I'll try and get started on time here. Great. Thanks, everybody, for joining us here. Second day of Piper Sandler's Annual Healthcare Conference. I'm Joe Catanzaro, I'm one of the Piper Biotech Analysts. It's my real pleasure to welcome Schrödinger. Joining us is their CFO, Geoff Porges. A lot to get through in 25 minutes, but I know, Geoff, you wanted to sort of, sort of intro, give a little overview what Schrödinger's been up to and what we have to look forward to, and then we'll jump into Q&A.
Great. Thanks, Joe. Nice to see you all. Just for those of you who aren't familiar with Schrödinger, it's a company that's developing advanced computational methods for discovery of novel materials and pharmaceutical products. This year we've guided to revenue growth for our software business, which is the company's primary business and original platform. That revenue growth is 15%-18%. We're looking for, and we've got a lot of confidence in the outlook for the year. Last year, we did $135 million in software revenue. We've got close to 2,000 customers globally for the software. It's very sticky and becoming an essential tool for the discovery of novel materials and pharmaceuticals.
We are also building a proprietary portfolio of our own programs that are in or on the cusp of going into the clinic, and we're increasingly pivoting our capital allocation towards those proprietary molecules. We're quite excited about the first clinical data for our first internally developed and advanced program that we should have by the end of the year, a Pipeline day that's coming up in December, December fourteenth. That should build on some data that we disclose at ASH and give a real overview of the disclosed programs and the opportunity we see for them, but also some of the earlier programs that we haven't disclosed in the past. Stay tuned for that.
Next year, we're looking at starting to see some of the data coming out for those molecules, hopefully, but also continue to advance the software business and grow our top line there. That's kind of the outlook for the company.
Perfect. So, you sort of touched on it, but I wanna see if you could elaborate sort of the company's concerted effort to shift more resources towards your internal proprietary pipeline. What were sort of the main drivers of that decision?
Yeah. So I think well, the first is that we do have the capital to do that now. After the IPO and follow-on, and then the distribution from Nimbus, we're pretty well capitalized, and we've concluded that, A, we'd rather retain 100% of the value of a program than 10% or 15% of the value of a program, which is what we have when we're involved in a collaboration. The second is we'd like to be in control of our own destiny, and like many companies in the biopharma industry, we've been sort of swung around by changes in, you know, strategic prioritization and management and perspective in some of our partners this year.
So if we retain the programs and take them further forward ourselves rather than partnering them, we kind of mitigate that. So, I think it's driven by confidence in the platform and the programs, by retaining more of the value and then also being in control of our destiny.
So how does this decision affect the other parts of your business? I'm mainly thinking about your drug discovery collaborations. Is your sort of bandwidth to do some of those deals a little bit more narrow now, or is it sort of still the same?
Yeah, we certainly have the resources to do collaborations, and we continue to be in discussions with potential collaboration partners. The way those discussions play out is often we have customers. Well, everybody in the industry is a customer for our software, but when those customers say, "We'd like to scale up," frequently there's a discussion about, "Can we have a collaboration to see how you use the technology up close, as well as increasing our software purchase?" So we are seeing the opportunities for collaborations to occur in the context of an increased software purchase. So we certainly wanna have the capability to support that. It's consistent with growing our software.
But what we're realizing is that the way we add value in the collaborations is at a very specific point in the discovery of the novel molecule. We don't really add a huge amount of value in terms of, for example, pre-IND studies or, you know, CMC or some of those things. It's really around the point of where we design the molecule and validate those design hypotheses, and we think that we can narrow down our collaborations, still leverage that competitive advantage and that capability, still generate value, but not have to take on the full burden of doing all of that preclinical development.
So, we continue to have discussions, but we're trying to focus more on what we call relatively skinny collaborations, where the economics is still good for us, but we don't have as much scope of involvement.
It sounded like you touched on this idea of, you know, partners who are using your software, then coming to you and saying, "Hey, maybe we could do some drug discovery collaboration." Maybe it speaks to sort of the synergies across the business.
Yeah.
Where does the synergy come from when we think about sort of the software and the proprietary pipeline and the, and the drug discovery pipeline, collaborations? Like, so where does the overlap lie?
Yeah, there's a lot of synergy in that area of deploying the technology at scale to discover really novel compounds. And we find that putting the software out into the industry gives us a huge amount of feedback on the utility of the software, and we don't see the programs that people are deploying the software against, but we do see some of the challenges they run up against and what we need to do to make the software even more useful. So we get that feedback from those customers. We build those capabilities into the platform. For example, now we're in the process of identifying a lot of proteins that are associated with off-target tox, and building the software capabilities to predict those off-target tox liabilities of molecules.
So that's, that's a good example, and there are a bunch of other attributes that we're building as well. So we get that feedback, improve the capabilities of the software, effectively beta test that in our collaborations, where our own drug discovery group is working with some of our software customers on a specific project. Does this work? Is this adding value? Is this helping accelerate drug development? Then we see how that's working, and then we say, "Okay, there's some low-hanging fruit in terms of us deploying this ourselves against particular targets," and it enables us to accelerate our own programs. Those capabilities are then fed back out into the market through our software organization, and made available to our customers, but there's definitely a lag.
It's usually 12-18 months before our software, our commercial software customers get full access to a capability in the software that we are deploying internally, just 'cause it takes a while to really industrialize it, validate it, and, you know, make sure it doesn't break.
So I wanna maybe shift, specifically to the software business and ask a bunch of sort of what some key drivers are there. But first, I wanted to ask maybe a high-level question. I think Ramy has been very vocal about this AI/ML-
Yeah.
-and how-
Yeah.
How you guys are leveraging it versus how maybe everybody else thinks about AI/ML within the context of drug discovery. So maybe you could help us sort of understand that, that difference?
Yeah. We have been pretty outspoken that AI is only as useful as the training set that creates the machine learning algorithms that underpin the AI, right? If you have a training set that is narrow, then that's going to be the limits of what the AI can tell you to do, and if the training set is very, very broad, then maybe you can have an AI that's very, very broad.
Our contention, and I think it's still valid, is that there is no universal pharmaceutical discovery training set, that the representations of the universe of possible chemical compounds that we have tested is such a small fraction of the 10 to the 60th or so of different possible chemical compounds, that if you, if you believe in the universal pharmaceutical training set and you apply that, you will come up with molecules that look like existing molecules. Because the training set consists of the existing molecules, and this is what we see, because we see, and, you know, when we're involved in both collaborations and proprietary programs, we look at everybody's molecules, and we mine the patents and, and, you know, what's been disclosed. And, and we see molecules that are... They're, they're just sort of related family members, if you like, that are very, very similar.
So they could have been discovered by traditional medicinal chemists who say, "Okay, what if we just tweak this side chain in a few different ways and come up with an alternative?" Or could be discovered by machine learning, which, of course, is just accelerating what human beings do, right? So you... AI is just sort of massive scale of what, what humans do. So, if you want to discover really novel molecules, you have to go outside the existing training set. So the limitation of what everybody is using AI for, in our opinion, is that it is inevitably going to come up with me-too molecules that are not really differentiated from the existing molecules. And because we start with such an expansive universe of the entirety of chemical space, we...
Apply physical, physics-based methods to that chemical space, we come up with very divergent scaffolds that are different to the existing pharmaceutical compounds against the targets, and then we explore them and refine them, and so the molecules that we have in development are quite, quite different. So I guess to summarize, we don't believe that there is a useful, universal, pharmaceutical industry-wide, pharmaceutical, you know, molecular landscape-wide training set. We do think that there is utility in accelerating the discovery of novel molecules, but the way we use it is it's confined to a specific target.
So using physics-based methods, we can model all the possible scaffolds against a particular target, create a training set from that to optimize the selection of molecules that we're going to advance against that target, but then we throw it out. We don't believe that that teaches us anything about the molecules that you might develop against a totally different target.
Perfect. So, so maybe with that, we could shift to the software business and maybe work backwards and think about sort of medium, longer term drivers of, of continued software growth, and maybe to something you mentioned earlier, like how important is new product offerings to the growth trajectory in the software business, looking forward?
Yeah, to answer your specific question, it's actually not that important right now. The most important driver is scaling up, and we are seeing this in our large customers. We're actually even seeing it in smaller customers who are realizing that if we're going to get the utility of this technology, we have to deploy it at substantial scale. In some cases, almost the scale that we are deploying it internally at Schrödinger. And we've said again and again that we're deploying at an order of magnitude above our largest customer. So our biggest big pharma customer is still 1/10 of the level of our deployment of the technology internally.
And we think that we're deploying that technology against, let's just say, hypothetically, 20 or 30 programs, which is still a fraction of the programs that a big pharma company would deploy against. So we are continuing to have discussions about that scale-up, which is the big driver. Now, down the track, five years from now, I do think that some of those predicting off-target tox, predicting a host of other physical attributes of molecules, is going to be really important. But right now, it's about scale.
Yeah. Maybe along the lines of this scaling up, what's the biggest leading indicator for you guys that a company or partner is tracking towards scaling up their use of your software?
Well, the biggest leading indicator for us is, look, this is a. Once you get to north of $5 million, it's a high-level B2B selling exercise. So it's something. It's a discussion that virtually goes on continuously, and, you know, we will have, you know, months and months in advance, we'll be talking about, okay, how's the software being implemented? What's the experience of your users? What issues are you encountering? And we'll know how far they are up in terms of the scale of their deployment, and we'll know when they start to hit the limits of the number of licenses that they have. And so we have a lot of visibility to seeing what that scale-up is gonna look like and what the timing of it is.
Smaller customers, we could get a purchase order, you know, for someone saying they want to go from $500,000-$700,000 worth of software a year. That may be more of a surprise 'cause we're not sort of in there all the time. But the big customers, we have a lot of visibility to it.
So, yeah, I guess when you look out at your partners, big and small, like, where is the greatest area of underutilization of your software? And I think you guys have said that maybe 10%-20% of drug discovery is structure enabled which I think is key to using, leveraging your software. Like, does that need to continue to, to increase for, for the software business to also continue to grow?
Yeah. So first, first answer is, the greatest area of opportunity is all the large pharma companies and mid-sized pharma companies that are dabbling with the software, but not fully invested in the software. And, the, There is definitely not a tight correlation between dollars of drug discovery spend and dollars of Schrödinger spend. It's all over the place. The leading external indicator I would look to predict where we're going to sell a lot more software is changes in R&D leadership.
So when a company makes the transition from, I shouldn't say this, old-school pharma R&D head, who grew up with an army of medicinal chemists, to new-age pharmaceutical R&D head, who believes in the use of computation and digital technologies, when that transition occurs, we find that a very positive, kind of, leading indicator for us. So does that answer your question?
Yeah.
Yeah. So let me talk about the protein. That was the other question. So, in terms of proteins, yes, the structural characterization of proteins is occurring very rapidly. AlphaFold has been a huge advance, but it's not enough, and even the sort of latest version of AlphaFold doesn't have the sort of super granular protein structures that we need to deploy our technology, but it is a really useful starting point, for deploying our technology. Now, we've invested in structural biology capability, but every pharma company is as well.
So there are lots and lots of new protein, you know, actual protein sequences and atom-level structures being published, but equally inside pharma and even biotech companies, those protein structures are being identified, and people are coming to us and saying, "Okay, can we use computation to come up with a molecule?" So that 10%-15%, you're right, has been sort of gate limiting, rate limiting, but it is rapidly accelerating.
I think one thing you're often asked about is sort of how the biotech funding market has-
Yep
... has impacted your, your business. So I'm wondering if you could speak to that and whether we're at the point where that sort of dynamic has, has sort of stabilized and maybe the impact or not on, on Schrödinger moving forward.
Yeah. Certainly, you know, in terms of year-over-year comps, that's no longer really a big issue for us. We are not seeing a huge amount of flux in our customer base now, meaning new customers coming in or customers going out. But if we go back two or three years, there was a significant contribution to growth from new customers, meaning biotech companies that had been funded and showed up and said they wanted to discover drugs against a target. So putting grossly simple terms, we're not seeing the new Morphic, the new Structure kind of coming to us and saying, "We want to be a million-dollar software customer.".
That's no longer a growth headwind for us, but neither is it presenting itself as a growth opportunity right now, which is why we're sort of seeing the opportunity in the large companies who, you know, can go to much higher scale.
So, maybe now shifting to more near-term considerations around the software business, and specifically, I guess, thinking end of the year into 2024. At this point, how much visibility do you guys have into, you know, potential renewals, whether partners are going to increase their utilization? I would imagine at this point, you're at the point of, you know, there's a contract drawn up, and it's just a matter of maybe getting some signatures.
That is almost all circumstances that's the case, except that there are always surprises into the end of the year. We are closing software licenses, you know, December 30th, typically, and the finance organization is looking at contracts and making sure that we think about what the revenue recognition needs to be for a particular contract and the language and the clauses right up to the end of the year. Now, those very large purchase agreements, obviously we've been talking about them all the way through the year, so we have a high degree of confidence about how they're gonna land and what they're gonna look like, but it's the smaller contracts that still turn up and surprise us.
So, I wanna ask maybe some bigger picture questions on the software business. Maybe first, we've always sort of focused on small molecule efforts, whether there's opportunity for software offerings, whether currently or in the future, to leverage it within the context of biologics and biologic drug discovery, or are we sort of always gonna be sorta constrained to small molecule efforts?
No, look, we do have capabilities in our software stack that are useful for biologic discovery. For example, we can predict binding affinity for antibodies and for other sort of antibody-like structures pretty effectively. We can predict various attributes associated with pH and sort of duration of binding pretty effectively. One of our products is a thing called LiveDesign, which is an enterprise informatics platform, and that's proving to be very useful for biologics. So we are having more and more discussions about increased scale of software sale to both the biologics groups within big companies and then pure biologics companies themselves.
That being said, we are conscious of one limitation, which is that, you know, that there are 10 to the 60th possible structures for a small molecule against a particular target, in theory. There aren't 10 to the 60th possible structures for an antibody against a target. The possible range of structures for an antibody is already constrained by having to have a light chain and a heavy chain and, you know, binding epitopes and all those sort of things. So because the scope is so much narrower for a biologic, the utility that we confer in terms of coming up with a novel molecule is less. We can help, but we aren't actually going to deliver the specific molecule the way we do in a small molecule program.
My last question on this front, and it's, I'm admittedly naive to this, what's the competitive landscape look like within this sort of software, you know, physics-based drug discovery-
Yep
... offerings, and whether there are, you know, any competitors that can come and take market share from, from Schrödinger?
Yeah. There are competitors, and I've spent a reasonable amount of time since I've been at the company trying to figure out who they are and what they do. We can't find a competitor that has double-digit share individually, and when we do find competitors, they fulfill a small amount of the kinda real estate that we fulfill in terms of the drug discovery stack. So what's happened is there are competitors that do a little bit of one thing or a little bit of another thing, but none of them that have the breadth of our, our offering. And so what's happening is we bring all of these tools and integrate them and then put them in the enterprise informatics platform.
It becomes harder and harder for those companies that have this tool over here or this tool over there to be competitive. Then, when we look at what they're spending on R&D, our R&D investment in our computation platform, we've said, is close to 50% of our total R&D spend, so by implication, it's $70 million-$80 million. So our competitors are spending less than one-tenth of that. The difference, the gap between us and them, is only going to get wider. We, there are competitors. We pay attention to them, but they have small share, and they're underinvesting in the opportunity.
So in this last couple of minutes, and I know we'll be talking about this much more in the future, but I wanna touch on the proprietary- Mm-hmm. -pipeline. You mentioned the upcoming R&D Day coming up in December. What is your... What do you hope to be the take-home message?
Yep
... as you talk about sort of the proprietary discovery efforts?
Yeah, I think, look, we're really excited about our Pipeline Day, December fourteenth. I think first we'll be sharing our first clinical data. It's not gonna be, it's gonna be data from a healthy volunteer study, so, you know, great, we've treated some people who weren't sick. But it's a milestone for us nevertheless, and we do think that it's pretty important for that program to share that sort of safety and tolerability profile, given some of the limitations of other drugs against that same target. Secondly, we'll share some of the reasons that we're excited about both that program but also the other program that's in the clinic, the CDC7, and the other program that's approaching the clinic, which is the WEE1.
But we'll also share some of the other programs that we're advancing, some of which are in similar domains to those first three, but some which are in quite different domains. So what I hope people take away from it is, A, not just a preclinical company, but a clinical company. B, not just an oncology company, but a small molecule drug discovery company, with application in a variety of therapeutic areas. And C, a consistent ability to come up with innovative molecules with unique attributes. I think that'll fulfill that, it'd be great.
That's great, and I'm looking forward to it, and we're out of time. So, Geoff, I wanna thank you for your time and thoughts, and thanks, everybody, for joining. Take care, and enjoy the rest of your day.
Great. Thanks, Joe.
Thanks, Geoff.