Skip to content
No PriorsNo Priors

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

Biohub started with an ambitious goal of curing, preventing, and managing all disease by the end of the century. A decade later, thanks to the convergence of frontier AI and biological data, that goal may have been too conservative. In this episode, Elad Gil and Sarah Guo sit down with Biohub co-founders Mark Zuckerberg and Priscilla Chan, alongside Biohub Head of Science Alex Rives. Together, they discuss Biohub’s $500 million virtual biology initiative, which integrates frontier AI with wet-lab work to build predictive world models of cells, proteins, and systems. They also talk about their newly announced open-source engine for digital protein and antibody design, ESMFold2; why Biohub is a nonprofit rather than a venture-backed startup; and how hierarchical simulations will soon allow doctors to treat patients at an individual, mechanistic level. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @Biohub | @finkd | @alexrives | @ChanZuckerberg Chapters: 00:00 – Cold Open 01:02 - Mark Zuckerberg, Priscilla Chan, and Alex Rives Introduction 01:26 – Why Biohub and Their Mission 08:27 – Integrating Frontier AI and Frontier Biology 09:45 – Micro to Macro Biological Modeling 14:22 – Mechanistic Interpretiability 16:58 – Why Biohub is a Non-Profit 21:41 – Understanding How Biology Works 24:23 – Timeline for Curing All Diseases 26:25 – Translating Research to Patient Impact 28:04 – Launch of ESMFold2 32:13 – Tackling Off-Target Effects and Edge Cases 38:39 – Putting the Tech in Individual Hands 41:06 – Talent at Biohub 44:25 – What’s Next After ESMFold2 46:10 – Connecting ESMFold2 to Agentic Systems 46:51 – The Virtual Cell 49:33 – Defining Success for Biohub 51:52 – Biohub Strategy Update 56:20 – Conclusion

Mark ZuckerbergguestAlex RivesguestSarah GuohostElad Gilhost
Jun 10, 202656mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

Biohub’s open-source AI and biology platform to accelerate medicine

  1. Biohub’s core thesis is that the fastest path to medical breakthroughs is building shared, open tools and datasets that the entire scientific community can use, rather than trying to “cure diseases” as a single centralized organization.
  2. They argue biology is data-constrained in a fundamentally different way than internet-trained AI, so Biohub pairs frontier AI with “frontier biology” (new instruments, assays, imaging, and engineered systems) to create the missing data needed for robust models.
  3. Their modeling roadmap is explicitly hierarchical—proteins to cells to systems—aiming ultimately at a “virtual cell” that can generalize to new interventions and connect genotype→molecular mechanism→phenotype.
  4. A key research direction is mechanistic interpretability for protein language models, using model representations to uncover latent biological structure/function relationships and generate new hypotheses, not just predictions.
  5. ESMFold2 is highlighted as an open-source protein world model enabling fast structure prediction and emergent protein/antibody design, validated with wet-lab assays and cryo-EM, with implications for reducing off-target effects and accelerating preclinical iteration.

IDEAS WORTH REMEMBERING

5 ideas

Biohub optimizes for ecosystem speed, not ownership of cures.

They repeatedly stress the goal is to accelerate the whole field by releasing tools/models openly so many labs can iterate in parallel, including on niche and rare diseases that market incentives often neglect.

Biology’s bottleneck is often *creating* the right data, not just training bigger models.

Unlike LLMs with abundant internet text, many key biological phenomena aren’t observable with existing methods, so Biohub invests in new assays, imaging, sensors, and cellular engineering to generate novel datasets.

A “virtual biology” stack requires hierarchical models with connective tissue between layers.

They argue you can’t reliably model systems-level behavior (e.g., immune dynamics) without solid molecular and cellular foundations, and they design experiments (spatial transcriptomics, developmental imaging, cell–cell signaling sensors) to bridge layers.

Mechanistic interpretability is positioned as a route to *new biological knowledge*.

Because protein language models learn from billions of sequences (including poorly characterized proteins), interpreting representation spaces could link unknown proteins to known function/structure “grammar,” generating testable hypotheses beyond black-box prediction.

ESMFold2 demonstrates an “open discovery engine” pattern: general model → emergent capabilities.

Rives emphasizes they didn’t build an antibody-specific model; a general protein world model yields protein/antibody interaction competence and enables digital exploration of designs followed by small-scale lab confirmation (e.g., 96-well validation, nanomolar binders).

WORDS WORTH SAVING

5 quotes

We'll have a bigger impact by getting this in more scientists' hands quicker by doing it as open-source projects instead.

— Mark Zuckerberg

The theory isn't that we're gonna cure the diseases. We're not. It's that we wanna help accelerate the pace of progress for the whole scientific field.

— Mark Zuckerberg

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.

— Alex Rives

So my goal is to be able to treat the individual as an individual, understand the mechanisms, and be able to intervene.

— Priscilla Chan

Our vision is not that there's gonna be like some central super intelligence that solves all of science. I think, like, people are really important and I think will be more important in the future, and giving people more tools to be more productive is gonna be like a critical part of any kind of positive future.

— Mark Zuckerberg

Open-source science and tool-building as philanthropyFrontier AI paired with frontier wet-lab biologyHierarchical biological modeling (protein→cell→system)Mechanistic interpretability for protein language modelsESMFold2: billion-protein folding and rapid structure predictionProtein/antibody design via digital search + small wet-lab validation loopsVirtual cell vision and ecosystem/agentic integrationsOff-target effects, toxicity prediction, and rare-disease accelerationNonprofit rationale: long horizons, capital scale, broad participationBiosafety and responsible open release considerations

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.

iOSAndroidClaudeChrome