Nvidia Invests Further Into Healthcare And Releases The Largest Biology Foundation Model With The Arc Institute

Famed chip maker and hardware giant Nvidia has been investing significant resources into the healthcare and life sciences spaces. Its latest move was announced last week as it launched, in partnership with the Arc Institute, the largest biology foundation model developed to-date.

The new model, named Evo 2, was built on the Nvidia DGX Cloud platform and was trained on a dataset of nearly 9 trillion DNA and RNA subcomponents (nucleotides) and 128,000 genomes, representing perhaps one of the largest and most monumental efforts globally to understand the building blocks and genetic code for all domains of life.

Importantly, given how extensive the model’s training is, it has incredible efficacy in “predicting the form and function of proteins based on their genetic sequence, identifying novel molecules for healthcare and industrial applications, and evaluating how gene mutations affect their function.”

These insights into DNA, RNA and proteins are especially important in understanding organisms’ genetic codes, gene expressions and the ways that disease impacts those expressions—meaning that practical applications range from drug discovery and development to agriculture and other bio-engineering feats.

As described in a Nature article, the model “can write whole chromosomes and small genomes from scratch. It can also make sense of existing DNA, including hard-to-interpret ‘non-coding’ gene variants that are linked to disease.”

The Arc Institute, which pioneered this work alongside Stanford University, provides scientists with the tools and resources necessary to tackle complex scientific challenges and pursue a research-only mindset rather than worry about administrative tasks related to securing funding. Per the institution, Evo 2 represents “a biological foundation model that is trained on a representative snapshot of genomes spanning all observed evolution. Emphasizing generalist capabilities over task-specific optimization, Evo 2 achieves robust prediction and generation performance from molecular to genome scale and across all domains of life.”

Without a doubt, Nvidia is slowly but surely increasing its presence in life sciences and healthcare. I have written previously about its work with digital twins and other potential applications in healthcare, in addition to its venture capital arm which is also investing hundreds of millions of dollars into the sector. Given that Nvidia already provides the compute and hardware infrastructure for nearly every other company that is involved in healthcare AI applications, this focus makes complete sense.

Furthermore, healthcare and life sciences are both booming sectors with regards to artificial intelligence applications. Many other companies are also working at the intersection of technology and biology, given the numerous challenges that are present in the fields of drug discovery and protein folding. For example, Deepmind and Isomorphic Labs have made immense progress with AlphaFold, another leading foundation model ecosystem to better understand protein folding. Meta created something similar with its ESM Metagenomic Atlas. Given the increasing rates of catastrophic disease and the rapidly evolving nature of pathogens, scientists in these sectors hope to use the best of the advancements in AI to help solve some of biology’s toughest challenges.

Indeed, the immense progress that has been made thus far has paved the way for monumental scientific inventions and developments to emerge in the years ahead. Undoubtedly, this work is just getting started.