Recursion Pharmaceuticals Wants to 4X Its Supercomputing Power for AI With the Help of Nvidia. Here's Why It Matters.

For a biotech like Recursion Pharmaceuticals (RXRX 0.57%) that aims to use artificial intelligence (AI) and petabytes of data as tools for drug discovery and development, computing power is a key resource. Without a lot of computational capacity, there's simply no way for the company to comb through billions of different molecules and calculate whether they might be useful in the context of the billions of biological targets that could be relevant to treating diseases.

With the help of its collaborator, chipmaker Nvidia, Recursion now intends to become the biotech industry's computational powerhouse. If it succeeds, the benefits could be so great as to be nearly unimaginable. Here's what it's doing and why the upside could be so big.

Adding computing power is an important step

On Nov. 9, management shared a few key updates with investors that sketched out a plan for a major expansion of the company's capabilities. The first step was to sign a collaboration with Tempus Labs to access its vast trove of 20 petabytes of oncology data, which includes genetic information, biomarkers, and health records. It will pay Tempus $160 million in cash or equity over the next five years for the privilege.

Now, Recursion has a grand total of more than 50 petabytes of biological, chemical, and health data under its purview. But the leads for the therapies of the future that are doubtlessly hiding in that data will only be possible to identify and exploit with the help of a vast amount of computing power.

That's why it's deepening its collaboration with Nvidia and adding 500 graphics processing units (GPUs) to its BioHive-1 supercomputer, quadrupling its computational capacity in the process. As a result, the business will own one of the world's most powerful supercomputing systems, making it the leader in the biopharma sector's computing capacity.

The move builds on its ongoing attempts to develop specialized large language models (LLMs) and foundation models (FMs) capable of streamlining and improving the efficiency of the drug discovery process, and it will absolutely bolster a core element of its value proposition to prospective collaborators. But the scale-up will take some time to pay off.

But what can it actually do with all that power?

In the short term, essentially the faster computer means that Recursion can do even more of the same things that it has been doing so far. It will be screening for promising leads for development, then advancing those leads into pre-clinical studies, and eventually moving the best pre-clinical programs into clinical trials.

It currently has five clinical-stage programs, four of which are indicated for rare disease, and one for oncology. Given that it doesn't yet have any recurring revenue sources, and that it only has cash and equivalents of $387 million against trailing-12-month operating expenses of $315 million, it probably can't afford to increase the rate at which it initiates clinical-stage programs in humans at the moment. Therefore, it could potentially make some money by selling its most promising leads to other companies.

Collaborators like Bayer are likely to be the first in line. In November it amended its collaboration agreement such that the pair can now launch as many as seven new oncology programs based on the targets it identifies. A total consideration of $1.5 billion is on the table, pursuant to reaching the specified development milestones, and Recursion could make some royalties on any commercialized programs, too.

In the long term, the financial returns could be far more vast. Management's vision is for AI to trawl through its data to find high quality leads, follow up on those leads with robotics-enabled confirmation experiments, and then pass the vetted candidates on to human laboratory staff for further late-stage preclinical follow-up investigations.

In other words, it wants to fully automate many of the high-effort and low-return activities inherent to drug development. If Recursion succeeds, it could either be a pharmaceutical powerhouse, with dozens of drugs in development and commercialization at any given time, or the most popular collaborator in the biopharma sector. Both outcomes would make it a highly lucrative investment.

But today, despite its ongoing efforts, Recursion is an exciting business with an unproven thesis. Nobody doubts that drug discovery efforts can be accelerated with the help of extra computing power, or that having a lot of data to sift through leads to a longer list of potentially useful targets. The question is whether it's possible to make a profitable business after accounting for all of the various costs involved.

Until it can answer that question by making more than it spends, this is a risky stock that isn't right for everyone. Still, if you're usually comfortable enough with the risks inherent to biotech stocks, Recursion could be right up your alley. Just be aware that it might be quite some time before its revenue flywheel is operating at the capacity it will need to survive in the long term.