Welcome, everyone. This is the Fireside Chat with Schrödinger. Thanks for joining us. My name is Vikram Purohit. I'm one of the biotech analysts with the research team here. Happy to have with me, Ramy Farid, CEO, Geoffrey Porges, CFO from Schrödinger. Before we get started, I need to read a brief disclosure statement. For important disclosures, please see the Morgan Stanley Research Disclosure website at www.morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales representative. With that, let's go ahead and get started. We have roughly 30 minutes, a good amount of material to cover.
Yeah.
So, Ramy, Geoff, maybe we could just start with some brief opening remarks from your side on what you think some of the, the key inflection points have been for the business this year. And just for those who may not be fully acquainted with the business, just a quick sense of how the business is currently set up.
Sure, sure. Happy to. So the business is set up to be able to leverage sort of an extraordinary platform that we've been developing over 33 years. The company was founded in 1990, and this platform has allowed us to compute properties of molecules with very high accuracy. So we have this platform, been developing it over a long period of time, and we leverage it in several different ways. One way is to actually make that software available to the industry, and by industry, I mean the drug discovery industry, so pharma companies, biotech companies, worldwide, everywhere. We have sales offices all over the world.
Also, to material science companies doing all sorts of different things that involve designing novel molecules, and where it's really beneficial to predict the properties of molecules before you go randomly making a molecule and spending all that time and money doing things by trial and error, just using experiment. The other way that we leverage that platform is to help other companies to advance their drug discovery programs or material science programs. So where we are using our software internally and helping to advance those programs. In those types of collaborations, the partner has a lot to do about, for example, in drug discovery, picking the targets, you know, making decisions about when to take the program, you know, to development candidate stage, when to take it in the clinic.
We're playing a big role, and in fact, many of these companies we've co-founded, so we were involved in the formation of those companies. That business also generates revenue in the form of upfront payments, in the form of preclinical milestones, clinical milestones, royalties on sales, right? You know, that's interesting. Another way of monetizing the platform. More recently, we've started to work on our own programs that we have retained all the value and all the ownership, so they're wholly owned programs. Some of them are partnered, but many of them are wholly owned. Of course, there we're gener... You know, we're selecting the targets, advancing the programs, taking them to development candidate, taking them into IND-enabling studies, and even more recently, taking them into Phase I clinical trials.
We have one program that's there now, another one that will be there soon, and a third one where we plan to submit an IND, you know, soon. So that, I hope that helps-
Sure.
Understand. And I'll just say one last thing, just a few seconds.
Sure.
It's very important, as we're having this discussion, to remember, these are not three separate businesses. There are tremendous synergies between these businesses. We're always looking for new ways of leveraging those synergies, and so it's, it's a very deliberate thing that these are all in the same company and benefiting from each other and sort of growing, growing together.
Mm-hmm.
Vikram, you asked about this year?
Oh, yeah. Good, thanks.
We started the year with a major validation event, which was the distribution from Nimbus after the sale of the TYK2 to Takeda. It was really nice to get $147 million, and that left us in a really well capitalized position. The second thing that we've signaled is that we are seeing opportunities for really substantial growth in the software business, despite some uncertainties in sort of biotech funding environment that we're all aware of. Our largest customers in particular are having conversations with us about significant step-ups in the deployment of our technology into their drug discovery organizations, and those discussions are continuing. We are advancing our collaborations.
As I signaled, and we've talked about in many calls, we are transitioning our R&D expense and our internal teams from the collaboration programs to proprietary programs, and that's what increasingly we're going to start seeing data from. We're a clinical company this year for the first time, which is fairly remarkable. We now have two programs that are in Phase I. We have two clinical trials with MALT1, one trial that we're just opening up now for our CDC7. So the company is definitely transitioning. We're not walking away in any way from the software business. In fact, we see lots of opportunities there, but we're definitely adding on and investing in the proprietary medicines as well.
Got it. Got it. That's helpful. There are two kind of broader macro issues that I wanted to get your perspective on before we discuss the software business and the drug pipeline, because I feel like, having a perspective on that set up for what's going on more broadly is going to be additive to that discussion later on.
Mm-hmm. Yeah.
So I think one of those questions strikes me like one for Geoff, one strikes me like one for you, maybe, Ramy.
Okay.
But, on the topic of biotech funding-
Yeah.
And both of you feel free, feel free to chime in, but, what has been the impact or, or what is, generally speaking, the impact of biotech funding fluctuations on Schrödinger's top line? And how can you best try to structure the business to insulate from these kinds of cycles, if that's possible?
Yeah. If you look back at the growth rate of the software business, I think last year we grew at 21%. The year before, I think it was 30% or so.
Mm-hmm. Mm-hmm.
And the year before that, it was close to 40%. There's no doubt that in 2020 and 2021, there was some extra growth contributed by emerging companies, newly funded, saying, "We're going to do drug discovery against target X or against class of molecules Y, or in some disease," and they open up their drug discovery efforts by signing a contract with us. So new companies emerged in those years that definitely contributed. Now, if we look at the number of new companies showing up, it's gone substantially down. But the companies that took on our technology then, by and large, have stuck around. Now, not all of them. There's been some attrition, there's been some consolidation, there's been some acquisitions. Small companies have sort of gone into big companies or shut down discovery. So there's been some small amount of attrition there. But the biggest change is that we're not seeing the new, new customers showing up.
But another interesting thing, the companies that seem to have made it, and I don't know off the top of my head when they went public, whether they were in the 2021 cohort or slightly before that, but I'd say, you know, Morphic is a good example of that. The companies that sort of got out and into the clinic seem to continue to be well enough capitalized to not only afford our software, but in many cases, to be increasing their adoption of our software. So, amongst our top 20 customers, there's definitely biotech companies in there, and it's remarkable that we have biotech companies who are using, on an ongoing basis, more software than some of the largest pharmaceutical companies in the world. Which gets to the question that we're always asked about: What's the opportunity?
The opportunity is for those big pharma companies to step up. So, that's kind of what's going on with the funding environment. It's not that people are dropping off, it's not that we're losing a lot of customers, but it's the absence of the new customers that's causing our growth rate to be lower than it was in the past.
Understood. That's, that's helpful perspective. The second question then is related to all the recent focus on AI and ML across every sector, including drug development.
Mm-hmm.
I think it would just be helpful to get your perspective on, you know, as people think about companies like Schrödinger and other AI-enabled biotech players, for lack of a better word, what are some key nuances to keep in mind when people think about what's really possible?
Yeah.
-versus where expectations might be on any given day? And where do you feel like AI, ML, tech-enabled drug development is truly validated and has a true use versus where it's still aspirational?
Yeah. So I think the most important thing, to start the conversation, is to remember that AI, ML is actually a technology. It's, it's an algorithm. It has, like every technology, it has a domain of applicability, it has things it's really good at, and it has things it doesn't. I think the problem is, with LLMs in particular, when that came out, that to a lot of us, including me, by the way, you know, it doesn't matter if you're an expert or not, when you start to use these technologies, they start to have, they take on characteristics of being a bit magical, right? It's sort of like you can't even imagine, how the hell is this working? And you have to realize, by the way, in that case, it's just trained on a massive amount of data. So yeah, it's doing what it's supposed to do.
But it's an actual technology. You can understand it. It may take a little extra time, but that's what it is. Again, what that means is it has a domain of applicability. There are certain things it's going to work in really well, and there are actually certain things that it won't. So what that means and what that there's a way of describing that now in very general terms. In cases where you want to predict something that looks very similar to things you already know, it works really well. That's what machine learning is. In cases where you want to learn about something completely or make a prediction about something that's completely novel, for example, that's not in the training set of these systems, it will not work.
It doesn't matter all the sophisticated words you would assign to it, generative, deep learning, whatever the fancy word of the era, you know, of the time is, it still is. That's what I mean by it's a technology. Don't let yourself be fooled by, you know, the kind of magical talk around it. Okay, so now where is it having an impact? It's having a huge impact at Schrödinger. We have quite a number of programs that we've been working on. A large number of them are in the clinic. I mean, these are advancing. We have a really fantastic track record.
Well, it turns out that we have been leveraging physics-based methods, that's the other type of method where you're using first principles, nothing to do with machine learning, with machine learning, in a way that actually leverages the advantages of machine learning. Machine learning has advantages and disadvantages. The disadvantage we're just talking about, it needs a training set. Where is that training set gonna come from? It can actually come from ab initio physics-based methods, because you can't produce large training sets by just doing a lot of experiments. That would take hundreds of years. That's not practical. But you can do hundreds of years' worth of experiments in a day using these physics-based methods that we produce. So there's an example where it really is having an impact.
It's allowing us to explore huge amounts of chemical space by understanding the technology and leveraging the advantages that each different technology has. The companies that are, and there are a lot of them, hundreds of them, that are claiming they have a black box solution they can't tell you much about, they can't tell you where the data comes from, they can't tell you about the algorithm, they have no track record, there's nothing they... You know, I think it's okay to be skeptical of those, even though the words sound really fancy when they talk about it. It's okay. Your instinct is correct. It's probably nothing there. Yeah. Now, this is everything I just said is in chemistry, okay? I'm talking about chemistry. There are many other aspects of drug discovery.
There's drug development, there's patient selection, there's things on the other side, biology, path, understanding pathways. These are areas that actually turn out to be quite amenable to machine learning, to things like LLMs. That's not a space we're in. Since we understand the technology, we can say a little bit about it. There's some very interesting things going there, but use the same principles. Like, if it sounds too good to be true, it probably is. If it's a black box, and they're not gonna tell you where their data is coming from to train it, again, it's probably a good idea to be skeptical. But there's some really, really interesting things happening outside of what we're working on.
Again, in the biology and understanding that, and reading literature, and doing what a biologist does, and of course, in informatics, you know, which is really quite amenable to machine learning because there's a lot of data. That's what it's always about. It's about how much data you have and if it's good data. I hope that helps.
Very helpful.
Yeah.
I think that's a good, good segue into talking about the software business.
Yeah.
So, with your last earnings update, you had increased your software guidance.
Mm-hmm.
That was based on visibility you noted having into some potentially large contracts coming up. And I think, this has been a topic of focus for a while now. How much visibility do you truly have into your business backlog? So could you just kind of tease that apart for us?
Yeah. Maybe I'll say a few things and hand over Geoff-
Sure
Both of us have been thinking a lot about this. So first of all, it's important to remember that we have a very, very high customer retention rate. I mean, it's essentially 100%. The few customers that don't renew are ones that stop doing research or get acquired by another company. So they're very, very large, I mean, essentially 100%. So that already provides some level of transparency in knowing they're gonna have to renew. They become dependent on the software. So it's not all this uncertainty: Gee, I wonder if they're gonna continue to use the software.
The other thing that we're really encouraged by is that we have noticed that when we start having conversations with heads of research, right, that people who are actually really making these decisions and are really thinking about the efficiency of the company versus sort of maybe, you know, if you go too far down, you know, end users just trying to get through their day, and they're more focused on, does this software easy to use or not? You know, things like that. Anyway, when you're talking to the heads of research, and they're saying: Hey, we're seeing the success of the TYK2 program and the alpha-4 beta-7 program from Nimbus and Morphic and so on, and the GLP-1 program from Structure, we want that. That's important to get in turn.
And then they're saying: "Okay, not only do we want that, but we're, we're prepared to have you train us because we, we provide a lot of training. It's something we've really invested a lot in. That's a pretty good... That's starting to feel pretty good, right? If they're coming to us, training, they understand that the technology is having an impact, and they're just trying to figure out how to get over all the inter-internal barriers, IT barriers, knowledge barriers, that's, that's a - that starts to give us some comfort that, okay, this is a customer that's saying, 'All right, we, we need to be using this technology at scale. We, we need to be using it the way Schrödinger is using it.' That's where the visibility comes in.
Yeah, and as you can imagine, Vikram, we have a database of every one of our customers, and we know when their contract's up for renewal, and they will know if they haven't renewed it because it'll get switched off, and then they won't be able to be testing any molecules. So, you know, we have a commercial organization that's in conversations with those customers, at least the ones who are spending more than a few thousand dollar a year. Your contract's coming up in September or in December, and we start the discussion, and then they say, "Okay, this is how much we've been using it." And typically, there's a step up in that contract, but more importantly, they're certainly about renewing it.
As we've said in the past, a number of them, the largest companies, elect to have multi-year contracts with us. They... when they're building a business process-
Right
a fundamental business process and an organization around using a particular technology, they need to know that's locked in. So they say, "Okay, we want guaranteed pricing etc., for three years or two years." As you know, the revenue recognition for that forces us to recognize that revenue upfront, but we can see that renewal coming years in advance. And short of the company sort of being acquired and shutting down drug discovery, it's pretty inconceivable that they will engineer a business process and organization and then stop using the technology. So we have a lot of visibility to those renewals of those multi-year contracts in particular, but of all of our contracts. What we don't have certainty about or even, you know, really tight confidence around, is how much they'll step up.
Right.
The trend across our business is that the deployment of our technology is scaling up as the customers get more and more experience with it and gain more and more utility from it.
Got it. Got it. Okay, that makes sense.
Okay.
Building on something you just said, Ramy-
Mm-hmm.
when you're speaking to heads of research about their decision-making around the use of Schrödinger software, also tying in something you said previously about the TAM for Schrödinger software being much bigger than what it is currently, right? And that all implies that there are use cases within biopharma that are underutilized, where Schrödinger software services could be plugged in, where they're currently not plugged in.
Yeah.
What do you think are a couple of, like, good tangible use cases where it's just not being used as frequently as it should be?
Yeah. Okay, here's a good example. We would estimate that of all the programs that pharma's working on, about 10%-20%, depending on the company, are what's called structurally enabled. In other words, they know the structure of the target. They know the identity of the target, they know the structure, they have a lot of details about how the molecule binds. They got that from experimental methods. They went, they got lucky, they solved the structure pretty quickly using X-ray crystallography. That's good. That's great input for our software, and they get great results with those. Okay, what about the other 90%, 80%, 90%? Well, it turns out there are ways of actually enabling those targets, but it requires a lot of work. You may have to go out and get a cryo-EM structure.
You may have to go get an AlphaFold structure and then refine it to higher resolution using our platform. It's, it's a lot of work. It's pretty complicated. Protein structures are complicated. Every time a molecule binds to a protein, the protein adopts a new conformation. These, these are complicated, and you have to understand not just the structure of the atoms and all the atoms in the protein, you have water molecules associated. I mean, this is an enormously complicated, flexible thing, but it takes a lot of work and a lot of licenses of the software, a lot of technology and a lot, a lot, a lot of knowledge to enable a target.
But if you're motivated to do it because you know that if you have a structure, you're going to be able to leverage our platform at scale, and that's going to have a huge impact. So I think this is an area where the work that it takes to structurally enable programs, even picking the target. So you will find in pharma, this is all segregated. You know, biologists pick the targets, hand it over to the chemists, and the chemists say: "Wait, there's no structure." They say, "Too bad. We need a..." But you know, that doesn't make sense. So at Schrödinger, these teams are all working really closely together, and we're saying, okay, what is going to be the path to identifying a great target, but getting a structure and doing all the work that's required?
And that has a profound impact on our business because once, of course, customers are enabling more targets, right, now they're, the demand for the software really increases substantially. So this is an area where we're focused on. They're not, this is an area we're not utilizing. The other, really quickly, I know we're running out of time, is they don't have enough licenses to actually explore enough chemical space. I'll just say this, drug discovery... I'm going to say some profound statement that you may not have ever heard before. Drug discovery is really hard. So it turns out, really hard. And what does that mean? It means you have to test huge numbers of molecules to find that perfect molecule that some, by some miracle, right, is potent enough and selective enough and soluble enough and permeable enough and doesn't hit anything else.
I mean, that's really hard. The way around that, obviously, is to explore a huge number of molecules. That requires a lot of licenses and a lot of commitment to using the technology at scale. They're not doing that either. A few companies are. That's the exciting thing. They're starting to get there. They're doing it on a few programs, and we're pretty sure that once one company, two companies, three are doing that, of course, they all will do that, but it's just going to take time. That's another really great example of, yeah, of I think what you're asking.
Got it. Got it. That's helpful.
Yeah.
Yeah, we have around seven or eight minutes left. Maybe this is a good, good, time to pivot to another line item in the business, drug discovery and partnership revenues.
Mm-hmm.
There's been a good amount of focus on that throughout the year.
Mm.
My first question for you there, Geoff, just to level set for all of us. Just walk us through exactly how that line is estimated, because there's obviously a good amount of gray area in when milestones could be achieved and when they could come through. So just kind of unpack that for us first.
Yeah, no, it's complicated. So the revenue that runs through what we call drug discovery is sort of a combination of different things. It's our portion in that quarter of the revenue from upfront payments that we are eligible to recognize in association with completion of work on that program, and that's based on an estimate of the time required to complete all of the work in the program, and that revenue is sort of spread across that period. So if that time changes, say, it goes from four quarters to six quarters, then the amount of revenue that you get to recognize goes down by roughly a third in each quarter. So the first thing is we have to recognize the revenue from the upfront payments.
Then we have to try and pick when the programs that are still in our portfolio, in the collaboration, are going to get to a certain milestone that will trigger a payment. So that might be getting to DC or LO or somewhere along the discovery pathway. And then we have to assume we will meet the criteria of the partner, so we have to estimate the likelihood that they will say, "Yes, check, looks good," and that that will then trigger a payment. Then there's a third component, which is the ships that we've set forth on the ocean that are in our partners' portfolios in active development, we have to estimate when they will hit some sort of milestone or kind of trigger that will trigger a milestone payment.
So maybe enter clinical trials or enter a Phase 2 or complete a Phase 2, depending upon the agreement. And in many cases, those are with third-party companies, meaning we did a collaboration with company X, and they've then partnered the program with company Y. So we have to try and estimate when company Y will reach that endpoint. So there is an increasing amount of uncertainty, the less it is in our control in those estimates. And the last piece that makes it particularly problematic is that the milestones get bigger and bigger, the less control we have, sadly, because as the programs advance into the clinic, as you know, industry norm is that the milestones get larger and larger, $25 million, $50 million, sometimes even $100 million, down the track.
We don't have much control over that at all because that's in the hands of the ultimate licensing company. So all of those factors contribute to the uncertainty about estimating the drug discovery revenue.
Got it. Okay, that's helpful. And building off that then, you can talk a little bit maybe about your partnerships and recent progress there. The BMS, really, you have partnerships with both of these companies.
Mm-hmm.
What would you cite as some of the most recent kind of concrete developments? To the extent you can talk about them. Looking forward, what do you think is, like, realistic in terms of the scope of disclosure you might be able to make as these things move forward?
Yeah, so, well, certainly a highlight for the year is the SOS1 program reaching DC and then transitioning into BMS's development portfolio. We understand that's going ahead pretty actively, but again, we don't get any more information now about that than you do. But that's pretty exciting. That's the first of the six programs in the BMS collaboration that triggered a $25 million milestone payment in Q1 that we disclosed. And we're continuing to work very actively on the rest of that portfolio. So, it was nice to see that collaboration that's, you know, three or four years old now, reaching that kind of milestone, particularly given the investment that we've made and all the efforts that we've put in. The Lilly collaboration is early. We only announced that at the end of last year.
It's pretty, I would say, high profile, high value target, and it's going really well. You know, we think there are opportunities that we could do more collaborations with not just Lilly but even with other companies. What we're continuing to say to them and saying to our investors is that we do have a limited bandwidth for taking on new collaborations. That you know, as we're better and better capitalized and more and more mature as a company, you know, we're shifting our resources towards our proprietary programs where we retain 100% of the value. So why should we be discovering really great molecules for a partner, where we're only retaining 10% of the value? So there's definitely that transition. We do believe in the collaborations. We're committed to them.
We think that we'll continue to have collaborations, but they're not gonna sort of really ramp up. We kind of think that there's gonna be a steady level of collaborations going forward. Maybe we'll add some, maybe some will fade away, but that's, you know, more and more we'll be doing our proprietary programs. Is that clear, Ramy?
Yeah, perfect. Exactly.
Got it. So, since you mentioned that there's kind of a cap on what you can do in terms of partnership capacity, how are you filtering what's interesting, what's additive to Schrödinger, what's worth pursuing, what's not?
Yeah. That's evolving. You know, in the early days, we didn't have as many capabilities. We didn't have a structural biology lab. We had one biologist in the group. We didn't have CMC people, you know, and so on. But those capabilities are evolving. What we're looking for, and I think we have found it in a few partners, is when a partner brings something to the table, right, that really is difficult for us to replicate. I mean, I think Morphic is such a good example. I mean, this is a company that is world-class understanding of integrin structure, you know, from Tim Springer's lab, and not a company has that. I mean, it's absolutely incredible, right?
Their ability to understand not just, just the structure, but the structure, function, relationship, and that's really, really hard. That's a hard thing to replicate inside Schrödinger. That's an example where it makes sense to form a partnership. A GPCR company, Structure Therapeutics, perfect example, right? GPCR Structure is getting more and more routine to get structures, but still really complicated. Understanding the biology, understanding the structures of these of these targets is still quite complicated. And so in examples like that, where there's a real complementary, you know, complementarity between the technologies. But again, I think it's kind of exciting that that's evolving and we're gaining more and more capability.
So I think there are more and more programs that we can take on from the beginning, you know, from the target selection all the way to IND and even in, into the clinic. So you'll... That'll continue to change. But we think there's still, there's still very smart people out there that are doing some exciting things and have been working on problems for 20, 30 years, that's a little hard to replicate, you know, in a, you know?
Sure. Understood. We are just about out of time. I'll ask you a final question. We didn't get a chance to talk much about the proprietary pipeline-
Mm-hmm.
But just to close out, maybe you can just recap some of the key-
Yeah
... data points we should be looking forward to in the next 6-12 months from your pipeline programs?
Sure.
Yeah.
So we have two programs in the clinic, MALT1 and CDC7. They're both in the hematologic malignancy indications, both in Phase 1 studies. The MALT1, we're actively enrolling both the healthy volunteer study, to get a lot more PK/PD safety tolerability data, and also, the patient study. We've expanded the study sites in Europe and B-cell malignancies. We think that we'll have some data, at least from the healthy volunteer study, sometime this year. And that should be really interesting, particularly given the context of some data from competitors already out there. The CDC7 trial, we're just opening study sites now, but there's a lot of enthusiasm for that trial. It's a tough indication. AML, we all know, is very challenging to develop novel medicines in, but there's plenty of unmet need there.
Then we're on track for filing the IND for the Wee1 next year. So, a lot going on there. We will have a pipeline day at the very end of the year in December, after we get more data from the MALT1, and hopefully then we'll share some other things in the pipeline that we haven't disclosed before.
Great. Let's close out with that. Geoff, Ramy, thanks so much for joining us.
Thank you.
Appreciate it.