Biohub: The Future of Biology is Open-Source with Mark Zuckerberg, Priscilla Chan, and Alex Rives

We just want to give tools to the whole scientific community.

We want to understand how biology works. I want to understand the genetics of this person. I want to understand the risks they have to different illnesses. My goal is to be able to treat the individual as an individual, understand the mechanisms and be able to intervene.

We'll have a bigger impact by getting this in more scientist hands quicker by doing it as open source projects instead. It's not just like there's some factory somewhere that you can pay to produce the data. You actually need to invent new novel scientific approaches. The theory isn't that we're going to cure the diseases. We're not. It's that we want to help accelerate the pace of progress for the whole scientific field. We folded over 1.1 billion proteins and predicted their structures. And we didn't design a model for antibodies. We didn't design a model to be able to bind one particular target. We just designed a model that could understand proteins. If we could design a protein to actually change the physiology, then we can actually cure someone.

No. No. And to be clear, we don't think that we're going to be the ones curing the diseases. Our goal was always to build tools that could accelerate the whole scientific field. That way, the scientific field collectively could cure all the diseases. But still, I thought that by the end of the century was a stretch. Now, I think it's too conservative. And so we kept being like, 'Okay, well, we had these series of funny, awkward, educational conversations where we're like, okay, but like why? Like why do you think it's impossible?' And being the person in the room is just like, 'Well, I don't know why. You tell me.' Finally, we got people to like they're like, 'Fine, if you really must know.' And we're like, 'No, we do. It seems important.' They were like, well we work in silos and when you publish information doesn't get shared, it gets locked up for long periods of time and we don't have tooling. They gave the example of we build a great tool by one postdoc in a lab and it lives on their computer and when they graduate the tool is gone. It was very hard to build shared tools to move science faster, build a shared knowledge base to quickly move science faster. That's where we began in thinking about okay, if those are the problems, what can we contribute?

Yeah, I mean so the original biohub model was basically focus on long-term tool development by bringing together engineers and scientists across multiple universities to focus on long-term tool development and it basically worked. We started off with CZI doing a number of different things and I think over time we just felt like okay, the science piece is really working and we just kept on investing more and more in it until now it is basically the primary and main thing that we're doing. We've expanded the original San Francisco Biohub to a handful now. There's New York, there's Chicago. The real focus and the unifying theme at this point is the virtual biology initiative around taking the unique data sets that are able to be generated in order to effectively model starting with the smallest pieces of proteins, but then eventually cells and whole biological systems. That's how we've evolved. This idea that some of this is an AI problem and you want to build a frontier AI lab, but you need to couple that with a frontier biology effort that can do the work of being able to understand and get the data that you need to actually be able to build these models. Because unlike language models, there's just a lot of data out there on the internet. That's not really the case with biology. There are obviously a bunch of different data sets that exist that academia and scientists have generated over the decades, but a lot of the stuff that I think we want to put into this, it doesn't exist. You want to be able to visualize things that people haven't been able to see before, which is why we're doing the imaging work. You want to be able to record things that are going on inside the body, which is why we're doing the kind of cellular engineering work. You want to be able to measure things like inflammation in ways that haven't been possible, which is why the Chicago Biohub is focused on building those kind of devices. That will fundamentally create new types of data sets that will allow new types of models. Going back to what you said, if the scientific field primarily needs tool development that now is going to empower scientists across the field to be able to do their work faster, that's what we think we can provide through this kind of long-term focus on tool development. But I think there's a fun through line from where we started to our work with Alex. Our very first request for application here was around single cell sequencing and we wanted to look at the RNA that is transcribed in individual cells. That was possible but it was still pretty early on in understanding how different cells were expressing their DNA. At the beginning we were just funding methods, getting people to describe how to do it so that others could share that methodology. Then that became us funding the human cell atlas, which is now one of the largest databases of single cell transcriptomes. It was getting hard for scientists to annotate the data, so we built cellbygene, which was a very simple annotation tool that scientists could use to make use of that data. A community came around cellbygene and started contributing more data that we had nothing to do with. Now cellbygene is a corpus of knowledge that a lot of transcriptomic based models are based off of and is used regularly by the scientific community. But there were always critiques that this is just stamp collecting, just gathering bits of data, and we're not going to be able to pull scientific knowledge and wisdom and insights out of it. We didn't have an answer for a while, and then imagine our delight when large language models became a huge topic of conversation that could make sense of large amounts of data. For me, what if we could actually understand how biology worked? Move it from a discovery based science to an engineering based science where we could systematically understand how living cells worked and understand why things go wrong. When we saw that moment, we're like, this is it. Something really big could happen here.

There are probably some differences. You can probably talk more to the specifics around this, but I think each layer is going to end up being somewhat qualitatively different. But you need to be able to understand the protein interactions in order to be able to understand how cells work. So you can't just go straight to cells without understanding the protein modeling. If you're trying to understand something like the way the immune system works or a bunch of cells interact together, it's tough to do that without first understanding cells. You might be able to simulate a system at a very high level of abstraction, but if you really want to understand how it's going to work, you want to build the simulations at each level hierarchically. That's basically the approach that we're going through, starting with the building blocks and the protein. There's going to be different types of data that you want to collect for each. The modeling techniques, I think we'll see. That'll all keep advancing across the board. But a big part of the strategy is this view that you need to build it up hierarchically. One of the things that's unique about us in the space is we were very intentional that the AI efforts and the wet lab efforts were a single effort. We've done a lot of work to bring them together. The really neat thing that we can do is try to pull and gather data that helps us connect across the hierarchy. You can look at transcriptomics with space within a cell and look at where it's localizing. We can look at translucent zebra fish and look at the development across different cells. We have sensors that allow us to look at cell cell communication and different molecules. So we can be strategic about the types of experiments and data we want to collect that helps us bridge across these and creates connective tissue that drives the modeling magic that happens.

Are you looking for a job?

Well, I think we just want to give tools to the whole scientific community. In order to have the biggest impact, it's not actually clear that we couldn't run it as a business if we wanted to. I just think that we'll have a bigger impact by getting this in more scientist hands quicker by doing it as open source projects instead. That's the approach. You have to raise a large amount of money in order to build compute clusters. In a lot of ways the data is even more of a constraint. The scale of these models compared to language models is smaller because the amount of data is less. To get the data, it's not just like there's some factory somewhere that you can pay to produce the data. You need to invent new novel scientific approaches to do the cellular engineering we're doing in New York or the types of devices in Chicago. When we talk about frontier biology and frontier AI, the frontier biology is you need to do real science to advance different biological methods to observe the things that create the data that go to the model. It's not an off-the-shelf thing. That's a pretty big effort. I don't know that there are many things like that done as biotechs. It's the scale of the ambition and the horizon over which we're committed. If you're building tools this complicated, you want a 10 to 15 year time horizon. There's no rule that said you couldn't do it as an incredibly well-funded startup, but this made more sense. It simplifies strategically to not have to think about making money. We release them as open source. The theory isn't that we're going to cure the diseases. We want to help accelerate the pace of progress for the whole scientific field.

And that's the power of the approach is you have these big models that you build that can then apply anywhere. I know that you mentioned earlier that you were going to try and cure prevent all diseases within a hundred years and you said it could actually be sooner now given all the advances in AI. Do you have some thought of when we think we'll be closer to that goal?

I'm optimistic it'll be sooner. The thing that's complicated is that it's a dynamic system. If you fix something, there will be future things that you need to work on. I don't think the current set of things we're aware of are going to be the only things that need to be worked out. The progress with AI is very exciting on this. The other thing I'd say is we look at more kind of systems than specific diseases. One area that seems really important to understand is inflammation. This is a big focus of the Chicago Biohub. There's a lot of data on that. It seems quite clear that it's connected to a bunch of different diseases, but rather than studying the specific diseases, we think that by understanding inflammation more broadly, other companies can use these tools to work on specific therapies. Another example is the immune system. I think it's a very good case to study for some of the work we're doing in cellular engineering when we lad up from proteins to cells to whole dynamic systems within the body. That one makes sense because cells can travel around through the body. Obviously that has a big part in addressing different diseases. How do you make the immune system function better? Connecting that last mile is going to be more something that biotech or other academics will be better suited to do. This is how we think about building out the tool set that helps accelerate all these other folks. Whether the timeline is 10 years or hopefully less than 100 now, I think it's useful for your average doctor or patient to think about what's externally visible in the progress here. You worked with patients for a long time at UCSF. What should doctors look out for? What should people look out for if you're actually accelerating progress?

Please, explicitly.

Yeah, so about a week ago we announced the new ESM fold, which is an open system for scientific discovery in protein biology. It's a world model of protein biology trained on billions of protein sequences, using a language model to learn emergent representations. We can predict atomic-resolution protein structure very quickly, and we've folded over 1.1 billion proteins, identifying features through mechanistic interpretability. The most exciting part is that it's a general model of protein biology, allowing us to search its space to design new proteins. It achieves state-of-the-art on structure prediction benchmarks, especially for protein-protein and protein-antibody interactions critical for therapeutic design. We've used it to design single-chain antibodies digitally, and after selecting from hundreds of thousands of trajectories, we synthesized and tested 96 proteins in a short experimental cycle, finding nanomolar binders. This shows that general-purpose models can yield protein design as an emergent property. It also highlights the power of open science, as we're releasing it as an open discovery engine anyone can build on, replacing intensive lab screening with computation.

That's right. It's critical to characterize these molecules in the lab. We have a structural biology center with powerful cryo-EM microscopes, allowing us to look at proteins biophysically and functionally. We designed proteins for several therapeutically relevant targets and confirmed their function and structure at atomic resolution at the binding interfaces.

Yeah, it's amazing.

We're able to look at the structure also at atomic resolution at the binding interfaces.

My hope in building a comprehensive model of how cells work is to also predict off-target effects. Some off-target effects occur because we didn't know a kidney cell expresses a receptor, leading to renal toxicity. With a single-cell atlas of all cell types, some not predicted before, we can see which cells have receptors for the target and predict downstream effects before human trials. That's an exciting application of a transcriptomic model to understand how different cells react to interventions. When it comes to delivery mechanisms and patient care, we need to choose the right disease first—some are easier to deliver therapeutics to. We were inspired by baby KJ, where the team at CHOP delivered a CRISPR therapeutic to edit a mutation that would have caused neurotoxicity. That disease was chosen because we could easily target his liver cells. The creativity and wherewithal to choose the right applications will unlock the first applications.

Yeah, recruitment could change. We have a program called Rare Is One. The idea is that many companies focus on common diseases, but there's a long tail where economics don't work. If patients can organize and say they'd take an experimental drug, and because the cost is huge, if you can flip that, the economics make more sense. You can generate something more easily and pair it with a group of people. In science and engineering, you often hit your head against the wall on common problems, but you learn a lot from rare or weird edge cases. That connects well here because now you'll be able to enable a long tail of new ideas and test them more easily.

But to some degree, you'll need something like this because many more new things can get created. That doesn't mean the general population won't want the same level of vetting. But giving people who want to be on the frontier the ability to opt in will also be helpful.

I think it's a similar philosophy. As Priscilla was saying, our focus is building tools that empower individuals. That's a common theme across a lot of things I work on—putting technology in individuals' hands. We don't believe in a centralized future with a small number of institutions advancing everything. Our vision isn't a central superintelligence that solves all science. People are really important, and giving them more tools to be more productive will be critical for a positive future. Progress has always been made through empowering individuals to try non-mainstream ideas. That's central to our ethos—it's why you create social media, to give people a voice. Open-source AI is one instantiation of that. In science, it makes sense, and we're deeply committed to open source, though there are considerations around biosafety we need to balance. Overall, this is deep in the ethos of our work at Biohub and other projects.

It's a hot market for AI researchers, but they're in high demand and can work on what they want. This gets back to frontier AI and frontier biology. The AI researchers here could work on language models elsewhere, but those labs don't have the frontier biology part. There's a mission component—they can do unique work here that they can't do anywhere else. I don't think there's any other organization in the world doing both frontier biology and frontier AI.

It's a really powerful mission, and you... yeah.

The other thing is you don't need a very large team. You can make progress with a strong group of a dozen or a couple dozen people. Finding people who care about this mission isn't hard because it's a super important thing in the world. People are drawn to different missions.

That's a great question. I'm really optimistic. Historically, people could spend an entire career on how to optimize a drug or get it through preclinical safety. With a new scientific paradigm, questions that were once hard become simplified. I'm optimistic that many core problems will be solved emergently through these models. For example, toxicity: if you can digitally simulate everything and predict where a drug distributes and binds across the body, you have the beginning of a solution. Once we have accurate molecular representations, we'll see rapid progress on these core problems.

It's been great to see it integrated into all kinds of things. One interesting thing is people connecting it with agentic systems to do automated design and automate the whole process. It's another example of bringing together agentic frontier AI with a world model for biology to reason about and automate the entire design process.

There are two things. We have a view on the next big challenge, which is the virtual cell—laddering up the hierarchy of biological complexity to the cell.

There are different views, but ultimately you want a system that can model each level of complexity: the proteomic, genetic, and transcriptomic layers, and connect that to the phenotype. It needs enough generality to answer questions about a new intervention in an untrained context. The gap we need to close as a field is making predictions that generalize, which will require an enormous data generation effort.

This has been the most dynamic period of technology I've seen. It's so exciting with everything happening in AI—every week something changes.

I'm both.

I feel like everybody's in a manic phase. Yes, it's a combination of invigorated and exhausted.

We have a clear view of this hierarchical set of world models around biology. We want to do the highest quality work in the world. With a world-class AI research team and biohubs that are world-class life sciences organizations, that's a setup no other organization has. But great ingredients don't guarantee success. Five years from now, I hope we produce something meaningfully better and a unique intellectual contribution. I also expect to see a lot more idea generation from people using the models, but I have faith that will materialize. For me, it's about making sure we do world-class work, and the rest will take care of itself.

From the last year, the biggest thing is that we formalized Biohub as the main focus of our philanthropy—a very big shift. Alex and the team coming in has been interesting. They're a world-class group who know each other and work well together. That stability provides a compounding benefit. Previously, Biohub leaders were primarily biologists interested in technology. Now we've flipped that: Alex is primarily an AI researcher with a background in biology. That reflects our expectation that this will drive more value in the future. So those are the biggest updates: new leader and team. On the industry side, it's on track. Exponential growth is accelerating, which has profound implications and validates making a big investment.

Thank you for joining us.