Technological advancement makes people uncomfortable, and artificial intelligence (AI) is no exception. It’s going to uncover things we don’t like, things we cannot explain, and things that truly impact mankind – maybe even all at the same time. How it can possibly add so much value may remain a mystery at times. For us to truly benefit from AI, sometimes we need to forgo explain-ability and focus on the end result.
Take Zymergen as an example. Their job is to use machine learning to change microbes so they do things like produce chemicals using natural fermentation. In their S-1 filing, Zymergen admitted that nearly 60% of the beneficial changes they discovered for their microbes defy known human explanation, even after the fact. One domain where such a technology proves extraordinary useful is in drug discovery, where the end result is the only thing that matters.
Back in 2017, we first wrote about Utah’s own Recursion Pharmaceuticals (now going by the name Recursion), a company that was doing something extraordinary – they were actually looking at cells using computer vision to see how they physically reacted to drugs. At that time, they had identified over 1,000 features to use that could determine if a sick cell was getting healthier when exposed to 1000s of drug compounds. Several years later, we checked in on Using Artificial Intelligence for Drug Discovery noting that Recursion had raised over $100 million in funding along with a full drug pipeline in the works.
Fast forward to today and Recursion has raised just over $465 million in funding with over half that coming in the form of a Sept 2020 Series D of $239 million led by German drug giant Bayer. Recursion has also filed for an Initial Public Offering (IPO) using the proper method, none of that SPAC rubbish. That means we have a proper S-1 with enough information to make an informed decision about whether or not to invest in the offer.
While Recursion is not pre-revenue, what’s trickled in so far doesn’t move the needle much. Investors who decide to pick up shares here will be basing that decision on what the company promises in the future – which is nothing short of changing how drugs are discovered.
Changing Drug Discovery Using AI
“The industrial revolution of drug discovery is here,” says the company, and by that they mean changing the drug discovery process so that more options can be considered at the onset while at the same time decreasing the time it takes for approval.
Software developers of the word, answer this. How can you possibly increase the number of compounds you’re considering (increase features) while at the same time decrease failures (increase quality), all while shortening the drug discovery life cycle (decrease time-to-market)? Two words. Big Data. Below, you can see how Recursion’s big data set is growing exponentially as their platform scales while simultaneously driving down the cost of each experiment.
Lots of data means lots of complexity, so let’s try and distill this down to three things for investors to consider from the horse’s mouth.
Three Value Drivers
When you’re dealing with something as complex as drug discovery, you need to try and keep things simple. Recursion does this by giving us three value drivers to focus on – pipeline, partners, and platform.
Recursion lays claim to “one of the largest, broadest and deepest pipelines of any technology-enabled drug discovery company.” A total of 37 internally-developed programs are focused on areas of significant unmet need, four of which are in Phase 1 clinical trials. Several could be blockbusters (over $1 billion in annual sales). The ten most notable can be found in the obligatory pipeline graphic. For investors, just one of these getting to market would prove the model works. Without beginning to sell one or more of our drug candidates, the company is unlikely to generate significant revenues. To be fair, they’re really eating their own dog food here.
With every new discovery, our platform gets smarter.
Using microscopic images of cells to determine how compounds affect them sounds cool because it is. But when combined with some artificial intelligence, it gets a whole lot smarter.
Recursion started out doing experiments using brute force methods. That seemed successful enough, given that it produced 8 of the 10 projects in the aforementioned pipeline. In mid-2020, Recursion moved away from brute force methods to inference. In other words, they’ve added more artificial intelligence into the process. What would have taken 1,000 years to execute using brute-force search can now be inferred in a matter of months. That breakthrough coincided with a five-year partnership being established with the fourth largest pharmaceutical company in the world by 2020 revenues – Bayer.
In the world of pharma, the size of someone you’re in bed with is a double-edged sword. The bigger the partner, the more prestige it gives you. But if that big partner decides to leave the partnership on ice or drop it entirely – could be for simple reasons such as a reorg or change of focus – then don’t expect others to be lining up. Not to mention such news is going to significantly increase share price volatility.
Bayer emerged on the scene in August of 2020 when they signed a five-year agreement with Recursion. Over that time, Bayer may initiate approximately ten research projects related to fibrosis across multiple organ systems, including lung, liver, and heart. These projects come with the usual revenue possibilities – option exercise fees, development and commercial milestones, tiered royalties, and the like. Initiation of these programs is at the discretion of Bayer, but payment of $30 million to Recursion in Sept 2020 helps show their commitment to the partnership. Not all of that money is reflected in Recursion’s revenues (seen below) because of the bean counters.
Bayer also contributed approximately 500,000 compounds from its proprietary library for Recursion to sick their algorithms on.
To Buy or Not to Buy
We’re adamant about not investing in companies pre-revenue, but Recursion tempts that rule. The sheer breadth of compounds they’re exploring, their bedmate Bayer, and the large amount of big data that’s growing at an exponential rate are all reasons we think Recursion won’t go the way of Bind Therapeutics. We’re keen to invest in artificial intelligence for drug discovery, but we prefer software-as-a-service business models such as what Schrodinger offers.
We’ve written about IPOs in 2021 noting that far too much irrationality is driving today’s stock prices. Many newbie investors are likely panicking right now as stock prices understandably go through a correction. (Experts have shown that people feel the pain of losses twice as much as they feel the joy of gains.) We’ve looked at quite a few AI drug discovery companies over the years, and leaders are starting to emerge (Atomwise is another example of an emerging leader). We like Recursion, but will wait for the dust to settle following their IPO.
St. Steve Jobs once said that the next big breakthroughs will come at the intersection of technology and biology. When we first read about how Recursion was using computer vision to actually look at cells and see how they responded to drugs, we were blown away by it. Clearly it was no gimmick. What this little Salt Lake City company does is remarkable, but then we need to remind ourselves. We don’t invest in stories, we invest in proven business models. Recursion still needs to prove themselves by using their platform to grow some meaningful revenues.
Should the IPO go through as planned, Recursion stock will trade under the ticker RXRX.
Want to see which five AI stocks we’re holding? Become a Nanalyze Premium subscriber and find out today.
View original post