Recursion lines up $50M Nvidia collab for AI-powered drug discovery

Recursion claims that its artificial-intelligence-backed drug discovery platform gets smarter each time it pinpoints a new drug candidate—which is why, even with several of those candidates already in clinical trials, the company is still making moves to strengthen the technology.

The latest of those moves saw Recursion unveil a partnership with Nvidia Wednesday. Under the terms of their agreement, Nvidia will give Recursion both $50 million in private investment in public equity financing, plus access to its own cloud-based tools for AI-powered drug discovery.

Recursion’s stock price more than doubled on the news of the team-up: It rocketed up to $14.66 when the markets opened Wednesday morning, a 116% boost over Tuesday’s closing price of $6.78. The share price remained above $10 as trading continued Wednesday—marking Recursion’s first time past that mark since mid-November.

Through the funding from and collaboration with Nvidia, according to Wednesday’s announcement, Recursion plans to develop new AI models for drug discovery on the Nvidia DGX Cloud—billed as “AI-training-as-a-service” software. Fueling that training will be Recursion’s own 23-petabyte dataset of biological and chemical data, which comprises information about 3 trillion gene and compound relationships.

Recursion will also have access to Nvidia’s other AI and cloud-computing services and expertise to help fine-tune and scale up the models.

The resulting AI tools could potentially be made available for licensing by other drug hunters via Nvidia’s BioNeMo, a cloud service that launched earlier this year with a specific goal of speeding up the use of generative AI to create potential new drug compounds.

“With our powerful dataset and Nvidia’s accelerated computing capabilities, we intend to create groundbreaking foundation models in biology and chemistry at a scale unlike anything that has ever been released in the biological space,” Recursion CEO Chris Gibson, Ph.D., said in the announcement.

Separately from the development of those new AI models, Recursion noted in the release that it also plans to use the BioNeMo software to advance its own drug discovery work, including projects conducted in tandem with current and future partners in pharma and biotech.

The Recursion OS platform has already helped the company discover and develop a handful of drug candidates spanning a range of diseases and therapeutic areas. The platform is equipped with machine learning algorithms that scan through Recursion’s massive database of genetic and pharmaceutical information to identify potential drug targets and the compounds that might effectively reach them.

AI-mined drugs already in phase 1 or 2 trials include Recursion’s candidates to treat a trio of rare diseases, plus another two oncology therapeutics targeting certain mutant cancers and familial adenomatous polyposis, a condition that causes dozens of polyps to develop in the colon and can progress into cancer if left untreated.

Meanwhile, Recursion is in the late-discovery and preclinical phases of developing several more potential new cancer drugs, and it’s also in the early discovery stage for “more than a dozen” programs across rare diseases, oncology, neuroscience and immunology, according to its website.