope is that by providing a medicine like this, you're giving patients the ability to make the decision for themselves about how they want to live and to be able to come back from maybe some disease that's already developed. And so I think we don't spend a ton of time worrying about how should we engineer other people's behavior, more thinking about how do we give people the choices they need to stay healthy longer and how do we help patients who maybe don't have any options today get an option to be healthy again.

Well, I have lived. I'm excited for you guys too. I hope this trial goes well.

So do I, actually.

I'm encouraged. I'm encouraged. And okay, I've got one more question that I'm going to go broader with you after that. And you addressed this a little bit earlier, but some of the critiques, I think, I think that people have voiced to me about transcription factors— actually, I have two more technical questions and then we'll go a bit broader— is that, and again, you did flick at this a bit earlier, but there are so many cell types, even if you're talking about neurons, you're going to have multiple types of—

Hundreds of types of neurons.

Yeah. And this is the trick with transcription factors is that there's just too much fiddling to do across all all these different cell types to get to somewhere super meaningful in something like the brain or the heart?

Um, yeah, I think for each one of those cases you could tell a different story, but the way I would frame this is that something super meaningful is a very subjective question. I personally think that the GLP-1s are super meaningful, helped millions of people live way better lives than they would have otherwise. The best health economics suggests that the median American gets a couple extra healthy years of life as a result of these medicines. And so to me, that is a tremendous, tremendous success. And, you know, the markets seem to agree. They ascribe about a trillion dollars in market cap to just that class of products. And so I think when people come at epigenetic reprogramming and they argue that, oh, it'll just like never be meaningful enough, I think they're often baselining to something non-existent. They're thinking like, well, unless this drug immediately lets me live forever, it's not worthwhile. We shouldn't even work on this. We should just keep grasping at the ether and try and find some panacea medication. I really don't think that's the way progress in any technology domain has ever come about. Really, I think what happens is that you develop early versions of something and you iterate. And so if the first versions of our medicines can give patients additional healthy years of life, even just a few in the first versions, imagine if I could offer you 3 extra healthy years of life with your family. Like, what's that worth to you? It's probably worth more than like everything else you have, and on the last day of your life, if I could offer you 3 more years of life with your family, it's probably worth literally everything you have access to in order to get that. And so I think when you run the mental experiment that way, you come to realize that even just reprogramming a few cell types is one of the most amazing opportunities we have in healthcare. And then I think this can be an infinite game where, as you alluded to, there's hundreds of types of cells just in the brain, hundreds in the— or hundreds throughout the body as as well, that if each one of these medicines could have such impact, we get to a place where we could provide potentially decades of healthy life with this technology. Maybe yes, it's true that seeing our way to infinite time horizons just with the tools we have in hand right now is hard, but I think that infinite time horizons really aren't on the table through any other technology. And this is our very best bet to have truly a transformative outcome. If we were able to achieve that, adding extra decades of healthy life, it would be as big of a transformation as the advent of antibiotics or handwashing, germ theory, be as big of a change in medicine as anything that had ever occurred. And so to me, I think that subjective pessimism really comes from a place of, are you baselining to reality or to science fiction? And I think if you baseline on reality, it's hard not to be stoked. Yeah. Yeah.

Yes. Yeah, no, that's very well said. I might embarrass myself again here. When I remember it was earlier this year in Japan, I think they've okayed a therapy for Parkinson's based on induced pluripotent stem cells, which is, is a form of what we've been talking about. When you take the, take the cell all the way back to its stem cell-ish state. And then I can't remember if they're giving them neurons based off those iPSCs. You're nodding yes, that that's what's happening. And then I had my my doctor friends shit all over what these people did and tell me this just does not work at all. And I, and I see other iPSC-based stuff for the spine startups going all these other places. Um, so like, okay, yeah. And, and compared to what you guys are doing, um, I mean, so what have they— in the Parkinson's case, they've done what? They've— so they've taken it all the way back, then they've made a neuron, and they're putting this neuron in your brain, and tissue's already damaged, but this is helping hopefully create kind of new connections and new pathways by putting these, these revived neurons in your brain. Did, did they have to do the same sort of work, or they're just using like the Yamanaka factors we already know about to, to make those?

So I believe they made the iPSCs just based on the Yamanaka factors, and then the real science in a lot of these iPSC, or more generally pluripotent stem cell derived products, is not how do I make the iPSC. Yamanaka did a pretty good job and kind of solved that. Um, it's more how do I make these other cell types from iPSCs. If you think about how they form in the body, these cells all get made in this crazy complex developmental process. They're formed in an embryo, then that embryo turns into the beginnings of a body, and they've got all these other cells around sending them just the right signals at just the right times. And then you think about trying to make these in a lab, and you have none of that. You've got some cells growing on a plastic dish that is something like several orders of magnitude harder than any substance they're ever going to encounter in a body. And you have to try and recreate all of the same signals that evolution discovered during— took place upon development over the course of billions of years. And you have to do that on top of plastic. It is amazing any of this stuff works at all, right? It's just so hard. And so they have found ways to make neurons that are maybe similar to dopaminergic neurons in Parkinson's, Ashlee Vance, but without having read through the exact details of the protocol they used in this paper, um, or this particular trial, sorry, it's very challenging to get cells that are exactly as mature as those that actually arise in an adult brain. And so the cells they have might be dopamine neuron-like, but maybe not quite dopamine neurons. And then they've got this other crazy challenge, which as you alluded to, is these are cells you're growing ex vivo in dish, and then you're transplanting them into the brain, you're shooting them in through a needle. So they've got to survive all that. Then they have to land in your brain. And then once they land, they have to like find a place to lock on, hunker down, survive. And then they have to figure out where to wire their axons. Who am I supposed to be talking to in this brain? Which cells do I take signals from? Who do I send signals to? All of this is very complex and it's very distinct from the way the brain naturally develops. So evolution doesn't give you any clear guarantees. There's no guarantee that if I just put a neuron in some random place in your brain, it's going to wire and fire the right way. And so I applaud folks trying to make an effort of many of these types of transplantations. There have been folks going back to deriving some of these cells from more classical embryonic stem cells, doing trials like this in the '90s. And I think there have been hints of signal in some cases, but I would agree with the general sentiment that it doesn't seem to be a panacea for Parkinson's today, probably for many of these reasons. You know, the cells aren't quite the right cells, the transplant doesn't quite let the cells survive, the cells don't quite wire to the right place, and sort of the same story for the spinal cord. All the same problems apply.

So we have one group of companies that are using this technology based on Yamanaka's discovery to take the cells back and then lab grow certain specific types of cells, be it a neuron on, or a heart cell or a lung cell. And then we have this other class of companies that are doing this experimentation around new types of transcription factors and new ways to reconfigure the body. And these are kind of like the two camps that have really grown out of this discovery.

Yeah, there's two sort of camps. And maybe one way I'd draw the line too is there's a group of firms like those you were just discussing who are trying to do this all ex vivo. Like, let's use Yamanaka's discovery and we'll reprogram some cells in a lab, we'll make another cell type, and we'll transplant it back into you. Sometimes if you're in the aging community, which is a fairly small niche, they talk about this as replacement. Rather than trying to fix the body, let me just make new parts and swap them out like a car. For all the reasons we were just discussing, I think it's an attractive idea, but it's hard, not totally trivial. Whereas the groups like ours, what we're trying to do is build upon this discovery and fix the cells where they are to actually make medicines that you would just take and actually have your existing cells perform perform better. So rather than trying to do some complex surgery where we're like ripping parts of you out and swapping other parts of you in and making parts ex vivo, I'm just trying to take the cells you already have and make them work like they did 20 years ago. And so I think that's a harder— maybe not harder, it's a very difficult biological problem. You have to figure out which transcription factors to deliver. But ideally, it's actually an easier administration problem. Once you've figured out what TFs to use, it's way easier to give somebody an IV than it is to do all this complex cell therapy Okay.

Okay. All right. We're gonna broaden out for a second.

Uh, I thought we already were.

Well, we're going even broader, man.

We're going broader. Okay. We're going to the moon.

So I can't tell if I'm putting you in an awkward position or the perfect— you're the perfect person to answer this. I mean, Eli Lilly is an investor, you guys, and so is Thrive, and Thrive is a massive investor in OpenAI. Um, because my question is gonna revolve around pharma and the foundation. Labs. Um, it's all of a sudden, in some ways, I mean, Demis and DeepMind have been chasing biotech for a long time, but over the last year we've seen just this flurry of moves. Anthropic acquired a biotech lab. OpenAI has been investing in this, this area. DeepMind, Google, just money. And I think they put another $2 billion into Isomorphic Labs, which which is pursuing biotech type stuff. And so it's been this interesting moment where the foundation, the big, big AI companies are signaling that pharma and biotech is a huge area of interest. I mentioned earlier, it seems like either everyone's lying to me or this AI stuff is actually working quite well in the biotech area. And then you got the traditional pharma, which has the actual, Well, in some cases has the labs, has all this years of data. And I don't know if this is crazy or not. When I play this through in my head, I'm just like, oh, what's been the story of like the last 10, 20 years? It's been tech moving into other areas that were formerly dominated by, by different companies. I mean, tech has taken over Hollywood. You know, they are like the major major studios now, transportation, you name it. This is, this is what happens. Are we heading to some moment where really the, these foundation labs, because they have the biggest computers and the most expertise and the best talent, are, are like our dominant pharma players? Or are the pharma people, they just have too much expertise and data and knowledge here and wet labs to, to, to be overtaken? Taken.

Yeah, I think this is one of the most interesting business model stories we're going to see play out over the next 10 years. So the short answer is I don't know. I'm going to guess with you. And so—

Tell me where to put my money, man.

Yeah, I'm going to guess with you and speculate. But I think it's a really interesting experiment to see happen. And I could imagine many possible futures. The possible future I think is most likely is maybe the one informed by history. So at least within therapeutics, there was a similar moment to the one we're in now in the late 1970s with the advent of recombinant DNA technology. So historically, pharmaceutical companies were really chemistry companies. Most of them actually started as companies that made clothing dyes, believe it or not. And then one day they started shooting clothing dyes into mice and hoping it fixed diseases. And that's actually how they found one of the first antibiotics, which was like shooting clothing dyes into mice. And they're like, we got a bunch of dyes, let's try and make some drugs. And so they're really chemistry firms. They weren't biology companies in the way you might think of them today. And when recombinant DNA was invented, not far from where we're sitting in South San Francisco, near Stanford, up in— up at UCSF. That was really the origin of biotech as we know it. Companies like Genentech and Amgen and Biogen and Vertex, which are all, you know, other than Genentech now being acquired into Roche, huge firms today, um, all began with that core technology that allowed them to clone genes and produce genes as drugs in a way you had never done before. That was really the last true technological shift that our industry saw. And what happened was not that the pharma companies ate the technology and then, you know, totally subsumed it. It also wasn't that the biotechs totally replaced the pharmas. You know, Eli Lilly still exists. Rather, what happened was that the biotechs produced a whole new class of products, and then they initially partnered with the pharmas because there were things the pharmas were still better at than the biotechs were. So it turns out clinical trials are really complicated. Running them is very hard. Manufacturing at scale is pretty complicated. And then distribution is also hard. Running around to all the physicians' offices and all the payers in the United States and making sure everyone's got a pen with the drug name on it and they remember what to prescribe. This is probably one of the most complex go-to-market motions in all of the economy, and the pharmaceutical firms are masters of it. And so at least in the early days, all of those functions at, say, Genentech were totally orthogonal from learning how to splice genes. And so they actually partnered with Eli Lilly on their first products. And I imagine we're going to see something similar if you use that story as a template. I really do think artificial intelligence in biopharma is the, the first revolution we've seen since recombinant DNA of the same scale. It's the first technology disruption on the same order of magnitude. And I think what we're going to see is some of these firms that are developing models that help us design molecules molecules, partner with pharmas to develop those molecules and make medicines that just couldn't have existed before. Some medicines maybe could have existed before, but they'll make them way faster, way cheaper along the way. And then I think maybe what is unique relative to the recombinant DNA revolution is these AI systems can actually provide benefit across the stack in drug discovery. Not only can they help you find the right targets to work on, can they help you make the molecules faster, cheaper, or in ways you couldn't have achieved previously. Obviously, but they can also help you with the downstream aspects of clinical trials. Which patients should I enroll? How should I process a lot of the logistical and operational challenges downstream? And so I think you'll start to see those initial partnerships around individual medicines, and then you'll also see these lateral partnerships maybe between large labs and large pharmas where they actually take the pharma's core expertise and expedite it using some of these AI systems. But what I don't think will happen, what I think is plausible but not likely is that we see OpenAI, the drug company, overtaking Eli Lilly. I think it's much more likely that these groups work together and that you end up with a new category of companies which are really AI native. They're therapeutics firms which take on the mantle of the Genentechs, the Amgens, the Vertexes of old in the same way that that epoch gave rise to a whole class of companies. I think this one will as well. Those will be firms that exist at the intersection of medicines and AI. In the application of those tools.

I see what you're saying. All that makes sense. I also see, I see a company like Google that, that does take these big swings. I mean, if you look at Waymo and all the money it's taken to put into that, if you were to try to invent some new mechanism here and to reach people directly, and I know they're acquired YouTube, but you look at the scale of what they brought to bear there. I could see these tech companies are so deep-pocketed and then tend to move faster and be willing to take losses for a while on things. I mean, it's not implausible that they could barge in here.

I think, I think to your point, yeah, it's definitely not implausible. This, this could happen. I think maybe another way of framing this from history is to ask, you know, why hasn't this already happened? You know, arguably notably AI has not actually made money for anyone yet. I mean, asset appreciation, right? But until maybe very, very recently, these companies weren't cash flowing positive. Yeah.

Whereas it seems like Anthropic's making a ton of money, but they're also spending a ton of money.

Yeah. So again, until maybe very recently, there weren't tons of free cash flows. There have been tons of free cash flows in the broader tech ecosystem for a decade plus now, right? So why did these technology firms not move in? I don't think it's that they weren't willing to make the bet or spend the money. What I would argue is that they didn't have some core competitive advantage relative to the pharmas or the existing biotechs.

Yeah.

And so the question this time would be, is AI different? Does having expertise at the base model layer give you an ability to outcompete a traditional biotech or a pharma? And does your lack of expertise in the downstream aspects of drug development get compensated for by that upstream technology, right?

But doesn't, like, Demis winning a Nobel Prize signal something has changed?

I, I do think it's telling that the way ISO got built was they built a dedicated company at the intersection of artificial intelligence and therapeutics, and that they're actually building their own drugs and building their own pipeline. So Alphabet might be unique here in that they do have this sort of holding structure where they're building a separate company to do it. But based on the sociological structures and what it takes to bring those types of expertise together, I do think what you'll see is the rise of these new groups of companies that actually go and build types of medicines that couldn't have existed prior to AI systems. What I don't think will happen will just be that it'll be the first-party product of the labs. So maybe down the line, could I imagine a biotech that is 50/50 partnered with Anthropic and a large pharma and they do a JV together? Yeah, I can totally see that. Does OpenAI just subsume all of pharma and start inventing medicines and selling them first party itself? Again, plausible, as you said, but I think it's the far less likely path. I think there's really nice complementarity between the skill sets of these firms today where these AI systems can be really compelling for developing and creating new types of Ashlee Vance. And as of right now, the downstream clinical development portions have maybe seen less innovation from these AI systems. That will probably change over time, but there's a whole bunch of operational and logistical expertise that I don't think is just intelligence limited. There's a lot of tacit knowledge there the models aren't going to overcome tomorrow. And so you'll actually have complementarity between these actors. And I think all the people working on both sides of that equation are really aren't smart and they'll see that. And the sum of the parts will be greater than just either one trying to subsume the whole.

Okay. Eli Lilly, their market cap is around $960 billion, I assume. I don't follow it super close. I assume a lot of— we've got Retruti coming. It looks like it might get approved by the end of this year or next year, I believe. And I would guess some of that is baked in. Yeah, it's not my area of expertise, but it seems quite miraculous even compared to the GLP-1s.

The trials look really good.

Really good. Yeah. To you, is there— and my understanding is that there's lots of interesting peptides in these pipelines. It feels like we're in an era of— I mean, I want to use— I kind of want to use the word miraculous. I mean, you know, something's changed. I mean, these— the arrival of GLP-1s is incredibly powerful. It's affecting many, many lives. It seems like we have other stuff that people are even more encouraged about. Is there some scenario like 5 years from now where Eli Lilly— I think AI is still— it's obviously getting so much better, but is there— is the health story like actually the world's biggest story in 5 years? And we were kind of— not that we were sleeping on it, but AI AI was distracting us from the bigger thing that was actually happening.

I would like to argue they could be co-equal. I think it would be hard to say that it's larger than the advent of the first intelligent entities that are non-human. It's just like, it's really hard to beat creating aliens. But I do think they are of equal magnitude. And I think you see that even just in the market caps reflected, as you alluded to. And I really think that the story we're going to look back on in a few years is maybe not health versus AI, but the combination of those those two being bigger than a lot of the other intersections people are thinking about today. So when I see a lot of attention going is the story of, is AI going to demolish traditional software? Are we going to have SaaS going to zero? Where is the value going to lay in the app layer or the model layer? And I think those conversations are interesting, but when you think about the magnitude of that ecosystem, it's just much smaller than the potential value of creating new products that give people more healthy years their lives. And so what I imagine is that we have this existence proof that really massive products are possible that add health. These are the incretin mimetics that you alluded to, various combinations of peptide agonists. And I think we're going to see several more. You know, I would hope to say that our drugs will be some of these, but I think others will develop medicines like this too that were only possible because of AI that give you similarly large benefits. And I think we'll look back on this current moment and we'll be like, what were we talking about in in terms of like SaaS stocks and what P/E we should be paying for $5 billion companies. This was just the wrong thing to be focusing on. We should have been focusing on are we going to get plus 2 medicines that give us multiple healthy years of life? They're each worth a trillion. Like that was just clearly the story. I think that's the future I imagine we'll reflect on today rather than one versus the other.

Yeah, my question was a bit jokey, but that's, that's the heart of what I'm talking about. I feel like even though these share prices and these these pharma companies have shot up, I kind of think we might be underestimating. I keep picturing a world where, where like obesity and diabetes have been, you know, if they got eliminated in the United States and then I hear about what you guys are working on, this is a drastically different world.

Oh, absolutely. I mean, I think the craziest thing that has occurred in the past 5 years other than maybe AI is like the average person now has an option on obesity. Like you have an opportunity, there's a technology today that can help you in ways that were not accessible to you just a few years back. And I think we're drastically underrating the second and third order effects of that sort of change.

Like isn't the, God, isn't the medical system gonna, It's gonna be a shock, man. I mean, it's like, oh, we kind of needed these people to be unhealthy to, you know. And then I heard, I— somebody told me like the food companies are trying to— now they're trying to like out-engineer the pharma and make, make food so like calorie dense. Yeah, but it's like, I was like, oh God, man, Jesus. It's like people aren't eating enough, now we have to trick them even harder into eating more. That sounds awful, but—

I do think you point out like the healthcare system, I don't think we've totally baked in what things will look like in even just 10 years if this trend continues. You know, healthcare is something like 15% of US GDP.

Yeah, yeah.

Only about 10%, a little less of that is therapeutics. It's mostly services, operations, diagnostic tests are like maybe 7%. It's mostly administration and care of existing conditions. And so if we're able to dramatically reduce reduce rates of obesity and metabolic disease through drugs like this, I think you start to see meaningful shifts in where that healthcare spend goes. Maybe some of it transmits more from services back to technology and products, and that would be as big a change as occurred in that industry and a huge chunk of the economy has ever occurred.

Are people talking about this? It just seems like—

I mean, we're talking about this.

I know, but it's like these drugs until very recently were were incredibly expensive and people were still using them. Now they're plummeting in price. There's better drugs coming. I'm just like fast-forwarding this 10 years in my head. Like you said, I mean, you would almost have to be trying to be obese at some point because it's very available to you to not be. And I feel like most people will take that, that option. And so I don't know. Yeah, when I fast forward 10 years, I was like, God, there's got to be— walking to a hospital, there's got to be a very different thing going on day to day for the most part. I don't know.

Yeah, I mean, the optimistic take, like, if I can paint the halcyon future, is that if we do move more and more of healthcare to therapeutics from services and it moves to technology, then it becomes much more about how do we keep you healthy the way you already are longer. And, you know, in a way, this kind of harkens back to this notion of preventative medicine, which has never really gotten much traction. Because we never really had the tools to help you. All we could do was shame you into better behaviors. Oh, you shouldn't eat the foods you like. You should work out more. Of course we all should. But if we actually have technologies that help you stay healthier longer, medicines for some of the first time, you know, we only have a couple examples like this, statins, GLP-1s. If you had a whole palette of those tools, then I think the way we experience healthcare shifts a lot. It's not something that you only— you're only exposed to when you're already really sick. Really sick and you have no decision-making authority because by the time you're really sick, you just kind of have to do what the doctor tells you. You're not in a position to argue with it. You're not in a position to make selections for yourself. And starts moving to look much more like other aspects of the technology industry where you and your preferences get to shape a lot of your experience. You don't have to buy a MacBook. You're not in a position where you can't actually make the selection between this and several other options the same way you are now where you only get exposed to that technology after it's already too late.

Jad, when I, when I went to your office the last time, and maybe it's different now, you know, you had robots pipetting and stuff like that, but you guys are— so you're running all these compounds. We just did a story on Medra, which I, I know you're familiar with. I mean, so it's robots doing wet lab experiments. It's my understanding this happens in China all the time. You know, they have these huge experiment centers to do these things. Yeah, like, why do, why do you guys not have of, uh, are you shaming me on lack of any more robots? I'm trying to like, bro, why don't you have more robots? Well, the question, it's, it's, it's, it's, it's part of the earlier discussion of like who's doing what in this, this big pie. And it's kind of like you're searching for therapies. I guess we have what, what does CRO stand for?

It's, uh, Contract Research Organization, right?

So CROs are doing these experiments for people, running these, these tests And then, and then we've got the pharma, we got the AI. I'm trying to figure out if somebody is super vertically integrated across all these things, or if all this stays a little bit divided.

Yeah, so I'd say we're quite vertically integrated by the standards of a biotech, in that we don't use many contract research organizations. Almost every experiment or piece of data we've had a chance to chat about right here was generated in our labs. You know, we have our animal facility right above our experimental lab that you've been in. Whereas by contrast, I think the trend even just 10 years ago was more virtualization. Biotechs were trying to really just be a wrapper layer on top of a bunch of CROs. A friend of mine likes to joke that in the same way that people used to pejoratize certain types of AI app layer companies as ChatGPT wrappers, biotech for 20 years was like a Wuxi wrapper at some point, Wuxi being a large Chinese CRO. And so in terms of how things are verticalizing, I I do think we do still have a division between drug developers and then tool producers. And so the folks who make the best robotic systems are not therapeutics companies. I do think that's one layer deeper of vertical integration than anyone has right now. There are probably only a few companies in the world who can both build a product and build the robot to make the product. Tesla is one good example of that, but they have to get to a much larger scale than most of the therapeutics firms. And so what we try and do is take access, access many of the best tools in the industry, the robots you saw working with folks like Medra and Illumina who do DNA sequencing so that we don't have to invent quite every step, but then we put all of it in the same place and we can vertically integrate from there. Um, to the question, I will try and answer, you know, bro, why don't you have more robots? What I will say is—

robot shamed you.

Well, yeah, it's okay, it's okay. One of the reasons I— this is one of my hot takes— I, I think industrial robotics in biology have actually been a pretty big disappointment. What I mean by industrial robotics is the classic way we've tried to automate experiments is by hard coding what the robot does. Move on the x-axis by y inches, move up on the z-axis by some other number of inches. And unfortunately, it turns out that most protocols in biology are one, somewhat heterogeneous. Things change a little bit. It's not exactly the same every single time. And making those modifications is something that's difficult to on the fly. And then two, the vast majority of all of our experimental scientific knowledge was developed over the past 200 years, assuming that whoever was executing the experiment had the dexterity of human hands and the perception of human eyeballs and the reasoning capacity that's integrated in there. And so we've tried to re-engineer all the protocols onto these old industrial robots, and it just turns out that every single protocol you port over is really burdensome. It's really slow. And so unless you're doing a protocol basically day in, day out, 24/7, it doesn't make a ton of sense actually to try and put it on the industrial robot. You'll spend more time getting it to work than you will actually deriving benefit from the system. So where I'm actually pretty bullish on the future is that while many people like to make fun of humanoids as not the ideal form factor for every task, and people like to say, oh, you're just trying to ape Isaac Asimov and iRobot. I actually think in the context of molecular biology, there's a very strong first principles case for why you want maybe not an exact humanoid with two legs, but something like a humanoid-like form factor. You could either port every protocol to an industrial robot, which is a semi-infinite task that runs in O time, or you can solve the O task of making one robot that has human dexterity and human perception. Then you just get to use all the protocols you already have.

Yeah.

And so such robot does not exist today. I'm really hoping Michelle at Medtronic builds one for me or somebody else. I'll take robots from anybody. If You got a robot that has these properties, let me know.

We like Michelle.

Yeah, yeah. Michelle will get there first. I feel confident. But as of right now, that doesn't exist. And so what that means is that unless you're doing one thing basically over and over and over again, which we do a bit of, but we have far more complex and heterogeneous protocols too, it's just hard to get to the risk-benefit making a lot of sense.

Okay. Okay. You doing okay on energy?

I'm good. We can keep going forever, man.

All right. I'll do a couple more for you. I guess this is a big one. And then everyone knows this is where I go on my podcast is the how fucked are we compared to China?

Oh, okay. Okay. We can, we can riff on this one for a while.

I mean, you're doing your clinical trial in Australia. This is never-ending comedy for me. Every American company has to do their clinical trials in Australia simply because they have some, like, paperwork advantage over us. And I think we seem to be like possibly addressing this, um, slowly but surely. And then, and then clearly can do bigger scale, cheaper trials of all sorts in China, um, if you're able to. And then, and then like we were just talking about, as far as the main, you know, the industrial end of biotech, um, I guess I hadn't fully appreciated how far China seems to be ahead on that already. Yeah, it just seems a little bleak. It seems like we're heading into this fantastic era of all the things we talked about where AI and biotech are combining and the US actually happens to be the scientific leader in both of those fields, but that the do the work end of this is elsewhere. That seems like a massive fucking problem. Problem to me.

I do agree with you that it is a problem. I think there's reasons for optimism, but to, like, to restate the problem, as of right now, it is much more difficult to do an initial first-in-human trial in the U.S. than it is in some overseas geos. We can talk about exactly why that is. Once you've got a trial started, it's often much faster for that trial to complete if you do it in China. It actually takes longer to start. This is kind of a misconception people have, that they're just, like, fast and loose with regulations. And for most drugs, That's not true, but they do have a much larger population. There are more people in Shanghai than in the entire continent of Australia. And so if you're trying to find patients to enroll in your trial, you're much better off being in Shanghai. And so that means you're able to finish the trial more rapidly. And then finally, once you've actually developed that medicine, you have to manufacture it. And we've actually lost a lot of the manufacturing capacity in the US that we used to have. It's moved over to lower-cost geos like China, like India. But that means that it's also more difficult to vertically integrate and produce especially some of these more complex medicines like we make where we're making RNAs wrapped in LNPs. There's just very few places in the US that can still do all this, especially very few places that can do it in one roof. And so that's the case for pessimism. The other case for pessimism on this is that a decent fraction—

We're going to be the optimistic case. Oh, no, no, no.

We've got one more. These are some of the hard sort of constraints. There was a business model in US biotech for a long time, which was basically building a company around a single type of drug. You would in-license some intellectual property. This was usually happening in Boston or the environs. And then you would develop your company basically around this asset, the asset being the drug. And you would run trials, then eventually you'd sell that company to a larger pharma. And it turns out that firms in all nations can read the same patents. You can find them on patents.google.com. This is not a secret, and that people work really hard and are very smart in all nations of the world, especially China, and their ecosystem is able to actually outcompete that model of biotech on a lower cost basis than the traditional Boston ecosystem, which I'll use as a metonym here, has been able to thus far. And so on just a dollar-for-dollar basis, it's really hard to build a business that way in the US anymore. Okay, now optimism.

I'm crying.

Problems. Clinical trial timelines and starting up your first inhuman. The optimism on that is that it's almost entirely a regulatory decision. There's not something in the water in the US that makes it harder to do trials. It's not like we don't have enough people. It's just the paperwork is set up in a different way.

Yeah, I kept asking people why Australia, they just make it simpler.

It's not even just simple. In a way, it's simpler than just make it simpler. It's even more straightforward. In the US, we have a very centralized system. System. So every single first-in-human trial goes to the FDA in Bethesda, Maryland. They have to review everything. It takes a while. They're really overburdened. They don't have enough people to review everything. So I really empathize with the regulator there. And it also means that you end up with kind of a homogeneity of thinking around what parts of the trial do we focus on. The Australian system basically says, let's decentralize it. So individual academic hospitals and clinical trial centers can approve protocols based on their institutional review committee. They have what's called an track, a human rights and ethics committee. I'm probably getting that right. Then the central drug regulator, their equivalent of the FDA, is called the TGA in Australia. They act as a final check, but they really defer a lot to each one of these hospital systems. The notion is that we're better off with each one of these really premier institutions being able to make their own decisions. Each one of them is a responsible actor and have them provide a diversity of different types of services, get different expertise, different types of trials, and then TGA acts as a final check and makes sure they're not doing anything crazy. And because that means that you then have a diversity of people to review, and these organizations are all competing with one another because they all want to be the best place to do a trial in Australia, you get natural market dynamics where people are trying to do better than their neighbor and trying to be the number one site. The great thing about this regulatory problem is that Australia has kind of already done the experiment for us. They've been running this system for a decade. It's great. Terrible things don't happen every day. And so I think we have a blueprint that we could copy and paste over to the US in large part and actually dramatically reduce our time and burden to—

why don't we do that?

I think in any institution, when you've set something up the way that it has been, it's hard to make change. And I don't want to claim to be the sociologist to understand exactly why FDA hasn't made changes, but I do think there have been really important encouraging science recently, like very recently in the past few months, FDA has discussed that they might want to adopt some of the aspects of something like the Australian system in the US, making first-in-human especially easier. Because the more of this industry we can keep in the US, the better. It's one of the very few industries we still lead in. And even just the trials themselves are a pretty big economic component. I mean, the reason Australia made this so easy is they wanted to help develop the industry there. I don't think we should just take that lying down in the the US and totally lose it. So there's reason for optimism in that one. It's paperwork. We can change paperwork. In fact, somebody else already has the right paperwork. Maybe we just copy that.

Yeah. And the relying on wild-type Australians is a problem, right? I mean, they're kind of— they're quite similar.

Well, the challenge, to your point, is they're quite similar. Some of them are Texan. I don't know. You never know how different they are. Part of the challenge in Australia— and this alludes to the benefit China has of just having a larger population. There just aren't enough Australians. Yeah, I really wish there were more of them. They should have settled all that desert. You guys should have terraformed at some point. I don't know why you didn't, but there aren't enough Australians. The beautiful thing about that is, well, the US has a lot of people.

Yeah.

And so if we're able to get to a competitive clinical trial regulatory regime, you know, we're not quite as large as China, we don't have quite as dense in terms of our cities, but we still have the world's leading academic medical medical centers, I think we can be quite competitive. And so many of the large pharmas, even if they do initial trials overseas, they come back to the US for the phase 2 and the phase 3. And I think the reason for optimism is we still have a lot of advantages there. Our healthcare system is still second to none in the world. And so that is a huge crown jewel that we need to continue to exploit and not leave for granted. On manufacturing, reason for optimism is that we invented a lot of this technology. Ashlee is the guy who goes around to factories. Got me some looks. On manufacturing, this is a solvable problem. Like, we know how to do this. A lot of these technologies were invented in the US, you know, in South San Francisco where I go to work every day. There's a sign that says birthplace of biotech in the same way that you and I are currently sitting in Silicon Valley. We don't fab chips here anymore, but that's not a law of physics. That's like, it's some mix of regulatory burden, Baumol's cost disease, and then financial incentives to set factories up here and make sure we do local drug development. But I think there's a strong case to be made that this is an area where policy interventions can be very successful. I think there are even national security reasons. You probably want biomanufacturing in the US. Yeah, if there is another pandemic, do we want to be in the position to make sure we can make our own vaccines here and not have to ship them internationally? Probably. Probably that's a better world. And so the optimistic case is that's again maybe paperwork plus some picks and shovels, but we have picks and shovels.

What about the supply chain? I mean, aren't all Are reagents made in China? And it's so many problems.

A lot of the reagents— so this is an interesting one actually. In terms of the long catalog of SKUs, a lot of reagents are made in China, sure. But I do think for most manufacturing, when you get down to the layer of a single type of drug, the SKU list you need is actually not that big. And so if one were to say, I would like to make sure we have the capacity to manufacture X doses of Y types of drugs in the US, I think one would be able to construct a supply chain that was pretty resilient. I don't think that's a lost cause quite yet.

Okay.

So a little bit of reason for optimism, some places more than others on each of those three. And then the final piece of this was, well, the biotech industry in the US has had a lot of these asset-first companies or asset-centric companies, and we're losing out to China in that game. I would argue that's never been where the US really built the majority of the biotech industry. Where we have been best is in innovating and creating new types of medicines that couldn't be created it before. Genentech, Amgen, Vertex, Regeneron were not single asset companies that licensed some drug that was sitting on the shelf and they put it in a different wrapper and they tried to sell that off for a couple million bucks a few years later. These were firms that invented fundamentally new ways to treat disease. And I do think the US still leads in that. That might not be true forever. And I don't think we should rest on our laurels. And a really important part of this is making sure that our federal biosciences funding is efficient and maintained at reasonable levels. One way of zooming out on this is to ask, why did the US have the premier biotech industry in the first place? As much as we would like to believe as Americans, it's not because we're just smarter. It's not like Americans are smarter than other people. It's that we had the crown jewel of bioscience funding. So we had the best universities, talent from all over the world wanted to come here, and we had the basic science that was the substrate for all of the foundational companies in this industry. And that is still true today. And I think as long as we are diligent about preserving that research enterprise, the Vannevar Bush's endless frontier, and we recognize that we need to maintain that, I think we can still be the most innovative place for medicines, even if one sector of this industry, which is really some of these relatively similar or iterative molecules, goes overseas. Over time. So the most important pieces can stay here. So those are my reasons for pessimism and optimism. I will leave it as an exercise to the reader to render how effed you think we are in your terms.

Well, I think you're right on all this. It's a matter of will and organization, and I remain skeptical on some of that. But it's not like you couldn't—

What's your answer to this question on the how effed are we scale? Where do you come out?

Pretty bad.

I mean, I just— 5 is like totally, like we got to bail and go to Mars in 1 one is like everything's, you know, I just don't know biotech as well as I think I know some of these other fields.

But, um, I, I don't know if you listen to the show a lot, I apologize if I'm repeating myself, but, um, you know, the— for— let's just do humanoid robots for a second. So, okay, like maybe they just don't work for a very long time, but let's put that case to the side. Let's assume we're gonna have humanoid robots, um, in a relatively near future. It's like we're so fucked, it's unbelievable, you know? It's like we, we don't make the motors, we don't make the actuators, we don't make the batteries. Um, you could sort of do like the base parts of, of the robot. So same thing, if AI takes off, which it is, and keeps going, and you're gonna have a physical manifestation of this in some form. I don't see how the US has any part to play in that. And you are right, like, we could do that. However, at the moment, if you wanted to do an actuator company just based on labor, um, most people tell me in the US to do the same thing, it would be 10 to 20 times as much as China. So like, we could still do it, but you're going to have to pay more. And then I just see a bunch of like play acting, really. I mean, if you go to Los Angeles with all these star startups is great. I mean, it's actually like a just myriad ideas, people doing such interesting stuff, but it's such a small scale. Like, if you look at BYD in China, it's just like this entire city of a company making batteries.

I—

it's exactly what you said. We could do this, but like, to go from where we are to that would be such a massive capital and philosophical change. And you would, you'd really have to have the entire United States committing to like buying this stuff and paying a bit more and just is like, we have to have this. And so, yeah, of course the government can help push that along and seems to be on some level. And then in biotech, you know, far, far, far, far infinitely more than I do. All I know is kind of like I get to survey so many people and it seems sad. It seems like China is very rapidly catching up on the innovation front. I kind of like will always always believe in the American spirit of innovation and thinking creatively and coming up with these things, but I think China will be neck and neck, and then they'll just have such an advantage of like how fast they could test it and make it. And seems like that's hard— it's hard to compete with over time.

Yeah, I do think that notion of, of competing with it is an interesting one to pull on here, maybe in a way that's distinct from other industries. Industries, this is a non-zero-sum competition in that if medicines are invented overseas that treat diseases, like, it's kind of a bummer for me as a blood-red-blooded American that those aren't American medicines with profits going to American companies, but it's still a way better world than a world in which we don't have a medicine for that disease. If you go to a hospital one day and they're like, "Hey, we have a medicine for you. Bummer, this is from a New Zealand-based company." Yeah. You're not like, "Oh darn, the Kiwis. I don't want that one," right? You're like, "Oh, thank God." Some people made this drug for me. And so I think patients, which is basically all of us one day, it's just a timeline until we all become a patient, are better off if we have a competitive ecosystem across geos. And I think what we'll end up with is some geos specializing in certain types of therapeutic development, at least in biotech, and other geos specializing in something distinct. I do think probably the US ends up leaning more into its strengths. We try really innovative new types of medicines that have never existed anywhere else, and maybe the nth version iteration of a mAb, a monoclonal antibody, that is meaningfully better than the one that came before it, maybe that gets made in China. But at the end of the day, we now have two good medicines rather than just one. And I do think that's something unique about this industry, is that we, we actually all benefit if there is more interaction, more engagement, and more competition across these different regulatory ecosystems.

Yeah. Okay. I mean, I, and I'm, as I try to state all the time, I'm not really like, I'm not some crazy patriot and I actually, I love China. I love going there. None of this really comes from that. It's just sort of like, what's the future of the place where I live?

And yeah, you don't want to live in a backwater.

Yeah.

Like, like it makes me sad from the perspective of like, we're playing a sport and I want to win.

Yeah.

But at the same time, I think it's important to remember that I think we should celebrate the fact that there are now multiple ecosystems that can produce new medicines that treat diseases that have gone untreated for millennia, since the advent of humankind. We should celebrate the fact there's now multiple entities on the world stage who are capable of doing that. And yes, we should use it as a lesson to ourselves to try and remain competitive and continue to push the frontier. But I think on net, you zoom way out and you're just orbiting Earth and you're asking, are things going better or worse? It's like, ah, now they're curing diseases in two places. I think net better.

That part I like. I mean, and this is just like my— way too idealistic dreams is, yeah, I mean, you look at the US versus the Soviets, it created this competition, you know, obviously in areas like space and things like that, we're like pushing each other to, to greater heights. But, you know, in so many areas, the Soviet system just couldn't keep up with what the US system was doing. And just, just immensely smart population kind of fell by the wayside from the system that it was operating in. It would be way nicer. Yeah. If the US and China could get— if we could have these, these powers, you know, all building together towards something for once instead of having this like opposing dynamic. I mean, maybe you can argue the competition is what helps drive stuff better. But yeah, it does. It does seem like there could be a cool world, especially in this moment of biotech where maybe amazing things are really happening and we're just cranking them out together. And we'll see.

I like, I like this optimistic ending.

I have one last question.

Go for it.

I'm sorry. It was the ending of that one. You, you always impress me on so many levels. You know, I mean, you're brilliant. You're brilliant, man. And but you know, both biotech and AI so well. And, you know, I've run across a couple of these companies now doing the biologic biological neurons doing AI.

Oh, like neurons for inference, like neural computing, organic computing.

What is it? Biological Computing, I think, is one that just came out. And I went, I actually have one of their dishes up there. They had all the rat neurons in this, in this, I don't know, a well, and we're running a model on it. And it was, they weren't doing text, they were doing video. And it was, it was like outflanking an NVIDIA GPU on predicting James and this one model, they thought of it as like RatGPT and it seemed really cool. And then you could, you know, like intellectually it seems like, oh yeah, I mean the brain is so much more efficient and clearly is doing something our LLMs aren't because I'm sitting here in this chair and not in a giant fucking data center, you know, to talk. So something decent.

Please pay no attention to the cables moving into my back.

Something decent is happening and I'm sure we could benefit from that. I was just curious if you'd looked at this and if you were encouraged by what you— it's early days, but what you'd been seeing.

Yeah, yeah. I haven't looked at this company in particular. I'm generally aware of this line of research. And the funny thing is it's a lot older than you would think. There's this guy named Tom DeMars. I believe I'm pronouncing his name correctly. He was at the University of Florida in the aughts, the tens. I don't know if he's still there. And you can find these old press articles. I know because I actually read them as a kid. Funny enough, one of your other friends, Max Hodak, read the same articles as me. Now, the difference between Max and I, why he's your first repeat guest, is I said, "That's really cool," and Max said, "I'm going to build the same thing in my house." That's right.

He did it at Duke in his dorm room, I think.

Yeah. Yeah. So Tom DeMars had built this system where he could grow neurons on an electromicroarray, and then he was actually able to use them to keep a plane held constant in Microsoft Flight Simulator, which is actually a very accurate flight simulator of what it's like to fly a plane. Pilots use it to practice landings. Certain runways and things. And so I think we've known for a while there are both relatively simple training loops you can run on biological substrates, and of course we have the top-down evidence that the brain is a very efficient hardware system for running lots of compute. So I think it's really cool. I think it's an early-stage area of research. Maybe I lean a little more on the skeptical side. I've had a number of people pitch me on the notion of something, something, 3 years were going to replace Blackwell GPUs with neurons. I think probably the number of years between replacing Jackwell GPUs and neurons and today is greater than 3, potentially much greater than 3. Yeah. And I think these systems, what they've demonstrated thus far, is the ability to perform pretty finite tasks in a way where they're maybe more akin to an FPGA in the most favorable framing of it. I don't think they're even that advanced. And so I think we're still going to end up with the vast majority of the intelligence systems that are being rendered right now running on general-purpose accelerated compute, you know, things like the TPU or the GPU. I think we're quite far from using neurons to actually run those systems. But I'm glad that people are pulling at these early lines of research because we do have the existence proof in our own brains that lower-power compute is possible. And that's not essential for the largest AGI systems that might emerge in the future, but for edge compute, I think that's really interesting. You know, if you're ever going to run the most frontier system on your cell phone or your laptop, we're going to need to work on that power efficiency problem. And even if we don't use neurons to do that compute, can we learn some lesson about how the hardware should be structured from that system? I don't know, but because I don't know, it's a question worth asking. We should go dig.

So this company was using actual rat neurons. They did have like iPSC-grown rat neurons, but they were telling me the actual— for some reason they lean more towards the actual. Do you know why that would be at all?

I can speculate wildly, so just, you know, put a grain of salt on this. I think it actually goes back probably to something we described earlier, which is that making an adult cell type from an iPSC in a dish is really hard. We know the iPSC can make every adult cell type because you can clone a mouse based on one, and if you run development from an embryo and a uterus through pregnancy and eventually through, through postnatal development, it all works. So it's like not a cell problem. The iPSC IPSC is fine, but trying to provide all those cues in a dish is really hard. And so if they made an IPSC drive neuron, but it's like not really a neuron, it's kind of like a neuron in quotes or like a neuron with two tildes, squiggly signs on the sides. It's a neuron-like thing. Then potentially it's less functional than a real just primary neuron you can isolate from a cortex.

They were explaining to me that they can like create this model and then emulate it in software so we don't have to like harvest millions of rats. But this is what I pictured the entire time, was how bleak this gets.

It's like even worse than The Matrix. You're just like harvesting neurons in order to run computers.

You see what we do for— to get the GPUs. I was like, God, imagine if the, if the fucking rats were better, man. Good Lord, it's just like all of Wyoming would be dedicated to like breeding rats, you know? It's like, God, we will pay for our sins.

That would truly be wild. There actually was a therapy that some folks were working on where the plan was to grow it in rats. And then like for each patient you would have like 50 rats they isolated the therapy from. Yeah, they were going to make humanized cells, you know, similar to how we make human livers. They would grow cells inside rat livers and then isolate them all. And so the plan was like you'd have your own sets of cages of rats and you'd isolate your drug from it. It didn't get into humans, but I had similar thoughts of like, if there's someone watching this, this might be something they disagree with.

I mean, I would 100% want to go film all of this. This, but I, I, I'm fascinated by extremes. So, um, Jacob, thank you so much. I never really know how to end these, these shows. Um, I guess people can go to the New Limit website and see all the wonderful things you do. Please do.

You can see some news about the company. And if you know anyone who is a scientist or engineer looking for an important problem to work on, we're always hiring. So please get in touch.

Oh, there you go. We found something.

We found a plug.

All right, man. You're obviously one of my favorite people to talk to. Thank you for being so generous with your time, and thank you for coming to the studio.

Thank you. Really appreciate it.

The Core Memory Podcast is hosted by me, Ashlee Vance, and/or Kylee Robison, or both of us together. It is produced by me and David Mickelson. Our theme song is by James Mercer and John Sortland. And the show is edited always by the John Sortland. Thank you so much to Brex and Send Cut Send for all your support, and thank you most of all to everybody for listening or watching. We love you. Please leave us a like, a review, a subscribe, all those tremendous things. Thank you, and we'll see you again.