--:-- Host: Since the birth of this industry, we only had increasing regulation over time. I think there was only one time in history of biotech where we made it easier to develop an appropriate drug. So when George Yancopoulos started Regeneron, it cost about $10,000 per patient in trial. That's ballooned to $500,000. There is no law of physics that requires it to be $500,000 in terms of complexity and cost to dose a patient in a trial.
--:-- Host: Everyone will be using AI in biotech industry five years from now. Can it take $2.5 billion to prove a drug and make it into $500 million? Can it make it 4x more efficient in terms of timeline? To answer that question, we really should go back and look where most of that money is spent right now.
--:-- Host: We had this enormous wave of excess and now we're sitting on the other side of that. We have an enormous amount of EV negative public companies. There was a stretch of seven to eight months where there were no biotech IPOs. And so I'm really excited about that sort of opportunity to make things that are just net new in the industry. And that's where you have to go to really keep winning.
--:-- Host: The biotech paradox. One fifth of public biotech companies are trading below their cash balances. Seed rounds have hit record lows. The industry spends $2 billion per approved drug, and that number keeps climbing. And yet, we're designing antibodies from scratch with AI. We have drugs that are bending the curve on aging. The science has never been better. So why is the business of biotech collapsing while the technology is exploding?
--:-- Host: Today's guests have spent years studying this contradiction. Lada Nuzhna founded General Control, a startup tackling aging after writing the definitive analysis of why there are no trillion-dollar biotechs. Elliott Hershberg is a partner at Amplify, betting on the next wave of platform companies while watching American biotech companies flee to China and Australia just to run their first trials.
--:-- Host: Their diagnosis? We're competing on the wrong axis. China wins on speed and cost. The FDA adds friction while innovation accelerates. And the entire industry is structured around an equilibrium that no longer exists. But there's a path forward if we're willing to invent our way out. We cover why regulation might not be the real problem, what GLP-1s reveal about blockbuster drugs, whether AI can actually fix drug development, and why the next iconic biotech companies will look nothing like Genentech.
--:-- Host: This is the state of an industry at an inflection point. Let's get into it. Welcome, Lada. Welcome, Elliot, to the a16z podcast. Very excited to have you here. And I'm very excited to have this conversation because where I'd love to start is I want to talk about all things biotechnology. And let's take a quick pulse check on the state of biotechnology as it stands today in 2025. So question one for both of you is how do you see the state of the industry as it stands today? And how do you see the state of the science as it stands today?
--:-- Lada Nuzhna: Yeah, if you put AI buy deals aside, it's been quite a few interesting years in biotech. Earlier this year, we had something like one-fifth of public biotechs trading at or below their cash balances. Record low number of biotechs raising their seed rounds. The platform dream is judged more harshly. And if you take a step back and look at the bigger picture, and ask yourself, has it ever been different? It seems like this state has been ongoing for quite some time.
--:-- Lada Nuzhna: We have Eroom's law, which is Moore's law but backwards. We now spend more than $2 billion per approved drug. We have the rise of Chinese biotech threatening the state of early-stage biotech companies. And I guess the question is, did we expect it to be different? Because approving drugs is really hard. It's probably one of the hardest areas of deep tech broadly because it takes on not only engineering risk, but also scientific risk.
--:-- Lada Nuzhna: It does seem like the biotech industry started on a more positive note a few decades ago. We had Genentech, Amgen IPOs, their investors making huge exits, more so than in tech back then. And we had the Human Genome Project that sort of promised that all of biotech would become precision medicine and we'd develop medicines tailored to one patient. And it does seem like it led us to a slightly different ending. I hope it's not the ending, and I hope there are some positive things we can find in it all. But in many private conversations during this past two years, the mood has essentially been: will this industry ever recover?
--:-- Host: Okay, so you have a lot of reason to be pessimistic.
--:-- Host: At least as of now.
--:-- Lada Nuzhna: As of now, as of today.
--:-- Host: I want to hear Elliot's take.
--:-- Elliot Hershberg: I think there's never been a bigger disconnect between both sides of the market. As Lada was saying, we had this enormous wave of excess. Our last actual boom cycle in this industry was COVID. You had an enormous amount of value creation and success. Now we're sitting on the other side of that, and we're objectively in a downturn. We have, as Lada pointed out, an enormous amount of EV negative public companies. That's starting to change and resolve a little bit, but there was a stretch of seven to eight months where there were no biotech IPOs. There's a logjam that goes from the public markets to the growth investors to the late-stage investors to the early-stage investors that just changes milestones and makes everything super hard.
--:-- Elliot Hershberg: At the exact same time, there's never been a better point to actually see the progress of technology. You have people developing zero-shot antibody design in labs. You have incredible virtual cell projects. An enormous amount of translational potential at the early stage. For an investor or a founder, it's an incredibly exciting time to be building a new company because there's this huge disconnect in the market between the public markets and what's happening earlier. It's a tale of two worlds.
--:-- Host: What I was going to say is, I think what's interesting is there are a couple of reasons to be optimistic today in terms of the market at least showing some of the famous green shoots of optimism. Some IPOs are happening, or at least people are preparing to file. Some very successful M&As. The biotech index is starting to get well above the famous $100 mark for the XBI. So there's reason to be optimistic.
--:-- Host: I think the fundamental question is what bridges that disconnect between all of the advance and promise we're seeing on the technology side and some of the fundamental laws of physics of the industry that have resulted in this not being a particularly effective industry to invest in over the course of the last several years. You wrote a great post, "Where Are All the Trillion Dollar Biotechs?" What's your diagnosis of the situation? What is it about the fundamentals of the industry that make it so challenging economically?
--:-- Lada Nuzhna: Since the birth of this industry, we only had increasing regulation over time. I think there was only one time in history of biotech where we made it easier to develop appropriate drugs. That was during the AIDS crisis, where AIDS patients were just laying outside of the FDA and demanding drugs be approved on a more accelerated timeline. The FDA did make a big change back then that allowed for certain accelerated approvals. But since the birth of the industry, maybe for good reasons, we were always making it only harder to develop new drugs, all while the science was improving over time.
--:-- Lada Nuzhna: Our high-throughput screens for new small molecules and antibodies are a billion times more efficient than they were 20 years ago. And what Elliot said really resonates because it feels like the science continues to get better and the state of the industry continues to get worse. I think a lot of it is downstream of regulation. The way the FDA started to crack down on the drug development approval process is through a huge tragedy. I think it was the drug thalidomide that pregnant women were taking that led to a number of deformities in children being born. Since then, the FDA started to require not just safety, but also efficacy in the approval process. Whoever would come to the FDA to deregulate this process would have to take on enormous personal risk because regulation is a tension between safety and efficacy. If we deregulate the process, we would have to take on some type of safety risk.
--:-- Elliot Hershberg: I think it's interesting. There's 100% of growth of regulation. There's a bunch of low-hanging fruit that would be substantially better for starting trials. We have three companies in the next 12 to 18 months that will be starting first-in-human trials. Guess how many are actually doing their first studies in the United States?
--:-- Host: When you say "we," you mean?
--:-- Elliot Hershberg: At Amplify.
--:-- Host: Oh, none. Zero.
--:-- Elliot Hershberg: Right. We've just come to accept the fact that everyone goes to Australia, everyone goes to Asia to do their studies because the things that we invent here, we can't actually first test here. So there is a regulatory barrier. But I was writing about a business that was trying to innovate on a clinical research organization. One thing I learned in terms of the machinery of translating regulatory innovation into actually cheaper trials is: let's say there's a change, whatever variable we think about for trying to decrease the costs of trials. How does that get implemented?
--:-- Elliot Hershberg: It turns out there's been an enormous consolidation of the clinical research organization market that's come down to about a dozen providers who have each done roughly 40 acquisitions across 30 years to consolidate into these large clinical outsourced providers that most people rely on. So when the FDA says we want to modernize the standards for trials and have electronic tablets, you actually have to get the clinical research organizations to adopt those tools and technologies. And they aren't. They aren't incentivized to. So there is a structure to the actual process of executing on the trials that lags behind the changes in regulation itself.
--:-- Elliot Hershberg: It's a two-pronged beast when you think about the cost of clinical development. It's what the rule of law is and what you can do, and then also all of this industry entrenchment of what we are doing. I think there are some ways for people to just do things a lot faster, even within the bounds of the law, than we're used to paying for or expecting. That's in part just a cultural component of biotech: we just assume that these things are expensive and take a lot of time.
--:-- Host: So that argument would be that the challenge is not technological. On some level, the challenge might not even be the regulatory bodies. It might be something structural, something incentive-based, where the way the industry is structured, you have various parties—in this case, the groups that help run the clinical trials—that may or may not be incentivized to be more efficient. Is that the argument?
--:-- Elliot Hershberg: That's right. If you were to break it down into cultural, tech solutions, and just implementations and regulatory, you'd have to assign some percentage to each. I think it is a composite of all three.
--:-- Host: Okay. So now let's add to the mix. Let's add to the soup. The China question. For folks that may not spend all their time thinking about biotech, what is going on in China as it relates to the biotechnology industry? What do we see as the impact that is having already and may have in the longer term to the U.S.-based biotechnology industry?
--:-- Elliot Hershberg: I think it's worth backing up and saying, how is the biotech industry set up in the first place? How does this actually work? At the very outset, in about the 19th century, there were chemical companies that decided to get into manufacturing chemical drugs. These were actually dye manufacturers making dyes for textiles. They decided to take their chemical manufacturing and distribution and apply it to making drugs.
--:-- Host: You're talking about the Germans.
--:-- Elliot Hershberg: That's right. The first drugs were incredibly crude pharmacology. This was heroin, this was cocaine, this was morphine—higher retention than dyestuffs. That's a pretty interesting business. But obviously it got a lot more sophisticated and specific over time. These groups like Merck, in the 30s, set up vertically integrated research labs. They're in the business of manufacturing and distributing their medicines and also doing internal research to make new products. Then the FDA comes along. A lot of these top pharma companies predate the FDA and modern clinical development.
--:-- Elliot Hershberg: There's a three-part component for all these businesses: manufacturing, commercialization, and distribution; internal research; and then trials and clinical development. What happened is that, as Lada alluded to, there's this Eroom's law in the industry where it's just gotten exponentially less efficient to do the internal research and clinical development. It literally became the case that it was IRR negative to actually do internal research. So these pharma companies divested their internal research and clinical development out. This is the birth of clinical research organizations and outsourced organizations. The origin of drug discovery research startups and biotechs is that we are the lunatics that take on that IRR negative business and say we're in the business of making these new hits that could ultimately get bought by pharma companies and become their next armamentarium of drugs.
--:-- Elliot Hershberg: So that's the setup. These companies divested their R&D, and biotech companies are doing that research. What happened is that we're all doing that research with the same discovery technologies. I'd argue that beyond the modernization of small molecule discovery and the growth of biotech and recombinant DNA technology, we haven't had huge changes in subsequent modalities. People don't have internal technologies that the rest of the industry isn't using. Like any other technology, when it becomes commoditized, you can compete on speed and cost. This is where China enters the story.
--:-- Elliot Hershberg: China has enormous speed advantages in terms of regulation for starting and running human trials, and enormous cost advantages in terms of labor, work ethic, speed, and volume of people that can be put together on these projects. In this loose social contract between big pharma and biotech startups—which supply the drugs, biotech startups now make two-thirds of the drugs that go to market—we have this geographic arbitrage where on speed and cost, China can enter the equation and compete for these types of ideas to deliver the medicines.
--:-- Lada Nuzhna: What's interesting is China didn't start in this place 10 years ago. 10 years ago, no one was talking about going and running their trials in China. In fact, they were more regulated 10 years ago than the US is now. What really happened is several waves of deregulation that produced the current modern clinical trial infrastructure that everyone goes to Shanghai for. Some of the things they implemented were implied approval: when you file your IND, unless they issue a proactive hold on your IND, it will be default approved in a certain number of days. The US has the opposite model where you have to proactively approve every IND document that comes to the FDA. They also parallelize the review of different components of that IND. You can review the CMC section and the clinical trial design all in parallel, while in the US you have to review them in stages.
--:-- Lada Nuzhna: What's even more interesting is this whole model of investigator-initiated trials, which is actually what most people go to China for. The actual CFDA process is more efficient than in the US, but I don't know if it's more efficient than just running trials in New Zealand or Australia. With investigator-initiated trials, your review timelines are cut 5 to 6x. It's a very fast process, and it's really specific to new modalities and high-risk indications like cell and gene therapy. That's what everyone goes to China for.
--:-- Lada Nuzhna: The old model of viewing China as a place to buy a printer where they print a medicine that would already otherwise exist is somewhat outdated. They are now leading in vivo carts. They are leading in gene editing and gene therapy, partially due to this ability to do those investigator-initiated trials.
--:-- Host: Okay, so given that that's a reality today, what does that imply for the long-term future of the US biotech industry? Because you can make the argument that if we're competing on scale, speed, and cost, and there's still a lot of great innovation happening here in the US, at some point it becomes difficult to fund that innovation here. We talk about things being IRR negative. It becomes increasingly difficult to fund that innovation here if we know it's going to get outcompeted over the longer term in China. So does our industry get hollowed out?
--:-- Elliot Hershberg: No. I think the answer is somewhat simple. I think we have to invent stuff. There's a really good China analyst named Dan Wang who just wrote a really good book on the whole structure between China and the United States right now. The one-sentence meme version is that China is an engineering state and America is a lawyer state. I think that's simplistic—that's not Dan's full argument. But we forget that we're also an incredible inventor state. Some of our founding fathers, we have Benjamin Franklin as one of the US's founding fathers. We have the American research universities that are the envy of the rest of the world. We're incredible at going from zero to one.
--:-- Elliot Hershberg: If you're talking about something where it's fast followers and a specific arbitrage opportunity, that's super challenging in terms of speed and cost. But if you're talking about inventing totally new modalities and growing the pie—things on the scale of the recombinant DNA revolution, which is fundamentally a US thing, or the birth of immunotherapy, which is in Texas right now—these are the types of things that get us out of it. We have to invent our way out and actually change what's possible with biotech. When you get those big unlocks, that's when you start to really deliver value.
--:-- Elliot Hershberg: It does shift the risk profile. Part of where you go is really pushing the boundaries of new modalities. I spend a lot of time with Michael Fischbach, an incredible scientist at Stanford. We're always talking about what's the next big interesting set of modalities, what's the new interesting mechanism in biology. That risk profile shifts. At Amplify, that's what we always think about: those technical founders. Because it starts to be the case that it's like the Regenerons of the past—Leonard Schleifer and George Yancopoulos, these scientists that are actually going to tell us where the future is.
--:-- Lada Nuzhna: I do wonder, though, to what extent—I think it's a very compelling argument, obviously I'm very aligned with it. The counter argument would be something you said a few minutes ago: there was this nice equilibrium that the industry has had for decades, which is US biotech invents and global biopharma invests. They in-license it, they acquire it, they partner with it. That equilibrium basically meant that the inventions that come out of biotech had time to develop. But now you have a third player in the form of China that could disrupt that equilibrium meaningfully. So yes, we can invent the next great modality, but we have to implement it faster too. Effectively, the shelf life of an innovation has gotten far shorter before we deal with competition.
--:-- Elliot Hershberg: It changes one of the dynamics, which is secrecy. In the industry, secrets become more important. When you want that longer time horizon to actually invent something, you probably need to keep it a little bit closer to the chest initially. I do think there's some nuance in how we think about regulation for this. It's probably not Biosecure, where you are totally trying to restrict relationships and transactions between the US and China, because that's just net negative. But if there are really direct fast followers, and that half-life on an initial invention is shrinking, we might need to rethink how to extend that horizon for inventions.
--:-- Lada Nuzhna: To push back with one example, I keep thinking about this professor, Irv Weissman, a legendary cancer biologist and stem cell biologist from Stanford. He launched a few companies into existence five years ago. He published a new paper on a new cancer targeting mechanism, which is what we think US biotech should differentiate themselves on: new biology, new modalities. He immediately started a company, well-funded. They developed the asset, were taking it to the clinic. And before they were able to initiate their clinical trials, a Chinese biotech with the same identical mechanism beat them to the clinic and launched trials, not only in China, but also in the US.
--:-- Lada Nuzhna: I do think that secrecy really matters. I think many people are starting to ask themselves whether they want to present at conferences, whether they want to publish papers, whether they want to file patents. That wasn't the case when Genentech was created. Back then, people were just putting everything out there for everyone to see and share. That's a real negative externality in terms of the mean substrate of communicating science, of open source, of this proliferation of technologies. A lot of technologies are built on other technologies. If everybody has to keep it closer to the chest because your time to actually commercialize your own invention shrinks, that's a consequence of this trend that we have to think about.
--:-- Host: Speaking of inventions, one of the great inventions we've seen over the course of the last few years is the rise of artificial intelligence. AI is obviously being heralded as transformative across a broad range of industries and applications, creating new experiences, new consumer products, all kinds of things. Lots of use cases we would have never imagined. Where do you view the impact of AI when it comes to developing new drugs? Is that the invention that actually makes biotech competitive again, that makes biotech an investable asset again?
--:-- Lada Nuzhna: I think it's not the question of whether AI will be useful. I'm definitely in the camp that everyone will be using AI in the biotech industry five years from now. For me, it's more a question of: can it take $2.5 billion to prove a drug and make it into $500 million to prove a drug? Can it make it 4x more efficient in terms of timeline? To answer that question, we really should go back and look where most of that money is spent right now. Most of it is not spent in the preclinical stage. It's spent in validating safety and efficacy in humans. It's spent on the commercialization stage, which is what happens after Phase 3 clinical trials.
--:-- Lada Nuzhna: So far, and maybe Elliot's take would be different from mine, a lot of the effort we are seeing is concentrated on the preclinical stage: doing toxicity studies, making toxicity studies in mice much faster, having in silico tools for cell lines. A lot of those things are valuable, but they don't necessarily bridge the gap. They don't necessarily improve the failure rate of clinical trials, which is highest right now at Phase 2—efficacy. If I started to see those same companies use and generate more human data to apply to those models, I think I would become more optimistic. I actually think many big bio problems and open questions will be solved with AI way before we can predict efficacy for drugs. But I think virtual cells is actually not that far out. I think we have all the necessary data to generate something like that. The question is how can we make it useful for predicting efficacy of drugs in humans?
--:-- Elliot Hershberg: I think Lada is right. The answer is probably unequivocally yes. It's becoming consensus that AI is a pretty important new experimental tool for biology and biologists, just as software was. The way I like to evaluate where it's useful right now is if you break down the three horsemen of Eroom's law: the time and cost of clinical development. That's an enormous pillar that's hard to tackle. We talked about that: tech, probably software 1.0, can help a lot there. Regulation, culture.
--:-- Elliot Hershberg: One layer deeper, you have the challenge of Phase 2 failure, which is basically a readout on the fact that we don't understand biology.
--:-- Host: Right. And we can't predict what's going to work and what's not going to work.
--:-- Elliot Hershberg: Exactly. Efficacy prediction on net new targets or hypotheses for a mechanism. Then the third is that there's an enormous amount of interesting ideas that we have, but we actually can't express those ideas in molecules and make new drugs. That's the third pillar of why it's really hard to make exciting new medicines.
--:-- Elliot Hershberg: For all the reasons we talked about, it's hard to dramatically change the time and cost if you're doing preclinical discovery when it comes to clinical development. But when it comes to making things that are otherwise impossible medicines—either finding really interesting new targets and having higher confidence in predicting the efficacy of the drug—there are some exciting directions. Especially when it comes to expressing ideas in molecules, I think a lot of these platforms and capabilities to take new interesting data sets, to take what's coming out of molecular machine learning, and just make things that are unequivocally impossible to make without these tools, makes some really exciting medicines.
--:-- Elliot Hershberg: I think the ambition of a TPP will go up. I think that's also really important. The portion of Eroom's law is predicated on the "better than the Beatles" problem. We keep adding to this armamentarium of medicines that we have, and it stacks up over time. We're continually in pursuit of beautiful medicines. To actually make something more beautiful and more potent at this stage of the game, the alpha is probably in really interesting data sets and new modeling tools and capabilities. I'm really excited about the categories of medicines that could be made with this type of approach.
--:-- Host: So let's talk about the three horsemen of Eroom's law. I like this, the way you've framed it. The structural problem we have with getting drugs approved is a big horse.
--:-- Elliot Hershberg: Totally.
--:-- Host: The lack of understanding of biology and the ability to predict efficacy seems like a very big horse. Is the ability to—are we design-limited in our ability to make molecules? Is that a big horse or is that a pony?
--:-- Elliot Hershberg: I think it's pretty substantial. We've gotten really good at making monoclonals. We've been making monoclonal antibodies for 50 years, half a century. We've got exquisitely good experimental tools for panning and finding these types of molecules. If you have a specific, no-nonsense target, that's fine. But if you're talking about a really complex, polyspecific molecule that's hitting multiple components—you see the beauty of multivalency with PD-1-VEGF. What else is out there that is two, maybe three interactions that exquisitely tunes the immune system or the state of a cancer? As we learn more about moving cells around on their manifolds of different cell states, I think there actually are target product profiles and medicines that we just genuinely can't get to with our existing discovery technologies that could be really big products. You see this very low-hanging fruit: without this tool combination of PD-1 and VEGF, that's beating Keytruda, right?
--:-- Elliot Hershberg: Like that's an enormous step change. Our ability to potentially get those types of results from these models seems possible to me.
--:-- Lada Nuzhna: Yeah, I think there are many targets on pharma's most wanted list that have been on that list for many decades now. I mean, targets like p53 that so many companies tried to target, and no one really managed to get the molecules that really work. Or it worked, but the target product profile wasn't desirable. Until we clear out that list, there's definitely a bottleneck on the ability to design certain therapeutic modalities. The question is, do we always want to have an oral for every antibody that exists out there? Maybe injectables are actually not that bad. I think the whole GLP-1 story definitely altered my perception of how much people are willing to do injectable drugs on themselves.
--:-- Elliot Hershberg: If you take an injection every quarter and you don't have to think about it, you don't have the adherence risk of a small molecule. I think that's totally right — that's not always the canonical rule of thumb.
--:-- Lada Nuzhna: It's also true that new modalities often struggled, even though they were better in some shape or form. The whole generation of CRISPR companies — I don't think we've really seen through the CRISPR dream quite yet. Ten years ago, when the first CRISPR companies were launched, the dream was that we would develop a tailored gene editor for every rare disease that exists out there. It's partially bottlenecked by regulation, but even for well-known targets like PCSK9, the question is, do you really want to do a gene editor, or is Repatha or Inclisiran actually good enough? Many new modalities will be bottlenecked by the "better than the Beatles" problem.
--:-- Elliot Hershberg: It is a fascinating thing because the armamentarium has gotten so large that now you're really slicing at specific patient preferences. Someone may prefer an oral to a weekly injection. Someone may prefer an infusion once a year to taking an injection every month. It starts to get very challenging to tease out which product profile wins in the market. This is where I'm sort of an unequivocal platform bull still. I think one of the most exciting opportunities for AI and biology is this world where the platform is the product. If you think of Moderna's and BioNTech's cancer vaccines, this is a product in clinical trials that is part next-generation sequencing, part AI and machine learning. There's actually a neural network that processes the sequencing data, and then there is a specific mRNA cancer vaccine designed for that patient. In that case, it's hard to even disentangle — it totally breaks our conception of what the target is, what the drug is. The drug is part information product, part diagnostic. That type of really personalized approach could open up and actually be a total pan-indication solution, in the limit. One of the really exciting things is, if you have a fundamentally generative platform that is the product, does that open up a much wider indication base?
--:-- Host: Speaking of indication bases, big indications — you talked about GLP-1s. Obviously one of the most extraordinary drugs in terms of the impact it's had on broader society that we've seen in a very, very long time. It's also a drug that, if we talk about healthspan, longevity, all of these things, it might actually be bending the trend in terms of some of these chronic diseases, whether it's metabolic disorders, obesity, and the like. In your piece on "Where are the trillion dollar biotechs?" you talk a little bit about where the next big wins for the industry will come from. You talk about finding diseases based on genetic basis, rare diseases where you can have a big impact. But if I've read your conclusion correctly, where you land is the real big nut to crack for the industry is to go after the disease of aging and to solve longevity to the extent we can. What's your view on what the industry is doing right and where the industry is still lacking when it comes to all things aging and longevity?
--:-- Lada Nuzhna: Unfortunately, the incentive to develop aging drugs is still not quite there. If you look at the U.S. payer system, the payer that pays the most for diseases of the aging population is Medicare, which kicks in after 65. Before then, we have a multi-payer system where patients tend to rotate their insurance every few years. There's really not much incentive for someone to cover preventative medicine early in life. Once you hit 65, it's no longer preventative medicine — it's treating the disease itself. A lot of this would be downstream of fixing how we pay for aging drugs, how we pay for preventative care. I don't know if we have a way to do that now, not for chronic medicines, especially not for one-and-done solutions to age-related diseases.
--:-- I am very excited about GLP-1s. I'm probably not the first person in the aging space to say that maybe GLP-1s would be one of the first aging drugs. I think this month, Lilly is reading out their semaglutide in Alzheimer's trial, which to me is a real test of whether it's an aging drug or not, because it's well outside the metabolic spectrum of diseases. At the same time, Medicare refuses to cover GLP-1s for obesity care. So can we cover GLP-1s for aging diseases? It's not super clear to me.
--:-- Elliot Hershberg: There's one interesting component of incentives where it's a commercial "better than the Beatles" problem. Once you have such enormous revenue and sales generation from a product, you're going to fight like hell to have something in your pipeline that could potentially replace that. It has this property of driving and pulling on the ambitions of the industry. People are honestly thinking at Lilly and Novo: what is it going to take to actually fill the patent window? What's the act two? It has to be something that is a large enough indication to take that vision seriously.
--:-- Host: It's a great argument that GLP-1s have really achieved two important things. First, they've potentially given us ways to treat some of the most endemic, challenging conditions that affect society, whether it's metabolism and obesity and other factors. And second, at some level, it's given the industry its mojo back — to go after big problems, to go after the big indications, to find the act twos. Because you're absolutely right, you're going to have to replace this product with the next big idea. Alex Telford's written a great piece on the cyclicality of trends of what produces a blockbuster. You had the Lipitor era of these enormous pills for big indications that were phenomenal products. Then we moved into this era of specialty medicines, the birth of biologics, a very big focus on rare diseases — making up for smaller population sizes with larger price tags. With the general pressure on pricing plus the carrot of the success of GLP-1, there is this big swing back into big indications. Potentially there are such big indications that you have to take being a direct-to-consumer business really seriously. You think of Lilly Direct — these are so big that you'd actually break the payer system if you were distributing something for weight and metabolism for aging. The arc of our business model — I did not expect this year to have John Merriwether talking about a direct-to-consumer biotech company. But that's kind of where the vibes are right now, which is actually exciting.
--:-- What is the state of the science when it comes to aging? You talk about some of the structural challenges in terms of reimbursement and maybe even regulatory. But let's go all the way back to the beginning. Do we understand what aging is? And do we have credible theories as to how to intervene?
--:-- Lada Nuzhna: I wouldn't say we know what aging is or even how to measure it. If we had a way to measure it, we would have clinical trials run based on surrogate endpoints, and maybe we would have multiple agent clinical trials run in parallel. Right now, the way we move in the direction of approving an aging medicine is we run multiple trials for multiple diseases, the way we are doing with GLP-1s. Then we derive the conclusion that if it delays onset of multiple diseases at the same time, maybe it's an aging drug.
--:-- I do think regulation is lagging behind science. We have multiple drugs that extend lifespan in mice and monkeys that have never been tested for lifespan indications in humans. Some companies are doing very exciting regulatory groundwork — companies like Loyal, where maybe for the first time we would have an aging drug approved for dogs. Maybe it's not that far out for approving the first aging drug for humans. I think we would be able to treat aging before we understand how to measure it or what it is or why it happens.
--:-- The way I view the future of lifespan drugs is that it should come in several waves. The first wave is small effect sizes, very established therapeutic modalities — small molecules. If it's preventative medicine, it has to be squeaky clean, very safe, because if you're preventing some future disease, there is no room for side effects. After that, we would have more exciting therapeutic modalities — maybe genetic editors, gene editors, maybe gene therapies. As we progress, the variance of medicine and therapeutic modalities that we apply to aging will increase over time.
--:-- Host: If you guys could wave a magic wand, what would be the aging stack that you take every day? You know, the Brian Johnson "Don't Die" protocol. In your mind, what should the average person be thinking about that they should be taking on a regular basis?
--:-- Elliot Hershberg: It does seem like we've hit a point where our generation shouldn't die of heart attacks. We have everything to prevent high cholesterol — you can take antibodies, siRNAs, small molecules, gene editors. We have a full stack for that. Heart attacks are the primary cause of death in the United States. People die around 72. To remove that, people would start living to 75, maybe 80. A good benchmark is Japan. In Japan, people don't die of heart attacks, they die of cancer. The median lifespan there is close to 80, so that's plus 10 years to lifespan. Most of the things would be things that are already approved.
--:-- Lada Nuzhna: GLP-1s are obviously a big one. The only documented effect for lifespan that we have in monkeys is caloric restriction. We know we can add about two and a half years to a 25-year median lifespan in monkeys by calorically restricting them. It really depends on the controls used in the study. There were two big studies. One used monkeys on a high-fat diet as control. The other used healthy monkeys. If you compare it to healthy monkeys, caloric restriction doesn't add that much. But if you live in the United States, you're likely a monkey on a Western diet. So I think GLP-1s will be broadly impactful for everyone.
--:-- Host: All right, so magic wand — you put Lipitor and GLP-1 in the water.
--:-- Lada Nuzhna: PCSK9 inhibitors and GLP-1s in the water.
--:-- Elliot Hershberg: It seems to be the case that caloric restriction is important. Moving — we're a very sedentary society. There's a lot of benefits in just being active. It's interesting seeing the level of personal health monitoring downstream of whatever Brian Johnson's cultural movement is — people actually doing a lot of longitudinal self-measurement, doing blood work, being more proactive in their care. Even being more proactive in cancer care — early stage screening and having types of medicines with the right risk profile to actually dose people with if you are able to detect extremely early stage cancer. Right now, it's an ethical question whether our fairly barbaric approaches to cancer care actually have a meaningful ROI. If we had different types of medicines, that would be pretty phenomenal for longevity and health.
--:-- Host: Speaking of magic wands, we've spent a lot of time talking about the industry and some of the challenges it faces in getting drugs to patients. If you had a magic wand and could change something around the regulatory environment, or change something about the laws of physics of the industry, what would those be?
--:-- Lada Nuzhna: I think the cost per patient per trial should almost be a stat that the FDA cares about. When George Yancopoulos started Regeneron, it cost about $10,000 per patient in a trial. That's ballooned to $500,000. There is no law of physics that requires it to be $500,000 in terms of complexity and cost to dose a patient in a trial. If we want to see the next Regeneron, we need to meaningfully care about that as a KPI for regulation and for the industry.
--:-- I also think it should not be accepted as the default that innovative American companies go to other geographies to run their first-in-human studies. There is low-hanging fruit when you think about the regulation in place in Australia and Asia. Carl June, the early developer of CAR-T therapy, was asking this in a workshop recently with the FDA: why do we not have investigator-initiated trials for cell and gene therapy in the U.S.? If there is this distinction of them being the engineering state and us being the lawyer state, we should actually say, let's win on regulatory innovation. Let's be really creative in the way we regulate this industry. We should still be the beacon where people do their clinical development and where trials are approved.
--:-- Elliot Hershberg: The reality is that even if you run your trial in China, you still have to come back to the U.S. Even Chinese companies that run their trials in China all come back to the U.S. because it's the biggest market. Even if the Chinese population continues to grow, the U.S. will still remain the biggest market for biotech companies to exit at. There is no Chinese pharma that people are selling to. All of the pharma that are buying Chinese assets are U.S. pharma, European pharma that would then go and run those trials here. So I do think we need to solve that question, because eventually everyone will be running trials here either way.
--:-- My magic wand — I would cast a spell and ask for something that would be a version of orphan drug designation, but for common diseases. The Orphan Drug Act was enacted around the 1980s. Before then, we had less than 40 approved orphan drugs for patients with a population of less than 10,000 patients. Today, 50% of drugs approved in 2024 were all orphan drugs — for very, very small populations. I think we are at the stage of biotech development where we really need something like that to incentivize development of drugs for age-related diseases or for longevity itself. Right now, age-related indications have some of the highest failure rates in drug development because there are no genetic drug variants, the process is much longer, the trials are way more expensive. We need some kind of incentives to make sure that more biotech companies go and develop drugs for cancer where Phase 1s kill half of the companies at early stages.
--:-- Host: It's an interesting concept — an orphan drug designation for more chronic disease. The original spirit of the orphan drug was to create an incentive for people to develop drugs that face small populations because the market potential isn't there. But in this case, for chronic diseases, the market potential is enormous. So in your mind, the thing that needs the most incentive is to incent companies to go through the difficult development process because the failure rate is high.
--:-- Lada Nuzhna: Yes. There is a little bit of disconnect in what types of diseases affect humans and what types of diseases we approve drugs for. If 50% of drugs are approved for rare diseases and rare diseases affect only a small fraction of the population — I think that's important — but how about every disease that people are actually dying from? The fact that the population is aging should be a big push to do something like that, because an aging population is a less productive population, and that's where the U.S. has had it now.
--:-- Host: Okay, so that's a great argument that if you could address aging in a meaningful way, you could have massive societal benefit, and you have to find a way to incent that because it's not happening today. How about your magic wand?
--:-- Elliot Hershberg: If I had a magic wand, it'd be a combination of figuring out how we incent the innovation that happens here to stay here. A bit of what Lada was describing. We have this wonderfully effective pipeline where a lot of innovation happens in universities and in startups that get funded through investors like us. You get a lot of incredible novel approaches to tackling disease from that. But the challenge is going from that invention to an actual product still takes a lot of time and money. There are a lot of hurdles. I like this idea of being able to say, why don't we see what works in the rest of the world in terms of running that relay race from invention to an approved drug more quickly? Can we copy those processes to make sure our lap time is at least as fast as the rest of the world's? If we do that, I think we will find that we can maintain a lot of innovation here.
--:-- One thing that is promising is that the regulatory agencies, the FDA, are at least signaling that they want to find ways to really innovate, to modernize. I think that's very promising. When it comes to some of the other challenges we have geopolitically — whether it's with China or just in the rest of the world — what a lot of the administration is pointing to is saying, how do we incent innovation to stay here? How do we incent the supply chain to stay here or to re-onshore? If I had a magic wand and could replicate all of the things that are working elsewhere and bring them here to get us back up to speed in terms of being able to run the race as quickly as other countries can, I think that would be extraordinarily promising for the industry, for society, and arguably for the world. That is my reason to be optimistic, and that would be what I would use my magic wand for.
--:-- Host: The GLP-1s as a drug class are this incredible example. I've heard it described as the most important consumer product that we've seen in the last several decades, for obvious reasons — the impact it's having on societal health. What makes a drug blockbuster in your mind? Why is it that some drugs are so incredibly successful? Why don't we see more of them? People tend to generate certain wisdoms around what it takes to develop a successful drug. Sometimes it's "you have to be first in class or first to market." But I think successful drugs are kind of like the opposite of that Tolstoy wisdom — each happy family is happy in its own way.
--:-- Lada Nuzhna: In the case of GLP-1s, semaglutide and tirzepatide weren't the first GLP-1s to be developed or approved. In this case, Lilly and Novo took on the very contrarian bet that obesity is actually a real market, which now seems obvious. But 10 years ago, if you were a company trying to raise for obesity as a disease, you probably weren't having much success. Many companies terminated their obesity programs because it wasn't clear that the market was there, and especially it wasn't clear that people with obesity would be injecting themselves with drugs. Pfizer terminated their GLP-1 program because internally they decided that chronic disease injectables — they didn't think patients wanted that.
--:-- Humira was also not the first TNF-alpha antibody out there. It was third to market, the third TNF-alpha antibody to be approved. But it was the first human monoclonal antibody, one of the first to be approved. All the previous ones were antibodies from mice. In both of those cases, it wasn't a biological take that was unique. The targets were pretty consensus. Biology is one of the areas where you really don't want to be contrarian. You don't want to be the only company pursuing some obscure mechanism. Usually you want to have some literature validation. But in both of these cases, it was either a big modality differentiator in the case of Humira, or a big contrarian take on what indication to pursue.
--:-- I think aging might be a contrarian indication to pursue for some. I think muscle is in a similar space right now. For a while, people weren't treating sarcopenia as a real indication because it's muscle loss in elderly. But there is now a similar race to GLP-1s to go and develop drugs for muscle. We'll see if it succeeds.
--:-- Host: So you, with all the time you've spent studying and focusing on the industry, have decided to jump into a startup — still stealth as I understand it — that's going to tackle aging. What are you thinking?
--:-- Lada Nuzhna: Our take is more of a modality take. Every time I think about where the biggest breakthroughs and successes in biology came from, it was always served from some type of technology or process or technique, and rarely from discovering a new target. We are developing a new modality that should make it easier to tackle aging. If you look back at something like the Human Genome Project, it took us several decades and several billions of dollars to sequence one human genome. Once we discovered better sequencing approaches, we can now do it daily for a few hundred bucks. I think aging is in a somewhat similar space in that it's a massive multifactorial disease. If we rely on existing modality approaches, it would just be an uphill battle to try and treat it. If we develop modalities that allow us to target that complexity without much additional engineering every time we want to start a new program, I think that would be a big catalyst for success.
--:-- Host: We started this conversation talking about where the trillion-dollar biotech companies are. You are both students of the history of this industry. Where do you see the next wave of iconic biotech companies coming from? Obviously the industry started with the Genentechs and the Amgens and the Biogens of the world. Then we eventually got the Vertexes and the Regenerons of the world. Where do you see the next wave emerging?
--:-- Elliot Hershberg: I'm a big believer in modalities. If you look at the history of the industry, there is an enormous amount of value created from unlocking new types of medicines. I'm really excited for the fact that you have all of these new generative design tools and sequencing technologies and delivery tools that can all start to be stitched together into a composite specific product. There are waves within technology where specific problems are solved and you start to bundle a bunch of different components together. Then new ideas come and it unbundles the stack. This happens across software and different markets. I think in biotech we're in a moment where there's a lot of opportunity in rebundling. The types of platforms that I'm super excited about are this composite of incredible synthetic biology and genomics tools, plus modeling, plus other tools on top of them that unlock things that otherwise weren't possible. I'm really excited about that opportunity to make things that are just net new in the industry. That's where we have to go to really keep winning.
--:-- Host: So your bet is the next wave of great companies will come in the form of some type of new modality.
--:-- Lada Nuzhna: Recombinant DNA gifted us the first off-the-shelf insulin. mRNA vaccines gifted us vaccines that we can synthesize in less than a month produced from the genomic sequence of the virus. First human monoclonal antibodies gifted us a wave of cancer precision medicine. Something similar has to happen for chronic multifactorial diseases. We are starting to see light at the end of the tunnel with gene editing, epigenetic editing, more tailored targeting approaches where no longer do our LNPs just go to the liver. We can now target HCCs. We can target kidney. We can target potentially brain. I'm bullish on new modalities.
--:-- Elliot Hershberg: There is something really interesting in this argument right now within biotech. If there are hyperscalers that emerge, there's this question. There are a lot of AI and biology companies raising a lot of capital, and some have no aims to make drugs. There's this continual background cynicism in the industry. A lot of people are asking, what is that all about? I think we have to believe in net new market creation. At one point, Illumina sold exactly $0.00 of next-generation sequencing technology to the industry. That turned into over $10 billion of sales and independent, large listed companies where their cost of goods sold were primarily going to Illumina. There's this interesting question: the largest company in the world, NVIDIA, is an infrastructure company. Is it possible for there to emerge a really large and fundamental infrastructure company in biotech? There are questions of either going where others can't and making something that people can't make, or making the final arc of commoditization and building consolidated platforms that do all of discovery for small molecules and antibodies. As this technology matures, it's kind of two different possible poles of value creation in the end state of biotech. Biotech is dead and long live biotech.
--:-- Host: What would be your bet?
--:-- Elliot Hershberg: We'd make the orthogonal bets. We're big believers in new modalities, and we're big believers that there's going to be modern infrastructure that drives and underpins the ability to make modern drugs. There's a ton of value creation on both of those axes, and we're very optimistic about that future. Long live biotech.
--:-- Host: Long live biotech. Thanks for listening to this episode of the a16z Podcast. If you liked this episode, be sure to like, comment, subscribe, leave us a rating or review, and share it with your friends and family. For more episodes, go to YouTube, Apple Podcasts, and Spotify. Follow us on X at @a16z and subscribe to our Substack at a16z.substack.com. Thanks again for listening, and I'll see you in the next episode.
