Notes - CZI's Mission - AI and Basic Science Acceleration

November 5, 2025

The Chan Zuckerberg Initiative (CZI) Mission and Strategy

The Chan Zuckerberg Initiative (CZI) was founded almost a decade ago with the intent to cure, prevent, and manage all disease by the end of this century. This highly ambitious goal initially made most scientists "couldn't look at us with a straight face". The strategy is not to cure all diseases directly, but to help scientists and the scientific community cure all diseases by accelerating the pace of basic science.

CZI's focus stems from the realization that major scientific breakthroughs are usually preceded by the invention of a new tool to observe phenomena in a new way (e.g., the microscope or telescope). Priscilla Chan, trained as a pediatrician, realized the limitations of current medical understanding when caring for patients, leading her to champion basic science as the necessary "pipeline of hope".

Filling a Funding Niche

CZI addresses a funding gap in science. While the government (NIH) parcels out funding into small grants for individual, near-term investigations, CZI develops tools that are longer term and often more expensive. These projects can cost on the order of $100 million to a billion dollars over a 10 to 15 year period. Since developers often do not "get credit for the tools," this is a space well-suited for philanthropy.

The Role of AI and Grand Challenges

CZI's work pairs frontier biology with frontier AI. AI is seen as both the "most overestimated and underestimated technology ever". The Biohub, the current main focus of CZI's philanthropy, aims to produce specific data sets necessary for training AI models, particularly to build virtual cells.

CZI selects grand scientific challenges that fit a 10 to 15 year time horizon. These challenges must have a "credible pathway" but retain enough ambiguity and risk to offer substantial returns if successful.

AI advances suggest the mission to cure and prevent disease may be possible significantly sooner than the end of the century.

CZ Biohub Network and Focus Areas

CZI operates Biohubs across the U.S., partnering with local academic institutions to facilitate collaborative, cross-disciplinary work:

The organization is moving towards a unification under Alex Reeves, an AI expert who understands biology, to connect frontier AI and biology. This centralization is intended to complete the "flywheel"—where models identify data gaps that inform the Biohub's next instrumentation and data set generation.

Key Tools and Data Initiatives

1. Cell by Gene Atlas

The inspiration for CZI's single-cell work was the lack of a "periodic table of elements equivalent for biology".

Origin: The Cell by Gene initiative began 10 years ago by funding the methodology to standardize single cell science.
Purpose: The tool was originally built as an annotation tool to break a workflow bottleneck—scientists needed a faster way to annotate their single cell data.
Impact: By providing a standardized annotation tool and sharing data publicly, CZI fostered a network effect. The broader scientific community contributed 75% of the data to the Cell by Gene atlas, compared to the 25% funded by CZI.
Accessibility: The user interface for Cell by Gene was intentionally designed to have a low barrier to entry, allowing people without deep computational or biological backgrounds to contribute and use the resource.

2. Virtual Cell Models (VCMs)

VCMs are considered one of the most important sets of tools CZI is building. The goal is to develop an increasingly general model of a virtual cell by building a hierarchy starting from proteins, moving up to cellular structures, and eventually simulating complex systems like a virtual immune system.

VCM Utility:

Hypothesis Generation: VCMs help scientists generate hypotheses for their work.
Derisking Wet Lab Work: They allow for in silico testing and tinkering on the computational side, enabling researchers to ask riskier questions and obtain a directional signal before making costly and slow investments in the wet lab.
Precision Medicine: VCMs will be critical for enabling precision medicine by predicting off-target effects and connecting mutations to protein expression.

VCM Types and Technology: Current models include:

Variant Former: Trained to predict the outcome of applying CRISPR edits to a cell.
Diffusion Model: Can produce a synthetic model of a cell based on a description.
Reasoning Models: An emerging focus aiming to move beyond spitting out correlations to actually reason through how things would evolve and why they happen in biology.

Precision Medicine Vision

CZI believes that most diseases should be thought of as rare diseases because individual biology is unique. Current medical approaches often group people based on limited characteristics (age, demographics, ancestry), leading to trial-and-error treatments.

The ability to connect mutations to downstream cells and protein expression allows scientists to find specific targets and predict potential side effects. Ultimately, CZI seeks an "explosion of a community" deploying precision medicine.

Organizational Approach and Resources

CZI draws inspiration from physics, where labs historically rally around big projects and shared resources. CZI believes encouraging collaboration between disciplines—like having biologists sit next to engineers—is key to solving organizational problems and unlocking value.

The organization's expansion is not primarily focused on square footage but on compute. CZI is scaling its GPU cluster from 1,000 to a planned 10,000 GPUs. This extensive resource is used internally and made available to external scientists through an application process to seed collaborations.