A – Chris Gibson: Thanks, Morgan. That’s a fantastic question. I would say that the response has been really, really robust. We had many R&D heads of large pharma companies at our JPMorgan presentation, which we co-hosted with NVIDIA. We had CEOs of large companies there, both tech and bio. And what we heard from people is, how do I get access to something like this? And we are doing the work now to increase the robustness of the LOWE platform. We’re having conversations with potential partners around how we could put these tools in their capable hands in a way that would be helpful to Recursion and to the industry writ large. As far as guidance around revenue, I don’t think we’re going to give guidance around revenue in the near-term.
What I will say is that, we see the bigger opportunity in driving these companies towards really significant collaborations like the ones we’ve done with Bayer and Roche-Genentech, as they see the power of a tool like LOWE, probably that’s the bigger opportunity for us in the near-term compared to sort of recurring software revenue. But we certainly will take all the revenue we can get if we’re able to identify those questions. All right. Thank you. Next up, we have a question from Alec Stranahan of Bank of America who asks, how do you plan to utilize LOWE either internally or as an external offering? How does this fit into your existing full stack capabilities? This is actually a fantastic question because I think it highlights something that’s really important.
LOWE internally at Recursion is being used by certain teams on the BD side and elsewhere. It certainly is something we think pharma could use. But I’ll actually go to a slide from our other deck here to say that internally, we actually believe there’s a step beyond LOWE, where autonomous agents use a tool like LOWE to drive discovery as opposed to individual scientists and I think this is a great example of this. This is a plot of thousands of targets in human biology. And what I’m showing you here on the Y-axis is how we’ve used a Large Language Model that is based on public data sets, like the cancer dependency map, open targets, TCGA, et cetera. And we have profiled all of these different targets to assess their relevance in oncology.
Whereas on the X-axis, we’ve used a Large Language Model that’s looking only at proprietary data internal to Recursion. And so what you see on the top right is important targets like PIK3CA, BRAF, mTOR, EGFR, et cetera, where we see approved medicines for these targets in oncology. We see that these targets score robustly for oncology relevance based on both the public data and Recursion ‘s proprietary data. But we see hundreds of targets in the bottom right, in this blue box, that are now being automatically initiated as new pre-programs at Recursion without almost any human intervention based on our Large Language Model scores. And we see these as targets that have the potential to be totally novel. And so at Recursion, our scientists aren’t just using LOWE, they’re really using robust workflows that are highly automated.
And LOWE is more of a tool that we see to collaborate with partners, that we see to drive partnership progress through our pipeline. All right. Next question is from Jesse Brodkin [ph], who asks, why did you choose Tempol or REC-994 for your CCM indication when the vitamin D data looked better in our preclinical screens? Thanks, Jesse, for that question. There’s a circulation paper that you all can read about this work. And what we noticed was that both vitamin D and REC-994 had a robust response in the context of these preclinical models. However, REC-994’S response was additive on top of vitamin D. So there was vitamin D in the chow of the mice and the REC-994 treatment added to the effect that we saw. And given vitamin D is a very safe, widely available molecule that many people take in their everyday, you get when you stand out in the sun as well, we didn’t see a lot of added value in us bringing that program forward.
Whereas bringing REC-994, which was otherwise inaccessible to people, it was not approved, not available forward, we believe there was the potential for additive benefit. And that’s why we’ve driven that program forward and we’re so excited to read out the data in Q3. All right. It looks like next we’ve got a number of questions around our NVIDIA collaboration. The first from Harry Schoenberg [ph] at JPMorgan, who asks, what involvement will you have with NVIDIA in the near future and going forward? That’s a great question. I’ll go back to the slide on our NVIDIA collaboration here. And just to reiterate that with NVIDIA, we are really focused in three areas currently. The first is advanced computation. We’ve been working with the team there for many years.
We think they’re incredible and they’re helping us take the algorithms that we’re building and help scale them, help tune them. And there’s not many people in the world who have a lot of experience training multi-billion parameter models, but there’s a great team at NVIDIA that’s done just that. And so we are collaborating really closely on some of our larger models. What’s more, we’ve already demonstrated the use of our priority access from NVIDIA in our expansion of our BioHive supercomputer. And of course, there’s the potential for us to access the DGXCloud Resources in a priority way as well. And then, finally, we see the potential for us to put potentially additional tools on their BioNeMo marketplace as we continue to develop these tools.
And who knows? The collaboration with NVIDIA is very, very close and we know that our teams are constantly coming up with new ideas and we’ll be excited to try some of those out with our colleagues there in the near future. Next question is from Mark Simmons [ph] who asks, describe the relationship and investment with NVIDIA regarding AI and their products. I think we’ve really hit on this one already, so I will move on. Okay, the next question is anonymous. This is a good one. Why have insiders been selling shares each month? Do they not have confidence in the company? That’s a great question and I’m glad we’re addressing it. So I’ll speak for myself, because I think most people look to the CEO when it comes to insider buys and sells.
And in 2023, I traded a very small, relatively small number of shares. In fact, it was roughly about 4% of my holdings that were traded. And so all of these trades were done using 10b5-1 pre-planned sales and purchases. And again, I traded roughly 4% of my holdings. If you were to look at that at the grand scale, just on our volume today, all of the trades I did in 2023 represent roughly 6 or 7% of just the volume Recursion traded today in the market. And so I think while you see many of these sales, many of these purchases across insiders, the reality is that the magnitude of these is relatively small and we’re using these for making sure that we’ve got the right diversification in place. This is my first job out of grad school and so I have the vast majority of my shares that I’ve had from the beginning, the vast majority of my shares that I had at the IPO and I intend to keep the vast majority of my shares moving forward because I definitely believe in what we’re building here.
I’ve dedicated my life and my career to it. Next up, we’ve got questions in our fibrosis project. So Alec Stranahan asks, fibrosis has been a historically challenging area for development. This is true. How is the asset you unlicensed differentiated and what are the first disease areas of focus? Well, Alec, I really appreciate that question. I’m not going to share the first disease area of focus yet because the novel target we’re working on we think has the potential to be useful in multiple different areas. And so we’re going to probably hold that information back from a competitive standpoint for a while. What I will say is the differentiation here is that we used a very complex assay. We essentially looked for small molecules that were mimicking the effect of Pentraxin-2 in a complex fibrocyte assay.
And what we saw was a number of molecules. Since then, we’ve really optimized one of those molecules, 1169575, and additional molecules that we’re advancing as backups. And we think this novel mechanism, and if you knew the mechanism, I could tell you more, but we’re not going to share it yet, has a lot of potential to modulate the immune response that could be broadly useful across this space. So we’re aware of the challenging development space. We certainly could imagine partnering this program as we get into sort of the Phase 2 portion of the clinical trials. But we think this one is important and worth advancing because we’re unaware of anybody else taking this target or this target class forward in the context of modulating the immune system to drive a reduction in fibrosis.
All right. The next question comes from Jesse Brodkin, who asked, did Recursion pay bear any money to obtain the fibrotic disease lead candidate from the collaboration? So Jesse, this program was advanced under our original fibrosis collaboration and specific disclosures around the financial terms can be found in the 10-K. And we’ll be filing that 10-K here in the next 48 hours or so. So you can look there. But what I will say is we didn’t have to pay anything upfront. There’s some modest milestones that we think are very attractive as we drive this program forward. And I think both we and the scientific team at Bayer are pretty excited to see what we can do with Target Epsilon. All right. The next question, back to Morgan Brennan from CNBC.
And Morgan asks, what proof points can you share on AI, ML and medicine? And are AI applications in drug discovery happening as quickly and effectively as you anticipated? Morgan, it’s a great question. So I will share that I’m a founder and I don’t think any founder is ever satisfied with the pace that anything is advancing. So I can say no, things aren’t going as fast as I would have liked. But I think if you look back at where Recursion started in 2013, where other companies like us started, and where we are today, we now have developed at Recursion multiple tools that are state-of-the-art in terms of target identification, in terms of making ADME and Tox predictions. We have a pipeline of five programs in Phase 2, or nearing Phase 2.
I think we can be really proud of the platform we’ve built, the pipeline we’ve built, the partnerships we’ve built. Some of our partnerships are not only, our Roche-Genentech partnership is not only the largest partnership in TechBio today, it’s one of the largest partnerships ever disclosed in biopharma in terms of total kind of bio box potential. And so I think that while the next 12 months to 24 months is going to feel to all of us like we’ve kind of under-delivered, we’re on this sort of exponential curve where if we look back in five years to 10 years, we’re going to be amazed at how far things go. But the reality is, like with any new technology, it takes time. And if we run these virtuous cycles and we get 1% to 2% better each time, but we can compound those efficiencies through many, many cycles, I think over time we’re going to see a fundamental transformation of the biopharma space that over a decade is going to feel much more profound than most people believe today.
All right. Next up we have a question from Curtis Maxwell [ph] who asks, what is the backlog of projects that are in the pipeline for AI analysis and what is the cost per project and duration typically? So here we can actually look at, if you’re referring Curtis to our programs at Recursion, we can share some of these statistics. We believe at Recursion that we’re trying to shape this traditional V of the biopharma industry into more of a T, where one day we will be able to take all of our prior data and our algorithmic approach and predict the right molecule for each patient and drive it all the way to the market without any attrition. Now, that T is going to be impossible to actually completely achieve, but we want to move in that direction.
And you can see compared to the industry average, Recursion already starting to shape our internal funnel to look more like that T and less like that V. And what we’re able to demonstrate so far across our programs is that our cost to IND and our time to validated lead significantly outperform the industry averages. What’s next up is that we hope that we’re going to be in a position to demonstrate at least meeting the probability of success of the industry averages with a faster time and higher scale for the size of our company and every generation of future programs we hope will build on that. And one day we hope to be able to demonstrate to the industry that we can increase the probability of success of our programs. And we can drive them forward not only in areas of unmet need in rare disease and oncology, where we can be first in disease potentially, but also one day to leverage this platform to fast follow at scale, to be able to take programs at Recursion that we can drive extraordinarily quickly based on the incredible science that’s being done elsewhere in the industry.
So a lot of good work to come there. All right. Looks like we’ve got another question from one of our analysts here. Given the complexity and layering of data keeps growing on your platform, how would you define a proof-of-concept in a constantly moving platform? That’s a great question, Gil [ph], and I think this speaks to a difference in mentality across the tech and the bio industries. We believe that these virtuous cycles of learning and iteration must always be running. And that increases the challenge of keeping up with the latest tool — the latest version of that tool. But we want to make sure that every program at Recursion uses the latest generation of every tool that we’re building. And that’s why we talk about the generations of our clinical pipeline, the first-generation programs, which are by and large focused in rare genetic diseases before we had a chemistry team.
So most of those first-generation programs are actually molecules where we used our ML and AI platform to identify a new opportunity for a known chemical entity. And you’ll see in our second-generation, you’ll start to see the layering in of new chemistry and digital chemistry tools to these programs as we advance them forward. And as we run a third-generation and a fourth-generation in the future, I think, you’ll see we hope that this platform learns, that this platform improves and that every generation of programs will have, on average, an increasing probability of success and we hope increasing impact. All right. Let’s go now to some investor and revenue questions. We’ve got a question here from Eric Joseph at JPMorgan. How should investors generally be thinking about the company’s business model at this stage?
Eric, that’s a fantastic question. At the end of the day, in our industry, the currency of impact, the currency of success is assets in the clinic. And I think that’s why Recursion has not just focused on building software-as-a-service, not just focused on our partnerships, but has a robust internal pipeline that we’re advancing in a small — in areas of small niche corners of biology with high unmet need and partnerships where we can go after large, intractable areas of biology. We are always doing business experiments at Recursion. LOWE is a business experiment. Phenom-1 was a business experiment. And we don’t yet know how those will drive our business model per se, but what I’m confident in is that Recursion will always be focused in bringing new composition of matter into areas of biology with high unmet need or where we can drive down the costs of expensive molecules that have been advanced into the market.
So I think you can count on that being at the core of what we’re building at Recursion, but we’re going to do all of that with a much more tech-focused mindset than I think many other companies in this space. All right. Back to Gil, one of our analysts. Do you anticipate that over time, more value will be created from the company’s internal pipeline or through its partnerships? Well, Gil, if we’re talking about long-term, I believe Recursion is going to generate much more value from our internal pipeline than our partnerships. We expect to generate significant value in our partnerships today. We signed these partnerships with Roche-Genentech and with Bayer because we saw them as having transformational potential for patients and the potential for extraordinary impact in areas of high unmet need.
But as each of those partnerships finishes, we expect to have learned what we need to as a company to be able to build our own internal pipeline into those more complex intractable therapeutic areas. And until every disease has a treatment, we won’t rest and so I think you can count on Recursion ‘s internal pipeline being a robust primary driver of our growth if we’re to look out over the intermediate and long-term. All right. Back to Eric Joseph at JPM. What’s envisioned as its earliest and most significant lines of product revenue? I assume that Eric’s talking here about some of our software tools like LOWE. Eric, we’re having lots of discussions with biopharma companies today about how we might integrate a tool like LOWE and our teams at Recursion with them.
I think it’s too early to talk about the significance of these lines of product revenue. I don’t think it’s too early to talk about how Recursion leading the field with tools like LOWE is helping pull the industry forward, partnering with extraordinary companies like Roche-Genentech and Bayer to help move the entire industry forward. And I think over time, whether it’s through the software offerings themselves or whether it’s through new chemical entities that we discover with our partners or in our own pipeline, I think we’re going to drive a tremendous amount of product revenue leveraging these tools. All right. Next up, we have a question from Kareem Harrison [ph] who asks, when will the company be profitable? Well, that’s a great question.
I think we see the opportunity before us as a multi-trillion-dollar opportunity with profound potential impact for patients. There are few industries today where despite hundreds of thousands of really incredible scientists working really, really hard, on average, our industry still fails 90% of the time in the clinic. And what’s more, I think, there are roughly 20 or 25 biotech and biopharma companies today with market caps above $100 billion. That kind of lack of condensation of these companies, I think, is pretty unique to biopharma. And so we believe that if you look out 10 years to 20 years, there will be a much smaller number of biopharma companies and those companies will look much more like Recursion does today than they will look like a traditional biopharma company, and we hope and expect to be one of those.
And what that means is that we’re going to lean into growth in the coming years. We’re going to be good stewards of our capital, but we’re going to lean into growth. And so because we see the magnitude of that opportunity, while we hope to decrease with the upcoming milestones and revenue, we hope to decrease the losses on a quarter-by-quarter basis in the intermediate term. In the long-term, I don’t think we’re going to lean into maximizing profitability because we think there’s a multi-trillion-dollar prize and impact for hundreds of millions or billions of patients on the line over the coming decades. All right. Next up, we’ve got a question from Juan Fernandez [ph], who asks, what is the company vision? What daily actions are being taken to achieve it?
Juan, this is a great question. We believe that biology and chemistry are deterministic, that with the right data and the right technology tools, we will be able one day to predict how any biological and chemical interaction operate, not only in human cells, but in the human organism and beyond the human organism in any living organism. And our vision is to be the company that digitizes this space, that moves from wet lab one day entirely to dry lab, where our experiments are done only to validate the predictions we make at scale. And if we can achieve that vision, I think we have the potential to be one of the most impactful companies in the world. And so how do we manifest this every day? Well, we have a Recursion mindset that we teach our team.
We have events like Decoding Recursion. I just got back from one last week where we bring new and tenured employees together for a couple of days to talk about how we can focus on the experiment we are here to run. We don’t want to play the game everybody else is playing because we know what the probable outcome is. We want to play a different game. We want to test this idea that there could be a different way to discover and develop medicines and so we push that into every person at Recursion. We even push that into our partnerships, pushing our partners to adopt new tools, to adopt our workflows. And so it’s very front and center at Recursion and we lean into that vision every day. We still have all hands every week at Recursion. I’m still a presenter at all hands as often as I can be.
