Amazon And Nvidia Backed EvolutionaryScale Raise $142 Million For AI That Makes Proteins

EvolutionaryScale, a relatively new firm, has raised a sizable sum of money to develop AI models that can produce unique proteins for scientific study.

On Tuesday, said that it had secured $142 million in a seed round headed by Daniel Gross, the former CEO of GitHub, Amazon, and NVentures, Nvidia’s corporate venturing arm.

The startup also released ESM3, an artificial intelligence model known as a “cutting edge model” in biology that can design proteins for use in drug discovery and materials science.
ESM3 is a step towards the advancement of biology, where artificial intelligence is a tool for engineering from first principles, in the form of structural design, mechanics and mechanical engineering, and programming software.

Scientist Alexander Rives in a report

synthetic AI models — image credit-finimize

In 2019, Rives began building synthetic AI models to explore proteins at Meta’s AI research lab, FAIR, with Tom Sercu and Sal Candido.

Structure and composition

EvolutionaryScale makes all of its 98 billion model parameters available for non-commercial use through its cloud-based Forge developer platform, as well as offering limited versions of the model for offline use.
EvolutionaryScale plans to use ESM3 to produce a new type of green fluorescent protein (GFP), which is responsible for the glow of jelly and the color of coral lights.
For example, partner with pharmaceutical companies to integrate ESM3 into their workflows or share revenue from unexpected discoveries commercialized using ESM3 with researchers.
To achieve this goal, EvolutionaryScale said ESM3 and its products will continue to be available to select AWS customers through the cloud provider’s SageMaker AIdev platform, Bedrock AI platform, and HealthOmics services.
ESM3 will also be available to select customers using Nvidia NIM engineering services, supported by the Nvidia Infrastructure Software License.

According to EvolutionaryScale, AWS, and Nvidia, customers can enhance ESM3 with their data

In the company’s promotional materials obtained by Forbes last August, EvolutionaryScale stated that it would be ten years before AI models can help design treatments.

“The incredible speed of new AI advances is being driven by larger samples, larger data sets, and computational power,” said an EvolutionaryScale spokesperson.

That’s how life is

We found that as the size of language samples increases, we deepen our understanding of basic biological principles and reveal biological structure and function.