Alex Rives

1. Key Themes

Biology as an Information Architecture Problem, Not a Discovery Problem

The central intellectual bet of Biohub is that biology can be moved from a discovery-based science to an engineering-based science. The arrival of large language models was the inflection point that made this feel tractable.

"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 beings, living cells worked and be able to understand why things go wrong. And so when we saw that moment, we were like, this is it. Something really big could happen here." — Priscilla Chan 00:07:32

Hierarchical World Models as the Architectural Strategy

Rather than building disease-specific models, Biohub is deliberately constructing a hierarchy: proteins → cells → whole biological systems. Each layer depends on the layer below it, and the goal is to connect them with bridging data.

"You can't just go straight to cells in a way without understanding the protein modeling. And then if you're trying to understand something like the way the immune system works or a bunch of cells interact together, then it's tough to do that without first understanding cells... I do think that a big part of the strategy is this view that you need to build it up hierarchically." — Mark Zuckerberg 00:10:35

Frontier Biology Must Co-Evolve with Frontier AI — Data Cannot Be Purchased

A non-obvious structural insight: the models are smaller than language models not because of capability limits, but because the data doesn't exist yet. Generating it requires novel scientific invention, not just funding.

"In order to get the data, 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 to be able to do the, for example, the type of cellular engineering we're doing in New York or the types of devices in Chicago." — Mark Zuckerberg 00:18:21

The Open-Source Strategy Is a Deliberate Force Multiplier, Not a Default

Open-sourcing is a conscious strategic choice to accelerate the whole scientific field rather than extract commercial value, and specifically to address the rare and long-tail disease problem that economics would otherwise orphan.

"We'll have a bigger impact by getting this in more scientists' hands quicker by doing it as open-source projects instead... 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." — Mark Zuckerberg 00:17:24

Mechanistic Interpretability as a New Scientific Discovery Engine

Applying the mechanistic interpretability toolkit developed for large language models to protein language models allows researchers to extract actual biological knowledge — connecting unknown proteins to known ones through emergent representation structure.

"The models have been trained on billions of protein sequences. They've been trained on both known and unknown biology. And yet they're developing these representations that start to kind of capture things that we can really see correspond to that reductive picture of biology that's been built up over the centuries. So you can start to connect the dots between proteins where we don't really know anything about them with proteins where we do know something." — Alex Rives 00:15:53

ESM Fold 2 Demonstrates General-Purpose Protein Models Yield Therapeutic Design as an Emergent Property

The new model folded 1.1 billion proteins, hit state-of-the-art on structure prediction benchmarks, and produced nanomolar antibody binders — not because it was specifically designed for any of those tasks, but because it was trained to understand proteins generally.

"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. And you kind of get protein design as an emergent property." — Alex Rives 00:30:25

Personalized Medicine as the End State: Treating the Individual as an Individual

Rather than disease-area bets, the organizing vision is complete mechanistic understanding of each individual's genetics, protein variants, and disease pathways — enabling bespoke protein or drug design per person.

"I want to understand the genetics of this person. I want to understand the risks they have to different illnesses. I want to understand the mechanistic connection between, say, a gene variant, a protein, and a disease process. Because if you understand that through chain, then you can design a protein, design a drug, bespoke to them, and actually make an intervention." — Mark Zuckerberg 00:22:16

Inflammation and the Immune System as the Most Tractable Systems-Level Targets

Instead of studying diseases directly, Biohub studies the underlying systems — inflammation and the immune system — that connect to many diseases, and then empowers others to do disease-specific work from that foundation.

"Rather than studying the specific diseases, we think that by trying to understand inflammation more broadly, that will make it so that other companies that can then use these tools can work on specific therapies... The immune system, I think, is a very good case to study for some of the work that we're doing in cellular engineering." — Mark Zuckerberg 00:25:28

Leadership Flip: AI-First Scientist Leading Biology, Not Biology-First Scientist Leading AI

A deliberate and structurally significant strategic shift — Biohub previously was led by biologists interested in technology; it is now led by an AI researcher with a biology background. This reflects a conviction about where value creation will compound.

"Prior to Alex leading the effort, the previous leaders of the Biohub were basically primarily biologists who were interested in technology. And now I think this is the point where we really flipped that... you are primarily an AI researcher who has a background in biology. I think that that's like a deep reflection on the way that we expect that this is going to drive more value in the future." — Mark Zuckerberg 00:52:58

2. Contrarian Perspectives

Curing All Disease by End of Century Is Now Too Conservative, Not Too Ambitious

When CZI started 10 years ago, Nobel Prize–winning scientists laughed at the goal. The conventional wisdom remains that it's wildly optimistic. The Biohub team now believes the original timeline was too slow.

"People thought that by the end of the century was a stretch. Now I think it's like too conservative." — Mark Zuckerberg 00:02:17

A Small, Stable Team of a Dozen AI Researchers Can Outperform Larger Labs

Against the prevailing wisdom that frontier AI requires massive research organizations, the Biohub view is that a small, stable, mission-aligned team compounds faster than large headcount — and that the unique data and biology access is the real moat, not team size.

"You don't need a very large team... I think you can really make progress with a very strong group of a dozen or a couple dozen people." — Mark Zuckerberg 00:43:57

No Central Superintelligence Will Solve Science — Distributed Tools Beat Centralization

Against the popular narrative of a single powerful AI solving all scientific problems, Biohub's philosophy is explicitly decentralist: progress comes from empowering a long tail of motivated individual scientists, not from concentrating capability.

"Our vision is not that there's going to be like some central superintelligence that solves all of science. I think people are really important and I think will be more important in the future. And giving people more tools to be more productive is going to be like a critical part of any kind of positive future." — Mark Zuckerberg 00:39:35

Rare Diseases Are Better Scientific Teachers Than Common Ones

The conventional resource logic concentrates funding on large-population diseases. The Biohub view is that rare and edge-case diseases reveal more about how biological systems actually work, yielding disproportionate scientific insight relative to investment.

"Often you can hit your head against the wall on the common problems and in this case diseases, but a lot of times you learn a lot more about a system from finding some kind of rare or weird side thing that's happening." — Mark Zuckerberg 00:36:50

The Exponential AI Curve Feels Impossible to Sustain — But That Emotional Reaction Is the Wrong Signal

Most observers interpret the rapidly accelerating AI progress as unsustainable. The Biohub view is that the feeling of impossibility is intrinsic to exponential curves and should be read as confirmation of being on track, not as a warning.

"The way that an exponential curve feels is it's growing so quickly that the kind of emotional feeling is it can't possibly keep going. But the nature of an exponential curve is it doesn't just keep going. It keeps accelerating, right? Exponential growth is accelerating... when you look at the curve in the industry, the fundamental thing is it is on track." — Mark Zuckerberg 00:53:52

3. Companies Identified

Chan Zuckerberg Biohub (CZI Biohub)

Nonprofit scientific research institute with locations in San Francisco, New York, and Chicago. Mentioned as the primary philanthropic focus of Mark Zuckerberg and Priscilla Chan, with a $500M commitment to the Virtual Biology Initiative. Building hierarchical world models of biology combining frontier AI with frontier wet-lab science.

"We just want to give tools to the whole scientific community... We'll have a bigger impact by getting this in more scientists' hands quicker by doing it as open-source projects instead." — Mark Zuckerberg 00:17:24

Human Cell Atlas

Large international database of single-cell transcriptomes, co-funded by CZI. Mentioned as a foundational data infrastructure that emerged from early Biohub single-cell sequencing investments and is now a core corpus for transcriptomic models.

"We funded the human cell atlas, which is now one of the largest databases of single cell transcriptomes." — Mark Zuckerberg 00:06:48

Cell by Gene

Open annotation and data platform built by CZI for single-cell RNA data. Mentioned as an example of a tool that spawned a self-organizing scientific community and became a foundational corpus for transcriptomic-based AI models.

"Then a community came around Cell by Gene and started contributing more and more data that we had nothing to do with sort of creating or funding or making happen in the world. And now Cell by Gene is a corpus of knowledge that a lot of the transcriptomic-based models are based off of." — Mark Zuckerberg 00:07:07

CHOP (Children's Hospital of Philadelphia)

Leading pediatric research hospital. Mentioned for delivering the first personalized CRISPR therapeutic to a child ("baby KJ") for a rare metabolic disease — cited as a landmark example of how disease and delivery selection can unlock first clinical applications of gene editing.

"We were all inspired by baby KJ, when the team at CHOP was able to deliver a CRISPR therapeutic to edit a mutation that he had... because we needed to target his liver cells." — Priscilla Chan 00:34:12

Evolutionary Scale

The company Alex Rives co-founded before joining Biohub, working on protein language models. Mentioned to contextualize the significance of his transition to lead the frontier AI effort at Biohub.

"You started Evolutionary Scale and you'd raised venture and you were making progress in your models. What was the pitch from Mark and Priscilla where you said, like, that's actually the right way to go after the mission?" — Sarah Guo 00:08:27

4. People Identified

Head of Science at CZI Biohub; previously co-founded Evolutionary Scale, a venture-backed protein language model company. Described as primarily an AI researcher with a biology background — a deliberate inversion of the prior Biohub leadership profile. Led the development and release of ESM Fold 2.

"Prior to Alex leading the effort, the previous leaders of the Biohub were basically primarily biologists who were interested in technology. And now... you are primarily an AI researcher who has a background in biology. I think that that's like a deep reflection on the way that we expect that this is going to drive more value in the future." — Mark Zuckerberg 00:52:58

Priscilla Chan

Co-founder of CZI, physician trained at UCSF. Described as the person who articulated the vision of Biohub as an integration of frontier AI and frontier biology. Has hands-on experience working with patients and rare disease communities, grounding the abstract mission in clinical reality.

"I think I had developed conviction that this is really a new era of science that's just beginning — what's going to be possible with artificial intelligence... you would really need to have frontier artificial intelligence. You would have to have frontier biology." — Alex Rives (speaking as Priscilla) 00:09:04

Jennifer Doudna

Nobel laureate, inventor of CRISPR-Cas9 gene editing. Mentioned as a partner in CZI's CRISPR Cures program at UCSF, working on translating research to clinical deployment.

"We partner with Jennifer Doudna on a CRISPR Cures program at UCSF." — Priscilla Chan 00:27:50

Elad Gil

Host of No Priors; investor and entrepreneur. Noted for having a PhD in biology and nearly a decade of wet-lab experience, providing him unusual depth to interrogate the scientific claims. Recruited half-jokingly by Mark Zuckerberg during the episode.

"Are you looking for a job?" — Mark Zuckerberg 00:12:40

5. Operating Insights

Build the Closed Loop First — Open-Loop Systems Cannot Compound

Elad Gil articulated a crisp structural principle that applies well beyond biology: progress accelerates when feedback loops are closed. Code and AI research iterate fast because results feed back immediately. Biohub's unique contribution is forcing the same closure in biology by coupling data generation, modeling, and wet-lab validation into one institution.

"The most important aspect of what you're doing is you're actually closing the loop with the actual biology because with code and research, it's closed loop systems. And so they're very fast to iterate. This is an open loop system. So you're closing a loop. And that's crucial to progress." — Elad Gil 00:54:51

Sequence Research Agendas as a Constraint-Optimization Problem, Not a Vision Problem

Rather than pursuing all layers simultaneously, the Biohub team explicitly frames prioritization as constraint management: given limited compute and data, where on the Pareto frontier do you want to be? This keeps world-class focus rather than spreading resources.

"You can have a lot of great ingredients and that doesn't guarantee that you succeed... it's just kind of a cool thing about the world is that people obviously are drawn to different missions... constraint optimize and make enough progress to do world-class work at one thing at a time while planting some seeds that can blossom over the next couple of years." — Mark Zuckerberg 00:48:56

Validate Computational Outputs Experimentally Before Publishing — Cryo-EM as the Ground Truth

The Biohub workflow explicitly pairs every model prediction with wet-lab biophysical confirmation (cryo-EM, functional assays). This is an operating discipline that prevents the field from chasing artifacts and establishes credibility.

"I think it's really critical to actually go and characterize these molecules in the lab... we designed proteins for several therapeutically relevant targets and we're able to confirm their function... you can see atomic resolution at the binding interface is correct." — Alex Rives 00:31:34

Team Stability Compounds Disproportionately in Fast-Moving Research Fields

In an environment where every week brings a major update, having a small, tight-knit team that has worked together before is a structural advantage that is systematically undervalued.

"This is like an extremely talented group of people who also know each other and work well together and are stable and good. And I think that also is underestimated in terms of the compounding benefit of people being able to work well in a stable environment over time." — Mark Zuckerberg 00:52:33

6. Overlooked Insights

Agentic Systems Integrating With Biology World Models Is Already Happening — One Week Post-Launch

The most consequential early external use of ESM Fold 2 was not manual scientific queries but automated agentic loops that connect the world model to end-to-end protein design pipelines. This was mentioned in a single sentence but signals that the combination of agentic AI and a biology world model — not either alone — is the actual near-term breakthrough surface. Any company or investor thinking about AI drug discovery should be watching this integration point specifically.

"One of the really interesting things that we've been seeing is people connecting it with agentic systems to just do automated design and just automate that whole process. So it's really another example of how you can see bringing together agentic and frontier AI with the ability to have a world model for biology and actually reason about biology and really start to automate the entire design process." — Alex Rives 00:46:19

A 96-Well Plate Is Now Sufficient to Validate Hundreds of Thousands of Digitally Generated Drug Candidates

This throwaway operational detail reframes the economics of early drug discovery entirely. The traditional bottleneck — screening millions of compounds in high-throughput lab experiments — has been compressed to a trivially small and cheap wet-lab step. This means the cost and time curve for generating validated nanomolar binders has effectively collapsed at the preclinical stage, which has enormous implications for biotech formation economics, venture return profiles in the space, and the competitive moat of incumbents with large screening infrastructure.

"Really in a small number of experimental trials, basically like a 96-well plate, you know, select from hundreds of thousands of trajectories digitally. Actually synthesize, you know, 96 proteins, test them in the lab in a really kind of short, easy experimental cycle. And we found nanomolar binders there." — Alex Rives 00:29:57