Thank you. Is this working? Yep. Good. Yeah. I'm just going to present a few slides just to sort of kick us off, just sort of describe our platform and just high-level overview of the company. The goal of molecular discovery, drug discovery, and materials design is to predict the properties of molecules before you make them, and to predict them accurately, and to predict them on a really large scale. That's how you find the perfect molecule, is to explore lots of chemical space. How are we going to do that? How are we going to predict the properties of molecules really accurately? You've heard a lot about AI and physics, and I'm going to try and navigate that a little bit.
Over the last 35 years, we've been developing accurate methods using first principles, using physics for predicting the properties of molecules. It's remarkable how accurate it is. It's incredible. When you simulate atomic motion and get all the details right, all the physics right, you actually get the right answer. Kind of a nice thing. There is an issue, which is that the calculations take a long time. They're computationally expensive. It's hard to do them on a massive scale, on the scale that's really required to do drug discovery, on the scale of billions of molecules. What are we going to do? You hear a lot about machine learning and AI, and you hear a lot about how fast it is.
The problem is, it's not very accurate. The reason it's not accurate is because there aren't training sets that are available to actually train these models to produce accurate methods. By combining these two methods, we kind of get the best of both worlds. You can use physics to generate massive training sets for machine learning and really realize the power of AI. That is what we have done.
We have developed these physics-based methods that are really accurate, and we have combined them with machine learning to generate something that is really powerful, which is highly accurate methods for predicting the properties of molecules before you make them and doing it on a massive scale. How are we leveraging this platform at the bottom of the slide here? We obviously license the software to pharma companies, biotech companies, materials companies, all over the world. You see here we have quite a number of customers. We also establish collaborations, in some cases with biotech companies that we have co-founded, in some cases with pharma companies.
You can see there we have quite a number of collaborations in drug discovery and materials design. We also, more recently, launched our own proprietary programs. You can see there we have eight active programs, and a number of them are early in discovery, some are later in discovery, and of course, some of them are in the clinic. I'll tell you about those in a second. I think we're going to get some questions there, and we'll address those in more detail. This is meant to just be a high-level overview. The second to last slide here, we had a great fourth quarter. We just announced $80 million software revenue, 16% growth. You can see there the total revenue.
The full year was also a very good year. We exceeded expectations, 13.3% growth in the software business. You can see there a total revenue of $208 million. We guided to 10%-15% growth in the software business, and we guided to nearly doubling of the drug discovery revenue, $45 million-$50 million. We also guided to less than 5% OpEx growth. We're very excited. I think this is a reflection of the sort of success of our collaborations. We announced recently that we expanded our collaboration with Lilly, added another target, same with Otsuka. That's obviously a good indication of how well those collaborations are going.
Again, like I said, we're going to talk more about this, but we're very excited this year to be presenting for the first time clinical data on our three phase I studies. The first one that we're going to be reporting is in the second quarter of this year on SGR-1505. Really excited about that, obviously. We are really excited about the fact that we continue to advance the platform. We are really excited about the progress that we have made so far, but we continue to make really significant breakthroughs in the technology. Predictive toxicology is one of the things we have been talking a lot about.
A lot of programs fail because you may have the perfect molecule with regard to how well it binds to the target, and it is soluble and has good properties, but if it is promiscuous and it is binding to off-targets, that usually causes problems, and you generally discover those too late, often in the clinic. A lot of toxicity that is observed in the clinic is the result of binding to so-called off-targets. We have developed a physics-based method that leverages both physics and machine learning to predict the binding to large panels of off-targets. We are excited to be releasing that this year.
We think that's going to be not only generating revenue in the near term, but of course, helping to contribute in a really significant way to growth in the future. We have also introduced a biologics platform. Our platform is agnostic to modality, but one of the challenges in biologics research is dealing with massive amounts of data. We have built a solution for that. We continue to advance what is called our force fields, which are really the underlying physics of understanding molecular interactions. We call them machine -learned force fields.
This is basically a method that gives you the accuracy of quantum mechanics and the speed of molecular mechanics. Sorry if that's a little technical, but hopefully you get the impression that what we're doing is continuing to build solutions that are accurate but can be scaled to the kinds of scales that are required to do drug discovery and materials design. I'll just leave you with this slide.
The details of it may not be really important, but I just want to give you a sense that we continue to innovate in this space everywhere, from target validation to hit discovery to lead optimization to preclinical development, with a number of really exciting technologies that we think are going to contribute in a really significant way to growth in the software business, and of course, success in the collaborations and proprietary programs, not only this year, but going on into the future. Hopefully that's a good introduction, and we can get into the details. Yeah. Great.
Hear me. Okay. All right. There's obviously no shortage of content to get through today. We have a lot to cover, but I do want to keep it as interactive as possible. If there's any burning questions in your mind, do feel free to email me at Brendan.Smith@TDSecurities.com. Trying to make sure I'm seeing everybody here. Okay, this is great. This kind of sets the stage for a lot of the different aspects of the business that we want to dive into.
You touched on guidance, so maybe let's start with that heading into 2025. You're talking about the 10%-15% growth in the software business. Maybe let's just start there, and then we can kind of double-click into the pipeline and maybe the latter half. Maybe number one, what are customers responding to in particular with your software offerings? Maybe help us contextualize why that's the case increasingly this year, given everything that we're seeing in the evolution of demand for it.
Yeah. I think I'm coming out okay, right? Yep. I think what they're reacting to most, and I touched on this, is using the technology really, truly as a replacement for experiment. That's a really big deal. The industry has been trying to, has been looking for a solution like this for a long time. It's really frustrating to do drug discovery and materials design by trial and error, by just making a molecule and hoping that it has the properties that are required. It's a pretty amazing transformation now that there is enough confidence in the platform that a chemist will, if the calculation says it's going to have a property, the chemist will do anything they can.
Many months of effort may go into this making that molecule because they have confidence that it will work. The remarkable thing, and we were just talking about this earlier today, is we've gotten to the point now where these computational methods are so accurate, and this is a really big deal, saying what I'm about to say, that it's being used now to validate the experiment. When the experiment doesn't match the computation, more often than not, it's the experiment that actually went wrong.
I mean, that's a profound statement. That's really exciting. Relying on computation to predict the properties of molecules and doing it on a large enough scale to be able to identify molecules that balance all the properties that are required for it to be a drug or to be a particular material is a really big deal. Yeah.
Just to be technical for a second, is there any important considerations when we think about the cadence of software revenues over the course of the year? I'm flagging this just because it is a topic of conversation on the earnings call, and I think it could help us kind of contextualize the way you operate the business too.
Yeah. Our business is in a transition from what we call on-prem contracts, which are recognized at the point of signing the contract, where most of the value, not 100%, but most of the value is recognized in that quarter, to what we call hosted, which is ratable or SaaS-like revenue. That was about 20% of our revenue last year compared to 13% the year before. As that increases, it creates a base of revenue that basically recurs every quarter. We do expect over time that the sort of peak of revenue in the fourth quarter, which has been as much as 40% or 50% of our revenue, will gradually come down. That is not going to happen suddenly.
We did guide to $44 million-$48 million in revenue in the first quarter of this year. That is a big step up compared to $33 million in the first quarter of last year. There are a number of deals that we signed in the fourth quarter that give us revenue in the first quarter of this year that we're recognizing. That will not actually continue in Q2 and Q3, but Q2 and Q3 should still have a good foundation. Our Q4 concentration will go down this year, and will continue to go down in future years. It is not going to be a sudden transition or anything like that.
Okay. All right. Great. I know you touched a little bit on partnerships too, so I want to maybe transition just to that because you did have the update of the Lilly expansion, the Otsuka expansion, the Novartis collaboration recently too. I think when we zoom out and look across the sector, there's a lot of different partnerships in this broader space happening. I think more recently, when we're seeing the expansion of existing partnerships, I think that also speaks to kind of some of the progress being made within them as well.
That's right.
Maybe help us understand what kind of drove the expansion of the partnerships. What are they seeing that they're liking, and how is that kind of factoring into additional conversations that you're having?
Sure. I'll say that partnerships for Schrödinger have two elements, one of which is making drugs. We're working with Lilly and Otsuka to come up with novel therapeutics. I would say part of the expansion is certainly that those original projects are making great progress, moving to later stages, and there's capacity and desire to actually do it all over again. That's what you're seeing with Lilly and Otsuka in this case, is new projects actually being added to the collaboration.
The other aspect is that these collaborations and partnerships allow those companies who've already adopted the platform or are adopting it at larger scale to see how it works and to learn from us how it works on different types of programs. I would say across all three that you mentioned, Lilly, Otsuka, and Novartis, there is that fantastic opportunity for them to now really see firsthand how this works and to actually adopt it on programs we're not working on with them.
Yeah. So I guess when you kind of look forward into Schrödinger's future, obviously you have the software business, you have the internal pipeline, which we'll get to, and you have the partnership drug discovery revenues, right? So give us a sense of how you're thinking about a more mature version of the business that exists today and how you kind of balance those three revenue streams in the future.
Sure. I think that we're committed to the software business, and we're really excited about all the things that are on this slide, and we're continuing to invest probably more than the entire rest of the industry combined in terms of computational drug discovery into our platform. We see a very long runway in terms of the increase in use by large customers in particular, but also emerging biotech companies. People don't fully understand that in recent years, we've had companies come and say, "Your platform is our solution for drug discovery. We aren't going to build all of the chemists and all of the platform. We are going to use your platform to come up with drugs.
Those companies are being successful, and they're being acquired, and their companies' programs are going to the clinic. Biotech and pharma are adopting this technology at scale and will continue to drive sustained long-term revenue growth for software. We're always going to be in that business. Now, on the proprietary therapeutics, we're really excited about sharing our own data. We think that that gives us some validation in terms of our ability to discover drugs and pick targets. That sort of is another way of creating value for our shareholders from the platform. It's not instead of the platform. We're committed to both of these elements of the business.
Lastly, the collaborations, as Karen outlined, are essential to driving the growth in the software business. Showing our customers how to use the technology at scale, and we've said our scale is an order of magnitude larger than our largest customers. Demonstrating that inevitably results in them coming back and saying, we want to do that too. They come back and scale up. It is no coincidence that there is 100% overlap between our largest software customers and our successful collaborations. Right now, all of these things are working together.
Yeah. I think that's an important point too. I think this gets at, we had a shout-out to Dan Brennan, and I put out an R&D survey in December of last year, actually kind of looking at the relative investment internally and externally with some of these types of platforms, right? I think a lot of what people are trying to understand is where are we in the curve of not just adoption of the software broadly, but to what extent are companies trying to invest internally in their own capabilities, what are they willing to outsource, and how they're prioritizing that.
Is there any color you could maybe give us as you're thinking about growth drivers for your software business? What are those conversations looking like to you, and how do you expect that to kind of play out maybe over the near term in some of these contracts?
Yeah. I think it's obviously the feedback from our customers has been tremendous. They're telling us about, I mean, this is kind of cool. Not only are they telling us that it's having an impact, but they're talking about it in conferences, and they're talking to other customers. We often bring together the customers that are using it at scale and the ones that aren't, and it's amazing how influential that is. I think the feedback has just been tremendous.
Yeah. Okay. I do want to make sure we leave time for the pipeline. The internal pipeline, you guys have a bunch of really important, exciting updates coming this year in particular. Maybe give us just a layout of the catalyst for 2025, and then we'll kind of double-click into each one successively here.
Yeah. We have three programs in the clinic: our MALT1 inhibitor for B-cell malignancies, our SGR-2921, a CDC7 inhibitor for AML, and then thirdly, SGR-3515, which is our Wee1/Myt1 dual inhibitor for solid tumors more broadly. Each one of those has been in the clinic, and we're excited this year and too keen to start sharing data on our MALT1 program. This is the result of our dose escalation trial that's ongoing. Later this year, we'll be sharing information on 2921 and 3515. These are initial updates on those dose escalation trials.
Okay. Great. Maybe let's start with MALT1.
Yes.
What is it about your platform that lends itself, first of all, to this target and your compound specifically, how it differentiates versus other people who have gone after it in the past? Maybe just help us level set what are we going to see in the initial update?
MALT1, first of all, just to give some context, is in the BTK pathway. I think everybody's very familiar with BTK unless they've been living under a rock for years. It's a very successful class of drugs, clearly. MALT1 sits downstream of BTK and upstream of NF-kappa-B, and there are still patients who have hyperactive NF-kappa-B signaling. We chose this target because of the precedents that exist in the pathway, but at the time, there really weren't great small molecule drugs. They were actually peptidic, not very nice looking, and didn't perform well from a PK perspective. Our platform, of course, allows us to design beautiful allosteric inhibitors. This is an allosteric site in a protease.
One of the things that we thought was really important was having a drug that hits the target very hard and allows us to capture maximum activity in terms of reducing the proliferation of these B-cells. Preclinically, of course, we've shown that we have a very potent molecule, and we designed it with that in mind because some of the earlier generation MALT1 inhibitors were less potent. I think that has led to some concerns around dose-limiting toxicity. As you know, BTK inhibitors actually have some cardiovascular risk.
The early compounds that showed cardiovascular dose-limiting toxicity and also renal toxicity, we felt coming up with a molecule that did not have those features would be super important for this new class of drugs. This year, we're excited to share an update on the safety profile of SGR-1505, as well as the PK. Also, we'll be sharing some PK/PD information. Given that this is an early dose escalation trial, we're not powered to give you hard efficacy, but we'll be sharing our update on initial clinical activity in B-cell malignancies. We're excited to do that.
Great. Yeah. Look, I think another aspect of this that a lot of people try to think about when it comes to AI/ML physics-based platforms-driven drugs is how much read-through is there to the platform capabilities. I think fit for purpose, what exactly are you trying to show with this compound in this context often is equally as important as how does it compare to the drug of same mechanism that came before it, right?
I guess maybe we'll start with MALT1, but then maybe we can segue into the next drug too. It sounds like the primary aspect of what you're trying to do from a platform perspective with this drug is improve this tolerability profile, the off-target tox. If that works here, what does that tell us about Schrödinger's platform?
I mean, I think what it does is it doubles down on the messages that we've already received from prior drugs. Think about TYK2 acquired for billions of dollars last year. Was it last year? Maybe.
Maybe the year before, right? Yeah.
By Takeda of that best-in-class TYK2 inhibitor. I think it says more about that. We can do this over and over again. Take a feature of a compound that has shown activity, reshow that activity, but do so in a way that demonstrates that that molecule is progressible and potentially best in class. That is a feature, I would say, not just of MALT1, but also 2921. There are multiple CDC7 inhibitors. Exelixis and BMS had one over a decade ago, and it did not make it out of phase I. Showing we can keep a molecule in phase I and it does well and is developable, I think, is super important. Finally, Wee1/Myt1, which is a different type of situation because you have got two things going on there.
You've got the opportunity to create a beautiful Wee1 molecule, but also to rapidly leverage this emerging finding around synthetic lethality and not combine two molecules, which is a novel, novel combination in the clinic, very challenging, but to do this all in one molecule. Actually, we're really excited about the opportunity to demonstrate we can do that because actually having two activities in one molecule, I think, is becoming an interesting theme in the industry, actually.
Yeah. I guess that kind of leads right into maybe the Wee1/Myt1 conversation now. In terms of what data we'll see this year and what is a win for you guys this year on that molecule look like?
Yeah. I mean, I think it has to be said that that IND was filed mid-ish last year, and we started enrollment. Enrollment's going well. We're making progress. It is early in the dose escalation trial. This year, I think initial safety, PK/ PD, are the things that we'll be focused on. We'd love to also be able to share activity if we have that, but it is we've been in the clinic now for six-ish months, and it is still pretty early. Yeah, we'll be sharing the update on the profile of the molecule to the extent we have that.
Yeah. So look, I think a lot of exciting data updates this year too. As you're thinking about the capabilities of the platform, obviously, it's not siloed into particular aspects of oncology, for example, right? When you think about what to partner, what to hold on to internally, what to advance before others, how are you thinking about the next, let's say, five-year timeframe of what does Schrödinger's pipeline ideally look like five years from now if the next few data sets hit the way that you want them to?
Yeah. I mean, I would just answer that by saying that we've already been making those sort of prioritization decisions around what capabilities do we have. Obviously, we've got the platform, but as you think about selecting targets, running pharmacology studies, running phase I trials, and beyond, we recognize that there are companies who have very deep knowledge and expertise. We've elected already to take some of our early programs and partner those with companies like Novartis or BMS, where we think that's even an accelerant beyond the molecule to get these into the clinic and then into later stage trials.
In terms of what the portfolio is going to look like, oncology remains a huge unmet need, but we've been pretty successful with what we call modality switches. This is where there's an antibody out there, and it's been challenging to make a small molecule. We've successfully done that with Morphic. That asset was acquired. The whole company was acquired by Lilly last year. We see an opportunity there because combinations in immunology are increasing in importance. I think you'll see more immunology projects emerging from our early portfolio into the clinic.
We're going to be really careful about what we choose to put in the clinic versus what we choose to partner. We think that, as I said, there's an accelerant sometimes by working collaboratively early on rather than going the long haul and rebuilding capabilities that other people have.
Yeah. You also touched on, Ramy, you also did in your opening remarks about biologics, right? Obviously, it's not by any means limited to that too. I think there is a broader conversation now going on with a lot of these computational-based platforms, which modality is best and if that question is even necessary, right, at this point in time. Again, where do you see the best opportunity with that? Why are you where you are today?
Yeah. First of all, that's one of the beauties of physics. I mean, physics, of course, is completely agnostic, obviously, to the modality. I mean, physics is physics, interactions of molecules, interactions of molecules. It is even why our platform has been extended to design of materials. The problem is the same. That is one of the really significant advantages. One of the requirements for physics-based methods is a starting point for the simulation. I glossed over this here, and that's why the target validation and the structural biology part and protein refinement is so important. The starting point for these simulations is an initial structure of the protein.
That is actually kind of important to have that as a starting point. We just happened to choose at the beginning of our journey through proprietary programs to focus on programs where there were more structures available. That is why we are working on small molecules. It is not because the platform only is applicable to small molecules, but as the revolution of structural biology sort of becomes realized not only through experimental methods, but computational methods. You have heard about A lphaFold. You know about the work that we are doing to refine those structures, cryo-EM. There is just an explosion of the availability of structures.
As that continues to happen, some of it is work that we are doing, but a lot of it is work outside. The number of targets that are amenable to the platform, our platform, continues to increase significantly. It is not just biologics. It is peptides and ADCs and degraders. It is physics. That is the beauty of physics. You don't need training sets. It doesn't matter if there's any training. You can use these first principles methods right from the get-go to generate training sets using first principles.
I mean, you also touched on you mentioned AlphaFold. I mean, maybe zooming out just a little bit, obviously, AI is in every headline you see left and right these days. And you guys have spoken about the integration of AI into your computational-based platform. I guess help us understand a little bit how broader investments in the AI from even out of Washington, right? How that funnels and how that capital moves into healthcare, into what you guys are doing, and how you're positioned to kind of capitalize on that broader investment?
I think the way it's working is I think people just mean computation, algorithms, using computers when they say AI. To the extent that that's what they're saying, I don't think they care if you're actually using LLM to do drug discovery, which, of course, is silly. It's just computation. Obviously, just the attention, digitization of discovery using computers instead of just doing things by trial and error definitely has benefited us.
It's a little frustrating because we have to try and explain the difference between AI by itself and AI with physics, but that's okay. It takes an extra what? It took three minutes or something to was it two or three minutes? It's a little extra work, but we're up to the task. We just had to spend a little bit of time explaining the difference.
Yeah. I mean, look, I think as a lot of these computational-based platforms just continue to capture share, I think people are looking around at spending trends, right? Anywhere that you can try to save money, the biggest large-cap pharma to emerging biotechs, they're looking hard at their R&D spend realistically, right? I mean, what are some of those conversations with your customers looking like? Which stage of the development process do you think that they're more invested in trying to save that R&D? How are you guys able to capture that today?
Yeah. Do you have any thoughts on that, Geoff?
I think we're particularly tapping into those top-down-driven budgets that are generally related to AI for R&D. Different companies are approaching this in different ways, whether they're approaching it with sort of a digital officer or head of new technologies for R&D or something like that.
We're finding it's easier to get traction when there's a top-down-driven mandate and budget than when we're having to sort of work through a bottom-to-top-driven mandate or budget because that's where they're trying to sort of economize, whereas the top-down, they're saying not quite carte blanche, but there's a lot more flexibility. We're kind of approaching most of our customers at both levels, and that's where we're finding most success.
That's right.
All right. I know we've covered a lot of ground, and we're just about at time. I want to thank you all for joining us. Thank you guys in the room for listening in. A lot more exciting AI content to come today and tomorrow. Do stick around.
Thank you.
Thank you.
Thanks.