Newsletter
Despite all the concerns surrounding generative AI, the technology has proven to be useful if implemented correctly. Now, in a recent development, the Chan Zuckerberg Initiative (CZI) has announced its plans to develop a state-of-the-art GPU cluster with the goal of curing, preventing, and managing all diseases by the end of this century.
At its core, the Chan Zuckerberg Initiative will acquire over 1,000 H100 GPUs to train AI large language models (LLMs) specifically designed for human cells. This approach aims not only to deepen our understanding of how the body reacts to diseases and novel treatments but also to enable researchers to guide a “virtual cell” through a series of simulations.
“LLMs have done an impressive job at helping us understand protein structure, and we think they will be equally great at helping us understand more complex structures like cells,” said Jeff MacGregor, CZI vice president of communications.
Making the tools publicly available
What sets the initiative apart is its commitment to making these tools accessible to a broader spectrum of scientists, considering the exorbitant costs associated with utilizing such powerful computing resources. Moreover, to develop these AI models, CZI will source data from existing datasets, including information curated by a Chan Zuckerberg software tool that profiles approximately 50 million unique cells, as well as resources from CZ Science research institutes.
“AI models could predict how an immune cell responds to an infection, what happens at the cellular level when a child is born with a rare disease, or even how a patient’s body will respond to a new medication. We hope that this collaborative effort will generate new insights about the fundamental characteristics of our cells,” said Chan Zuckerberg.
However, it is important to note that the CZI’s Biohub Network is responsible for acquiring the GPUs, and the project aims to address these scientific challenges over a 10- to 15-year timeframe. Furthermore, the project’s scale is not comparable to similar systems deployed by private sector entities for commercial applications.