Main Topic: The podcast episode discusses the potential of using artificial intelligence (AI) to accelerate scientific progress, particularly in the field of biology. The guests, Patrick Hsu and Jorge Conde, co-founders of the ARC Institute, and A16Z general partner, Jorge Conde, explore the idea of creating virtual cells and simulating human biology using foundation models.
Key Points:
Participants:
Takeaways:
Additional Insights:
Can we make science as fast as software?
In this episode, Erik Torenberg talks with Patrick Hsu (cofounder of Arc Institute) and a16z general partner Jorge Conde about Arc’s “virtual cells” moonshot, which uses foundation models to simulate biology and guide experiments.
They discuss why research is slow, what an AlphaFold-style moment for cell biology could look like, and how AI might improve drug discovery. The conversation also covers hype versus substance in AI for biology, clinical bottlenecks, capital intensity, and how breakthroughs like GLP-1s show the path from science to major business and health impact.
Resources:
Find Patrick on X: https://x.com/pdhsu
Find Jorge on X: https://x.com/JorgeCondeBio
Stay Updated:
Find a16z on X
Find a16z on LinkedIn
Listen to the a16z Podcast on Spotify
Listen to the a16z Podcast on Apple Podcasts
Follow our host: https://twitter.com/eriktorenberg
Please note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.
Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.
Host
I want to make science faster. Our moonshot is really to make virtual cells at ARC and simulate human biology with foundation models. Why are we so worried about modeling entire bodies over time when we can't do it for an individual cell? If we can figure out how to model the fundamental unit of biology, the cell, then from that we should be able to build. My goal is to really try to figure out ways that we can improve the human experience in our lifetime. There are a few things that if we get them right in our lifetime, will fundamentally change the world.
Today, we're talking about making science move faster. My guests are Patrick Hsu, co-founder of the ARC Institute, and A16Z general partner, Jorge Conde. We get into virtual cells and foundation models for biology, why science gets stuck in incentive knots, what an alpha-fold level movement for cell biology could look like, and how breakthroughs translate into actual drugs and business outcomes. Let's get into it.
Patrick, welcome to the podcast. Thanks for joining.
Patrick Hsu
Thanks for having me on.
Host
I've been trying to have you on for years, but finally I could get your time.
Patrick Hsu
Here I am. I'm excited to do it.
Host
It's going to be great. For some of the audience who aren't familiar with you and your work at ARC and beyond, how do you describe what's your moonshot? What is it you're trying to do?
Patrick Hsu
I want to make science faster. We can frame this in high-level philosophical goals like accelerating scientific progress. Maybe that's not so tangible for people. I think the most important thing is science happens in the real world. If it's not AI research, which moves as quickly as you can iterate on GPUs, you have to actually move things around. Atoms, clear liquids from tube to tube to actually make life-changing medicines. These are things that take place in real time. You have to actually grow cells, tissues, and animals.
The promise of what we're doing today with machine learning and biology is that we could actually accelerate and massively parallelize this. Our moonshot is really to make virtual cells at ARC and simulate human biology with foundation models. We'd like to figure out something that feels useful for experimentalists, people who are skeptical about technology. They just want to see the data and see the results. We want it to be the default tool that they go to use when they want to do something with cell biology.
Host
Okay, well, hold on. Let's back up. Why is science so slow in the first place? Whose fault is that?
Patrick Hsu
Now that is a long one. We should get into it.
Host
We should get into it.
Patrick Hsu
It's really multifactorial. It's this weird Gordian knot that ultimately comes down to incentives. People talk a lot about science funding and how science funding can be better, but it's also about how the training system works. How we incentivize long-term career growth, how we try to separate basic science work from commercially viable work, and generally the space of problems that people are able to work on today.
Things are increasingly multidisciplinary. It's very hard for individual research groups or individual companies to be good at more than two things. You might be able to do computational biology and genomics, or chemical biology and molecular glues. But how do you do five things at once? It's increasingly hard. We really built ARC as an organizational experiment to try to see what happens when you bring together neuroscience and immunology and machine learning and chemical biology and genomics all under one physical roof. If you increase the collision frequency across these five distinct domains, there would hopefully be a huge space of problems that you could work on that you wouldn't be able to.
Obviously, in any university or any kind of geographical region, you have all of these individual fields represented at large across these different campuses. But people are distributed, and you want everyone together.
Host
Okay, if I may. I would have thought a university was an attempt to bring in multiple disciplines under one roof. You're saying it's not. It's too diffuse. It's across an entire campus. So literally the physical distance creates an inefficiency.
Patrick Hsu
That's part of it. The other part is folks have their own incentive structures. They need to publish their own papers. They need to do their own thing and make their own discovery. You're not really incentivized to work together in many ways in the current academic system. A lot of what we've done is to try to have people work on bigger flagship projects that require much more than any individual person or group or idea.
Host
That's cool. So the original hypothesis for the ARC Institute is if you can bring multiple disciplines together to increase the collision frequency, as you said, and if one could remove some of the cross incentives that may exist in traditional structures, the combination of those two things will make science faster.
Patrick Hsu
These are absolutely part of it. We have two flagship projects, one trying to find Alzheimer's disease drug targets, the other to make these virtual cells. It's not just the people and the infrastructure, but also the models will hopefully literally make science faster. You could do experiments at the speed of forward passes of a neural network if these models could become accurate and useful.
Host
So that will be one thing that solves the length of discovery. You compress the time discovery takes naturally, by just throwing technology at the problem, at the risk of oversimplifying.
Patrick Hsu
Well, we're tech optimists here, no?
Host
We are. Yeah. Why has AI progressed so much faster in image generation and language models than biology? And if we could wave a wand, where are we excited to speed certain things up?
Patrick Hsu
To be honest, it's a lot easier. Maybe that's a hot take.
Host
You mean technology is easier than biology. Natural language and video modeling is easier than modeling biology.
Patrick Hsu
To some degree, if you understand and learn machine learning and how to train these models, you have already learned how to speak. You already know how to look at pictures. Your ability to evaluate the generations or predictions of these models are very native. We don't speak the language of biology. At best, with an incredibly thick accent. You're training these DNA foundation models. I don't speak DNA natively. I only have a sense of the types of tokens that I'm feeding into the model and what's actually coming out.
Similarly, these virtual cell models, a lot of the goal is to figure out ways that you can actually interpret the weird fuzzy outputs that the model is giving you. That's what slows down the iteration cycles. You have to do these lab-in-the-loop things where you have to run actual experiments to test with experimental ground truth. Increasing the speed and dimensionality of that is going to be really important.
Host
How much of this is the fact that if you're training on an image, we can see the image, so we can see how good the output is? What about all the things in biology that we can't see or don't even know exist yet? How can we create a virtual cell model when we're not even sure if we understand all of the components that are in a cell and how they function?
Patrick Hsu
People talked about this a lot in NLP as well. There's this long academic tradition in natural language processing. Then it was just weird and non-intuitive and intensely controversial that you could just feed all this unstructured data into a transformer and it would just work. We're not saying this will just work in all the other domains, including in biology. But there is this controversy around what does it mean to be an accurate biological simulator? What does it mean to be a virtual cell?
It's true. We can't measure everything. We can't measure things like metabolites with really high throughput and spatial resolution. There are going to be different phases of capability where initially they model individual cells. Then they model pairs of cells. Then they model cells in a tissue. Then in a broader physiologically intact animal environment. Those are length scales and layers of complexity that will aggregate and improve upon over time.
The other non-intuitive thing in many ways are the scaling laws that you get in data and in modeling. I'll give you an example. There's a lot of discussion in molecular biology about how RNAs don't reflect protein and protein function. We don't have proteomic measurement technologies that are nearly as scalable as transcriptomic measurement technologies today, certainly not at single cell resolution. But we're getting there, and you can layer on certain nodes of protein information that you can add on top of the RNA information.
In many ways, the RNA representation is a mirror. It might be a lower resolution mirror for what's happening at the protein layer, but eventually what is happening in protein signaling will get reflected in a transcriptional state. For an individual cell, this may not be very accurate, but when you imagine the massive data scale that we're generating in genomics and functional genomics, you start to gather tremendous amounts of RNA data that will read in what's happening at the protein level at some sort of mirror echo. That can be the case for metabolic information as well.
Host
So it's a low pixel image, but if we zoom out far enough, we'll get a sense of what's going on.
Patrick Hsu
You have to bet on what you can scale today. We're able to scale single cell and transcriptional information today. We're able to add on protein level information over time. We'll need spatial information, spatial tokens, and we'll need temporal dynamics as well. I bucket things into three tiers: invention, engineering, and scaling. There are certain things today, biotechnologically, that are scale ready. Then there are things that we still need to invent. That's part of why we felt like we needed a research institute to be able to tackle these types of problems. We weren't just going to be an engineering shop that's just trying to scale single cell perturbation screens. That would be interesting, but in three years would feel very dated. There's a lot of novel technology investment that we're making that we think will bear fruit over time.
Host
Can we flesh out the virtual cell concept? Why that's the ambition we've landed on and what it's going to take to get there?
Patrick Hsu
The most famous success of ML in biology is AlphaFold. This solved the protein folding problem of when you take a sequence of any amino acid, what does the protein look like? It's pretty good. It's not perfect. It certainly doesn't simulate the biophysics and the molecular dynamics, but it gives you a sense of what the end state is with 90% plus accuracy. That's the AlphaFold moment people talk about, where anytime you want to work with a protein, if you don't have an experimentally solved structure, you're just going to fold it with this algorithm. We want to get to that point with virtual cells as well.
At ARC, we're operationalizing this to do perturbation prediction. The idea is you have some manifold of cell types and cell states. That can be a heart cell, a blood cell, a lung cell, and so on. You know that you can move cells across this manifold. Sometimes they become inflamed. Sometimes they become apoptotic. Sometimes they become cell cycle arrested. They become stressed. They're metabolically starved.
If you have this representation of universal cell space, can you figure out what are the perturbations that you need to move cells around this manifold? This is fundamentally what we do in making drugs. Whether we have small molecules, which started out as natural products from boiling leaves, or antibodies when we injected proteins into cows and rabbits and sheep and took their blood to get those antibodies. We were basically trying to get to more and more specific probes. We had experimental ways to cook these up. Now we have computational ways to zero-shot these binders.
Ultimately, what you're trying to do with these binders is to inhibit something. By doing so, you click and drag it from a toxic gain of function, disease-causing state to a more quiescent, homeostatic, healthy one. The thing that is very clear in complex diseases, where you don't have a single cause of that disease, is there's some complex set of changes. There's a combination of perturbations that you would want to make to be able to move things around.
People talk about this classically as things like polypharmacology. We're moving from "this thing happens to have a whole bunch of different targets by accident" to "we have the ability to manipulate these things combinatorially in a purposeful way." To go from cell state A to cell state B, there are these three changes I need to make first, then these two changes, and then these six changes over time. We want models to be able to suggest this.
The reason we scoped virtual cell this way is because we felt it was experimentally very practical. You want something that's going to be a co-pilot for a wet lab biologist to decide what am I going to do in the lab. We're not trying to do a theory paper that's really interesting to read where the numbers go up on an ML benchmark. You want to practically decide what are the 12 things you're going to do in the lab in 12 different conditions, and actually just test them. That's how we enter the lab-in-the-loop aspect of model predictions to experimental measurements to improved model predictions again.
The goal is to be able to do in silico target ID, where you can basically figure out new drug targets, then figure out the compositions, the drug compositions you would need to actually make those changes. If we could do that, we could make a new AI, a vertically integrated AI-enabled pharma company. That is obviously a very exciting idea today, but in many ways, the pitch and the framing of these companies precedes the fundamental research capability breakthroughs. That's what we're really invested in at ARC, just making that happen along with many other amazing colleagues in the field.
Host
So if the goal is, I'm oversimplifying, if we wanted to get to the AlphaFold moment where it gives you a useful folded structure 90% of the time, to use your data point. We take that comparison to the virtual cell model and we say, okay, 90% of the time, if I ask the model, I want to shift the cell from cell state A to cell state B, and it gives me a list of perturbations. 90% of the time, those perturbations result in the experimental shifting from cell state A to cell state B. How far away are we from that AlphaFold moment for virtual cells?
Patrick Hsu
I find it helpful to frame these in terms of GPT-1, 2, 3, 4, 5 capabilities. Most people would agree we're somewhere between GPT-1 and 2. A lot of the excitement was that we could achieve GPT-1 in the first place, that you could see a path with scaling laws of some kind to make successive generations where capabilities would improve.
With our Evo DNA foundation models that we developed at ARC with Brian He, one of the things we've seen is that these genome generations are blurry pictures of life. We don't think if you synthesize these novel genomes, they would be alive. But we don't think that's impossibly far away either. We'll just have to follow these capabilities. We're taking a very integrated approach to attack this problem. You need to curate public data. You need to generate massive amounts of internal and private data, build the benchmarks, train new models, build new architectures, and do these things full stack. We'll attack this hill climb over time.
Host
What's the GPT-3 moment going to look like? By that, I mean a public release that alters the public's conception of just what's possible here from a capabilities perspective, and also inspires a whole new generation of talent to rush into biology.
Patrick Hsu
The good thing with biology is we have a lot of ground truth. There are entire textbooks that describe cell signaling and cell biology and how these things work. Even without a virtual cell model at all, if you went into ChatGPT or Claude and asked some question about receptor tyrosine kinase signaling, it would have an opinion on how that works. You would want the model to be able to predict perturbations that are famous canonical examples of biological discovery.
I'll give you an example. If you've loaded into the model an iPSC, an induced pluripotent stem cell state or human embryonic stem cell state and fibroblast cell state, could it predict that the four Yamanaka factors would reprogram the fibroblast into a stem-like state? Essentially rediscover from the model something that won the Nobel Prize in 2009. That would be one really classic example. Then you could do the inverse. If you have a stem cell, can it discover neurogenin 2, ASCL1, myOD? Can it find differentiation factors to turn that into a neuron or into a muscle cell?
These are classic examples in developmental biology, but you could also use this to try to recapitulate the mechanism of action of FDA-approved drugs. You could say, for example, if you inhibit HER2 in breast cancer cell states, you would get this type of response. Or it could predict the certain clones that will be able to be more metastatic or more resistant and lead to minimal residual disease. There are lots of biological evals that you can add onto these models over time that are really tangible textbook examples, as opposed to what the early generation of models do today, which are very quantitative things like mean absolute error over the differential expressed genes. Those are ML benchmarks. We want to increase the sophistication into something that you could explain to an old professor who has never touched a terminal in their life.
Host
By the way, you talk about textbooks as ground truth. Do you think we're going to find that a lot of the textbooks are wrong?
Patrick Hsu
I would say textbooks are compressed. When you look at these classic cell signaling diagrams of A signals to B, which inhibits C, that's a very two-dimensional representation of our understanding of a complex system.
Host
Right, right, right.
Patrick Hsu
Textbooks are what they are. They represent the corpus of reliable knowledge, but everyone knows that there are an incredible number of exceptions. Part of what discovery is, is to find new exceptions.
Host
Why don't you talk about the difference between simulation of biology and the actual understanding? What would it take to actually be able to model the extremely complex human body?
Patrick Hsu
Some people don't like the phrase "virtual cells" because it sounds too media friendly, not rigorous enough. I've always found it funny that many people are okay with "digital twins" and "digital avatars," which talk about modeling biology at a way higher level of abstraction. Virtual cells, if anything, is actually way more scoped and rigorous than modeling a digital twin or avatar. These are useful words because they describe the goal and the ambition.
In the long run, we don't care about predicting the perturbation responses of an individual cell at all, actually. Obviously we want to be able to predict drug toxicity. We want to be able to predict aging. We want to be able to predict why a liver cell becomes cirrhotic when you repeatedly challenge it with ethanol molecules. These chemical or environmental perturbations should be predictable. You just have to layer on the complexity.
Why are we so worried about modeling entire bodies over time when we can't do it for an individual cell? We broadly believe that this is a fundamental unit of biological computation, if you will. Let's just start there. Just like you have to start with things like math and code and language modeling, things that are easier to check. You can build a super intelligence over time.
Host
I think that makes sense. That's a very laudable, ambitious goal. If we can figure out how to model the fundamental unit of biology, the cell, then from that we should be able to build. In early AI, we just started with language translation, basic NLP tasks. This is long before the tremendous ambitious scope that we have today. I think we hopefully can mirror that type of trajectory if we're lucky.
It seems that biotech and pharma has been a shrinking interest in the rate of growth. What's it going to take for these innovations in the science to reflect themselves in business models and in growth for the industry?
Patrick Hsu
A lot of these biotech startups would try to initially sell software to pharma companies. Then they would realize, "Oh wow, we're competing for SaaS budgets, which aren't very large." Now they're realizing, "Oh, we have to compete for R&D budgets." There's this narrative from the current generation of these companies that our biological agents will compete for R&D budgets and replace headcount, just like we're seeing in agents across different verticals. Whether or not that will pan out depends on whether these things meaningfully allow us to build drugs more effectively in the pharma context. That's the most important thing in this industry.
We believe in virtual cells, not just because we think it will be a fountain of fundamental mechanistic insights for discovery, but also because if successful, it could be industrially really useful. We'll have to see over time. If we have 90% of drugs failing clinical trials, that means two things. One is we're targeting the wrong target in the first place. The second is the composition, the drug matter that we're using doesn't do the job. It's not clear for each individual failure which one it is, or if it's both, or what proportion of each. We'll have to sort that out over time.
You can imagine even in the case of success when we had 90% accurate virtual cells, you'll probably end up with suggestions like, "Okay, now you need to target this GPCR only in heart, but not in literally any other tissue." We don't have the drug matter that can do that today. That's why you probably need research to figure out novel chemical biology matter that allows you to drug pleiotropic targets in a tissue or cell type specific way.
Part of why biology is slow is because there's this Russian nesting doll of complexity in terms of understanding, in terms of perturbation, in terms of safety. The crazy thing is the progress in just the short time that I've been doing this is insane. I did my PhD at the Broad Institute in the heyday of developing single cell genomics, human genetics, CRISPR gene editing, and so many other things. The early 2010s papers on single cell sequencing would have like 20 cells or 40 cells. At ARC, in a relatively short amount of time, we're going to generate a billion perturbed single cells. How's that for a Moore's law?
Host
Yeah, that's remarkable. Jorge, I want to hear your answers to a couple of these questions too, as a lead of our bio practice. Both on the GPT-3 moment, what that could look like, and also if you think it's GLB1s or building off that, or if it's going to be something different. Also, what's it going to take for the science to reflect itself in the business for the industry to grow?
Jorge Conde
I'll take the second one first, if I could. In terms of where the industry is right now, one of the big challenges we have is as Patrick describes very nicely, discovery is hard and it takes time. The fail modes are exactly as you described. Oftentimes when drugs fail, which they do 90% of the time in clinical trials, it's because we're going after the wrong thing or we made the wrong thing to go after the right thing. Those are the two fail modes, and that happens all too often.
A lot of the stuff that Patrick is describing is going to basically improve our hit rate or our batting average on figuring out what to go after and then making the right thing to go after said thing. The challenge we have in the industry is that the bottlenecks are still the bottlenecks. The biggest bottleneck we have, which is a necessary one, is we have to prove that whatever we make, that we have the right thing to go after the right thing, so to speak. When we have it, that it's going to be as de-risked as possible before you put it into humans. And we have to be good at making them.
Host
And we got to make them too.
Jorge Conde
Exactly. That bottleneck is a necessarily important one. I'm not suggesting we've got to remove it, but are there ways to reduce the cost and time associated with getting through the bottleneck of human clinical trials? It's interesting because when we talk about all of the various stakeholders when you're making a drug, there are the companies, there's the science that supported the company that's trying to commercialize a product, and there are the regulatory agencies. Everyone is trying to ensure that first and foremost is the ability to discover and commercialize drugs that are safe and effective for humans.
That middle part of actually getting through that bottleneck is hard to speed up in a very obvious way. You can increase the rate you enroll clinical trials. You can use better technology to change the way we design these clinical trials so maybe they can be faster or shorter. But some of them just have a natural timeline. They have to go through. If you want to demonstrate that a cancer drug promotes survival, guess what? It's going to take some time to demonstrate a survival benefit. Or if you want to do a longevity drug, that by definition is a lifetime of a trial in terms of length. There are a lot of these bottlenecks that are really hard to get through.
What helps the industry? There are a couple of things. One is capital intensity will hopefully at some point go down over time as technology gets better. Capital intensity is something that our industry faces. In some ways, it looks a little bit like AI now, in terms of the cost of training these models. But the capital intensity is very, very high. That has not come down. So we got to get success rates up to impact capital intensity to get it down. The second thing is where can we compress time?
Host
So good models can help us compress early discovery time. We still haven't seen, and I think it's coming, but it hasn't happened yet. We haven't seen artificial intelligence or other technologies massively compress the amount of time it takes us to do the clinical development, the clinical trials, the enrollment of patients, all of those things. We're seeing some interesting things coming. We haven't seen the payoff there yet.
Jorge Conde
And the third thing is if we can make better drugs, going after better things, the effect size should be higher. So the answer should be obvious sooner. If we can get those three things right—reduce capital intensity, compress timelines, and effectively increase effect size in some very tough, intractable diseases—that is what I think fixes the industry.
From where we sit at the early stage, the reason that helps us is if the capital intensity goes down and the value creation goes up, it becomes easier to invest in these companies in the early days because you get rewarded for coming in early. The problem we have right now is that most companies aren't seeing rewards happen when there's a value inflection. So you come in early, you bear the brunt of the capital intensity. And even if a company is successful, that success isn't reflected in the valuation. So we're not seeing the step-ups that you see in other parts of the industry. That's really hard from an investing standpoint. So I think we need to see those various factors addressed for this space to really get fixed, to use your word.
Host
Yeah, that was great. I have a lot to add on to this. Please add away. A few simple observations. The first is the amount of market cap added to Lilly and Novo based on the development of GLP-1s is over a trillion dollars. Novo stock has decreased a lot, so let's say a trillion dollars. That is more than the market cap of all biotech companies combined over the last 40 years that have been started.
I think one of the interesting correlations of this is that when we have a 10% clinical trial success rate for a preclinical drug candidate, you tend to circle the wagons and try to manage your risk. The way you do this is you go after really well-established disease mechanisms where, if I develop new drugs that go after well-understood biology, it should work the way I hope it will in the trial. That is really expensive and costs a lot more than the preclinical research. The problem is you go after very well-validated disease mechanisms but with really small patient populations. So the expected value of this is relatively low.
One of the things we've seen with GLP-1s is the value you can create when you go after really large patient populations. I think that has culturally increased the ambition of the industry, both from the investor and from the drug developer side. That's something we should keep our foot on the gas for.
Jorge Conde
Yeah, and look, I think the trend on that is positive. I would argue the trend on that is positive. You're absolutely right. The demonstration of the value that has been created with the increase in use of GLP-1s and the value transfer that's gone to companies like Lilly and Novo is very merited, because they've cracked an endemic social problem in terms of managing diabetes and eventually helping manage obesity. That's remarkable. There's a lot of value that goes to that because they tackled a very challenging problem for society beyond just science.
I agree with you, the prize, the juice needs to be worth the squeeze. A lot of biotech has been about going after the low-hanging fruit because it's low risk and we got to eat today. You push off the big ambitious indication, the large population, or the really tough-to-crack disease. But I do think we're seeing more and more of that. By the way, some of these genetic medicines are going after some of the hardest problems—things you quite literally couldn't address but for editing DNA. I think that's incredibly remarkable and laudable and frankly inspiring.
But the fundamental elements of the industry have to work. The capital formation is there to support those kinds of things. And right now it's hard, because of the issues we talked about before.
Host
15 years from now, we're back in this room. We've barely escaped being part of the permanent underclass. We're reflecting on the GPT-3 moment or maybe the legacy of GLP-1s beyond where they are now. What do you think it could be? I'm curious to get your take on what is going to be the technological breakthrough that we're going to point back to and say, "This is really what set it all." Or do you think it's going to be a multi-factor combination?
Jorge Conde
Yeah, look, I think it's going to go back to where we started this conversation. GLP-1s as a drug are four decades in the making. These are not overnight successes. But I do think what we are going to see more of, and our hope is, when you combine the fact that we're getting better at understanding what to target and better at designing medicines to hit those targets in a whole array of new creative ways—small molecules, natural products from boiling leaves, as you said earlier—we're getting really good at designing smarter and better small molecules that do new things, that function in ways they didn't before.
We've gotten quite good at designing biologics or proteins with a lot of help from things like AlphaFold that helps understand how proteins fold. We're going to get a lot better at designing some of the more complex modalities like the gene therapies or the gene editors. When you can do that and combine it with our ability to use things like virtual cell models to really understand what to go after, we're going to have drugs. I would hope and expect that the industry will continue to bring forward drugs that have very large effect sizes for very difficult diseases that hopefully affect a lot of patients.
If that's true, then we'll start to see some of these really difficult diseases that affect all of society get tackled, hopefully one by one by one. We have obesity, we have metabolic disorder, we're dealing with cardiometabolic disease. We're starting to see interesting, promising things happening in neurodegenerative diseases. If we can tackle cancer, or at least several cancers that have begun to be treated more like a chronic condition than a death sentence, the more we see of that, the value to society will accrete over time. This should be an industry that is extraordinarily valued by society and candidly by the markets. We have to deliver.
Host
If we play this out, and let's say these AI models work, and you can make a trillion binders in silico that will be exquisite drug candidates, we still need to make these things physically and test them in animals and hopefully predictive models, and then actually in people. That will increasingly be the bottleneck in many ways.
My friend Dan Wang recently released a book called *Breakneck*, which talks about the US and China and the difference between the two countries and their philosophy, the way they approach markets. We're a country of lawyers or a country of engineers.
Jorge Conde
Exactly, that's right.
Host
China is an engineering state. The Politburo is folks who have engineering degrees. You need to build bridges and roads and buildings, and these are the ways that we solve our problems. Whereas from the first 13 American presidents, 10 of them practiced law. From 1980 to 2020, all Democratic presidential candidates, both VP and president, went to law school. You see the echoes of that in the FDA and the regulatory regime and all the bottlenecks that people talk about developing drugs stateside.
Increasingly, you see folks thinking about how we can run Phase Ones overseas, build data packages that we can bring back domestically for Phase Two efficacy trials. I think that's interesting directionally, but it's not enough. We need to figure out these two bottoms: the making and the testing. Even if we can solve the designing part.
Jorge Conde
Oh, I agree. Yeah, that's the bottleneck. We joke about it. You have to get a molecule that can go first in mice, then in mutts, then in monkeys, then in man. That takes a long time, and it's so hard to compress that. So when you do, you should make the journey worth it. When you fail on the other end of that, that's obviously horrible. Finding ways to make sure that when you walk that path, it'll be a successful journey as often as possible is what this industry desperately needs.
Host
AlphaFold solved a protein folding problem, but why didn't it solve drug discovery? More broadly, what would it take to get a drug? What is the bottleneck on the tech side, at least?
Jorge Conde
On the tech side?
Host
Yeah. Or maybe another way to ask the question is, because I always ask the founders a version of this question—the AI ones that are like, "Oh, we're going to do AI for life for drug discovery." My question I like to ask founders is, give me examples where you think AI is hyped, potentially overly hyped, where there's real hope, what we expect next, and where we already see real heft. So if I asked you, in AI, where is there hype, where is there hope, and where are we seeing heft today?
Jorge Conde
I would say there's hype in toxicity prediction models. That's the idea that we will say, "I'm going to show you a molecule, and the model is going to tell me if it's going to be toxic or not."
Host
That's right.
Jorge Conde
There's heft in anything to do with proteins. Obviously protein binding, but increasingly in protein design. I think there's real heft there. And then where there's hype is in multimodal biological models, whatever that means. Pick your favorite layers. It could be molecular layers. It could be spatial layers. Actually, I would say there's also heft in the pathology AI prediction models—automating the work of pathologists and radiologists. That's a powerful use case.
Host
Yeah, sure.
Jorge Conde
And there's a lot of stuff where you don't have to train weird biology foundation models, and you can write regulatory filings and reports and things like that. That's impactful and important. So now I go back to your question: why hasn't AI turned out drugs yet?
AI for drugs is one of these weird things where everyone who works in the industry is trying to claim that their drug is the first AI-designed molecule. Increasingly, in just a few years, this will just be a native part of the stack. Just like we use the internet and we use phones, we're going to have AI in all parts of the stack. So why hasn't it worked yet? It's this long, multifactorial process we've been talking about today. There's designing, there's making, there's testing, there's the approvals side.
Safety and efficacy as the two pillars in the industry are the two things we need to get right. We need to be able to figure out faster ways to predict whether a molecule will work and if it's going to be safe. There are ways AI can operationalize this. If you designed a small molecule, you could now computationally dock it to every protein in the proteome and see if it's likely to bind to off-target molecules. You can use this to tune binding selectivity and affinity in ways that might predict safety and efficacy. How well will that work? That's a feedback loop we'll have to actually test in the lab. That's part of what's slow—the testing takes real hours, days, months, years.
That's really why we picked at ARC the virtual cell models as our initial wedge. We think it can integrate a lot of these different pieces.
Host
In Dario Amodei's essay, *Machines of Love and Grace*, he predicts, among other things, the prevention of many infectious diseases and the doubling of lifespans, perhaps as soon as the next decade. What's your reaction to his essay, his bullishness, in some of his predictions?
Jorge Conde
I think the core intuition that Dario had was the idea that important scientific discoveries are independent, or they're largely independent. If they are statistically independent, then it would stand to reason that we could multi-parallelize. If we had models that were sufficiently predictive and useful, you could have not just a hundred of them, but millions, billions of these discovery agents or processes running at a time, which should compress the timeline to new discoveries and turn it into a computation problem.
I think that is a very futuristic framing for something that is actually very tangible today. If we can have virtual cell models at work that can start to do these kinds of things—help us with molecular design models, docking models, and then when you bind to this thing in this cell versus all the other off-target proteins, will a cell be corrected in the right way? These layers of abstraction and complexity start to get to things that feel very tangible to drug discovery. If you could actually traverse these steps reliably and in sequence, you can start to see how you can get the compression. So I think in the long run of time, this should be possible.
Host
One of the core suppositions in building a good virtual cell model is that we are feeding it all the relevant data.
Jorge Conde
The right data.
Host
The right data. So we'll work to get gene expression data, DNA data, any number of factors, protein and protein interactions, all the things you described. What if we're missing a core element? What if we just haven't discovered the quark or whatever? We just don't know what we don't know. And therefore what we're feeding the model is fundamentally or importantly incomplete.
Jorge Conde
I think that's almost certainly true. It seems almost obvious that we're not measuring many of the most important things in biology. You can find many important exceptions for any of these measurement technologies. In biology, we ultimately have two ways to study it in high throughput: imaging and sequencing. But there are so many other types of things you would care about that those things aren't necessarily going to do at scale.
That's really why the stuff we're talking about—the RNA layer as a mirror for other layers of biology—is one we spent a lot of time thinking about. There's a difference between a mechanistic model and a meteorological simulation type of model. For example, if you want to predict the weather, you can build AI models that will predict whether it will rain next Tuesday. It won't explain physically or geologically why and how that happens. But as long as it knows if it's going to rain next Tuesday, you're probably happy. Similarly, with a virtual cell model, it may not tell me literally why. Just like AlphaFold doesn't tell me literally why the protein folded this way and how, but it just told me the end state and it was reasonably accurate. I think that would already be very important.
Host
Shifting gears a little bit, we've been talking about science and biotech, but you're, in addition, an elite AI investor more broadly. So I want to talk about where your investment focus is right now as it relates to AI more broadly. Where are you excited? Where are you spending time? Where are you looking forward to?
Jorge Conde
My goal is to really try to figure out ways that we can improve the human experience in our lifetime. I think about the future we're going to leave to our children. There are a few things that if we get them right in our lifetime, will fundamentally change the world and how we live in it. I think synthetic biology is obviously one—GLP-1s, things that improve sleep, things that can improve longevity. These are all things that are easy to get excited about.
I think brain-computer interfaces is another area where we're going to see really important breakthroughs over the decades to come. And then I think the third is in robotics, both industrial and consumer robotics, that allow us to scale physical labor in interesting ways. You can see how each of these three things, even in the medium cases of success, really change the world. I'm very interested in helping make these kinds of things possible.
In the techno-optimist vision of the world, there are a few different types of scarcity. It's very easy when you do research to come up with important ideas. The hard thing is to tackle them in the right timeframe. Writing futuristic sci-fi things is not that hard. Being able to actually execute on it in the next five years or eight years is much harder. Academic discovery is littered with plenty of ideas that are interesting and important, but long before their time.
In many ways, the story of technology development is trying to use new technologies to solve old tricks. Most of our tools are for productivity. Whether that's the industrial revolution or the computing revolution or the current AI revolution, we're trying to do the same stuff. There's a relatively small set of very powerful ideas. New technologies give us new opportunities to attack them. There's a set of people and teams that are going to be positioned to be able to do that. They need to have technical innovation and then an intuition about product and business.
In the RPG dice roll of the skills you get in these three domains, people start at different base levels. You might have an incredibly technical founder who doesn't know how to think commercially, or someone who's natively a very commercial thinker but doesn't have very strong product sense, even though they could sell the crap out of it. These three broad categories of capabilities need to be brought together in a way that you can allocate capital to in the right times in order to make these ideas possible in a really differentiated way. This thing literally wouldn't happen if we didn't get these people together and fund it at the right time in the right way. That's really what motivates me.
These are the kinds of things I've been excited about backing: longevity companies like New Limit, BCI companies like Nudge, robotics companies like The Bot Company. These are some of the examples of things that I think must happen in the world and therefore should happen. How do we actually find the right people and the right time to go on the fellowship of the ring hunt?
Host
If not too difficult, I want to ask you a question adapted to these additional spaces—robotics, BCI, and longevity, if appropriate—in terms of three questions: what's overhyped, where do you see opportunity or path, and what's got heft already?
Jorge Conde
I think the cool thing about agents generally is that they do real work. Compared to SaaS companies that came before, agents replace real productivity. I think they have a lot of errors today. I would say the computer use agents will probably trail the coding agents by maybe a year. But it's coming, and we'll follow the trajectory as these go from doing minutes of work without error to hours to days. You're going to get a completely different product shape as we march through that across legal, BPO, medicine, healthcare, whatever. That's going to be really exciting. That's where we're going to see real heft, because most of the economy is services spent, not software spent. The reason we're all excited about this stuff is that it can attack the services economy.
Where is there hype? There's a tremendous amount, no doubt. The hype is in the model capabilities. We're working with an architecture that dates back to 2017. If you look at the history of deep learning, every eight years there's something really different. It feels like in 2025, we're really overdue for some net new architecture. There are lots of really interesting research ideas that are bubbling up that could do that thing.
There's a set of really interesting academic ideas, especially in the golden age of machine learning research from 2009 to 2015. There are so many interesting ideas, little arXiv papers that have 30 citations or less. As the marginal cost of compute goes down year on year, you're going to be able to take all of these ideas and actually scale them up. You don't see the scaling laws when you're training them at 100 million or 650 million parameters like back then. But if you can scale them up to 1B, 7B, 35B, 70B, you start to see whether or not these ideas will pop. That's very exciting because there's going to be a lot of opportunity for new superintelligence labs to do things beyond what the established foundation model companies are doing today.
These companies, in addition to these research teams, are in many ways becoming applied AI companies. They need to build product shape and do RL for businesses and make money, or build coding agents and make API revenue. That's important. It's a timely race to survive today. But I'm very bullish on the research of, say, a Sakana AI, which was founded by one of the authors of "Attention is All You Need," Llion Jones. They're doing incredibly interesting stuff on model merging and how you can have evolutionary selection of different models in an MoE. The opportunities in the long run to move beyond just RL gyms and figure out new ways to learn and find reward signals is going to be really exciting.
Host
It's a great place to wrap. Gearing towards closing, anything upcoming for Arc that you'd like us to know about? Anything you want to tease? For people who want to learn more, what should they know about?
Jorge Conde
AlphaFold came out of a protein folding competition called CASP. We created our own Virtual Cell Challenge at virtualcellchallenge.org, where we have $100,000 prizes sponsored by NVIDIA, 10X Genomics, Ultima, and others. It's an open competition that anyone can enter, where you can train perturbation prediction models and we can openly and transparently assess these model capabilities both today and in subsequent years, follow them to get to that ChatGPT moment. I'm extremely excited about this. We'd like more people to train models and apply, both bio ML experts and engineers in any other domain. I want this thing to exist in the world. Hopefully we're important parts of making that happen, but I'd just be happy that someone does it.
Host
That's an inspiring note to wrap on. Patrick, Jorge, thanks so much for the conversation.
Patrick Hsu
Thanks so much, guys. Appreciate it.
Jorge Conde
Thanks for having me.
Host
Thanks for listening to the a16z podcast. If you enjoyed the episode, let us know by leaving a review at ratethispodcast.com slash a16z. We've got more great conversations coming your way. See you next time.