Schrödinger, Inc. (SDGR) Earnings Call Transcript & Summary

Welcome, everyone, to back to our afternoon session of the 2026 Leerink Partners Global Healthcare Conference. I am -- for the audience, I'm Mani Foroohar, Senior Analyst at Jack Medicines. I'm hosting for this session, the team from Schrödinger. Ramy, Richie, how are we doing?

Great. Great. Great to be here. Thank you for having us.

Awesome. Welcome to my adaptive hometown.

Yes.

A little nicer weather-wise, perhaps than New York City.

Although apparently, it's 70 degrees there today. So we could have had this conference in just in New York.

We just had to pull you guys down. The delightfulness on the side. Let's talk about the evolution towards AC. I'm going to dive right in. Sure. dive right in. And I'm assuming people who know the general contour of the business. Let's dive into the transition to ACV. To what extent that is a change operationally? And to what extent that is a change in how existing contractual relationships are accounted for or recorded. And I think there's a little confusion amongst investors on, well, how are we able to change this? How much of just accounting for economic activity that would have happened anyways?

Yes. I'm sure we can clear it up. Richie, do you want to add...

Yes. No change operationally whatsoever. This is how we run the business. ACV actually gives a closer metric to how we run it on a day-to-day basis. The change here is really about moving to hosted contracts. We've been gradually going there over the last few years. This is just an acceleration of that movement based on what our customers are asking for, what our investors are asking for to give a better picture into the revenue visibility. And it actually helps us support and deploy to our customers in a more accelerated way. So this is -- it's just a transition within our existing framework. It does have the impact of declining revenue this year. And just to spend a minute on that because it is kind of an unexpected thing. ACV and revenue at the end of 2025, both about $200 million. We are expecting our ACV to grow 10% to 15% a year. That is the operational metric for how we drive the business. Because we're shifting to hosted contracts this year, a hosted contract is recognized ratably over the life of the contract, an on-prem deal is recognized mostly in the quarter it's booked. And because of most of our deals are in Q4, just given our customer budgetary cycles, the deals that we booked in Q4 by switching them to hosted, we will have limited recognition of revenue in that quarter, but it will even out over the course of a 12-month period. So because of that -- those 2 features, 2026, we expect revenue to decline and ACV to grow. Over the course of a 3-year time period, we think that will start to even out as we transition over most of the book to hosted.

So on a channel, some of the investors that are a little less familiar with these dynamics. They would say, for a rapidly growing business, you booking revenue upfront is most appealing because you have lots and lots of new ads. Does this move towards a model where you recognize revenue more evenly over the course of the life of a contract, would that not suggest that perhaps this is a mechanism for the company to cover up declining contract adds? Are they all just plotting against us? Are you -- is this what you're doing, Ramy? Are you plotting against your investors?

Yes, of course, not. I haven't heard that, to be honest with you. I've heard that. I

Tell me.

I see. Okay. So no, let's be very clear. Obviously, that is not the case. It seems to me, Richie, correct me if I'm wrong, but by actually giving guidance and tracking the business with ACV, that doesn't hide anything. That's exactly a reflection of the true growth of the business. If ACV grows, customers are adopting the software on a larger scale, new customers are joining. That's the only way you can grow ACV. You cannot hide anything. So it's actually kind of the opposite. Does that clear it up? That has cleared up.

There's nothing further I can add there. If ACV is not growing, the business is not growing.

And if ACV is growing, the business is actually growing. There's no trickery. I hate using that word at all in the sense, but I felt it seemed like we were forced to use it. So -- but the opportunity...

I agree. Yes. Cool.

So diving into another dynamic that I think investors find challenging. is thinking about how to model and predict some of these larger, chunkier, almost biotech partnership like... Fund flows... Which are a little different than the base sort of software business. How should investors track the health of that market, the demand for those partnerships and sort of the pricing power, i.e., the terms you can demand in those deals?

Yes. Well, I think the first thing to say is we have an unbelievable track record with these collaborations. And really, I think it's underappreciated. Probably that's our fault. We haven't talked enough about it. But we've done a very large number of collaborations starting from co-founding companies like Nimbus and Morphic and StruXure and Ajax and the success of those companies has been unmatched. You don't see this kind of success in biotech companies. We formed quite a number of pharma collaborations. Those have been successful with what did we report $650 million in cash from equity stakes, from milestones from upfronts in the last 5 years. That shows a real track record, which, by the way, is not an accident. I mean that doesn't happen by chance. That's happening because the platform is resulting in better molecules being designed and that's having better outcomes in the clinic. We have at the moment, 16 programs in the clinic for which we have royalties on sales. So this is a real business. This is not an NF1, it's not an NF2. It's not -- I mean, you can't catch -- it's 16 actually, right? What I just said as far as just that, and it's many, many more collaborations. So -- but the other thing I wanted to point out that's really important, I think, missed a little bit is that, obviously, all the things I just said, it stands on its own. But the other thing that's really important to understand is the impact that, that's having on the software business. We've changed the way drug discovery is done because of the success of companies like Nimbus and Morphic. Drug discovery is now done differently in pharma because of -- because we've demonstrated for the first time that you can actually use computation to replace experiment. You don't have to make molecules by trial and error. You can actually do a lot of drug discovery on a computer. That's new. That's because of our platform and the success of the program. So all that success and validation has resulted in a change in the industry and how computation is deployed. And obviously, we're the main benefactors of that right now. So I think that's what's important to keep in mind the synergies between the businesses. They're not separated. Does that -- I answer in any way what you were asking?

It gave you some perspective. Okay. I think one of the other questions that people have is the defense market. There is broad concern around most businesses that sell software of any sort. I'm speaking very broadly. Well, what's going to get disrupted by AI, what's going to get displaced by AI? Are 4 kids somewhere is going to vibe code, their own Schrödinger. Talk about why that's not going to happen. How you think about parts of the business or parts of your counterparties that are susceptible to that kind of disruption or replacement and reasons why part are not.

Yes. Good. So there are 2 aspects of that. One is, can an AI model trained on experiment replace physics. That's one. And then the other is, can AI be used to actually generate the physics engine. okay? That's the one you asked about, but I think both of them. Now I think we've addressed very clearly the first thing. You need physics. You need ground truth to train AI. So AI models are not going to replace physics. Great. Okay, we've put that one to rest. Now can a couple of kids with cloud code and a garage rewrite enterprise software of the level of sophistication that we've developed over the last several decades. The answer is absolutely not, and it's totally ridiculous. I mean it really is completely out of the realm of possibility. I don't think anybody who's understands these technologies or either the technology, the AI technologies or the technologies that underlie -- the science that underlies what we've done, think that, that's a serious question. Sorry, I don't mean to insult you. I mean because you're channeling the question.

This is nowhere near the worst insult.

Yes. I didn't mean it. But no, no, look, it's a legitimate question. Obviously, people are asking it all over. And by the way, hundreds of billions of dollars of value are disappearing because of a belief that this is -- so it's a real thing. You have to ask the question. I'm just answering it. The answer is unequivocally absolutely not possible. The amount of proprietary knowledge that goes into these technologies, the amount of deep understanding of novel things that the AI doesn't know anything about makes it impossible for AI to replace the kind of software we're doing. Now that's not true, [indiscernible] I think you were hinting at this. I mean, let's talk about maybe legal software. So I hope that doesn't insult anybody or anybody's family member. But sure, there are fields where it's possible to learn, right, that there's enough in the public domain in books, right? I mean we see what -- how powerful LLMs are, where you can start to imagine those companies might be a little bit worried. And not for a little while, it's still a number of years away, but those are going to be the first to get replaced. But what we're doing, we are so far away from that being a threat that this most definitely does not keep us up at night. Now here's the thing. We use this technology very extensively internally. It's helping to make our developers more efficient. So we understand very deeply what its capabilities are. So when we tell you this is not replacing what we're doing, I hope people believe us.

So you talked about the value of proprietary knowledge. I'm going to pivot over to proprietary data, sort of data as an asset and how we can think about the value either in resulting cash flows or improvement of the platform, how we want to sell, however we should measure it of the collective pool of data you have from the various experiments and the calculations you've done at scale for both your own programs and for your customers over the course of the life of the company.

Okay. I understand why that's being asked because everything always feels like it's about data, but let me put it into context. I don't think the amount of data, and I'm going to give you an analogy in a second, but the amount of data that's being generated even computationally is not going to power AI, and I'll explain why. The idea behind your question is that maybe if we accumulate enough data. Whether it's experimental or computational, we can start to build a foundational model that can explain all of chemistry, every property that needs to be predicted against any protein or any confirmation of protein. And that turns out not to be the case. You will always, always need to generate new data using physics for every new problem that you encounter. That is. You have a particular target. You're going after a particular pocket. In other words, a particular confirmation of that protein. And you have a particular family of molecules, it's called the chemotype. You need to generate hundreds of thousands of data points to train an AI model to be able to predict one of the properties for that system. That model gets thrown away. And then you need to regenerate another hundreds of thousands of data points for the next problem, new target, new pocket, new chemotype and that will always be the case. That will always be the case. There is no -- this idea of a foundational model for chemistry, for design of molecules doesn't exist. So it's not about the data. It's about the ability to generate the data, and that's the physics engine that's unique to Schrödinger.

While we're here, let's talk a little bit about -- not just talk about the engine. We talked about some of the debates around AI risk. I want to talk about how to think about monetizing your existing position. So how should we think about the roll-on of predictive toxicology and how that additional feature service, whatever phrase you want to use, how that affects the value to the customer and how you monetize that on a per contract basis?

Yes. So predictive tox or the ability to predict tox, toxicity is one of the grand challenge problems in drug discovery. It's probably the major source of failure, maybe aside from biology risk, but it's a big problem. And it's a problem that occurs very late in discovery. So right, in other words, you're designing a molecule and everything is all great and then you go and do that one test either in vitro or in vivo and you discover toxicity and that kills the program. That's the end of it. That could be 5 years, $30 million down the tubes. So it's a big, big problem. And we've come up with a way of addressing a major source of toxicity, which is binding to off targets. There is a huge amount of interest in this, obviously, given what I just said. We have now results from beta testing that is better than we expected. It's -- like I said earlier, it's now as good as doing experiment. The way this is done experimentally is you take a molecule and you actually put it into in vitro and you test against a whole panel of off targets. It takes a long time to do it. It's expensive and you don't do it very often. And now there's a computational way of doing that. So now to answer your question, sorry, a little background, but customers have to pay extra for this. It's a new module. So it doesn't just get thrown in. It's also tapping into new budgets. So if our budgets before -- if our software was being sort of purchased by research groups, this is now by the toxicology groups, which are a little bit further down and I think generally have bigger budgets actually. The total spend actually on predicting toxicity of molecules is hundreds of millions of dollars done experimentally. So it's a huge opportunity, really, really big opportunity. So we're excited to be launching it this year. We're excited about the beta feedback. We think this is going to be a major contributor to growth over the many years as we continue to develop it. By the way, it's not done. We have -- in our panel of off targets, we have maybe 60 or so off targets. There's probably on the order of many hundreds of off targets you have to worry about. That's the experimental panels are that big. So we will continue to expand and continue to improve the product and continue to generate more growth from it over the coming years.

I'll just -- I'll add maybe 2 comments there, which is this helps us expand our addressable market. As Ramy said, reaching additional budgets, but it's expanding our capabilities within an organization. And second is that our entire business today predominantly is monetization of on-target discovery. This is off-target discovery. And so it has the ability to expand our applicability significantly.

Yes. So let's talk about... How to think about that -- the scale of that -- I'm not asking for guidance. But how to think about the scale of that contribution as a new module and on what time horizon that shows up in ACV inflection, potential acceleration of growth rate of revenue, how we want to think about that?

It is built, by the way, into our guidance. We expect to generate revenue from it ACV, I guess, I should be saying. We expect to generate ACV. We have to get used to using ACV as our new metric for a little while. This year, given the positive feedback, extremely positive feedback we're getting from beta customers. But as is always the case with a new product, this has happened to us before. We have a lot of experience releasing new products. But we have a history of creating new markets. That's what we're doing. We're creating new markets. So this is a new thing. So it takes time to get customers familiar with it with the idea of doing something new, testing it, they have to test it. They have to -- they don't just buy it on our promise. They have to test it. So we expect there to be a ramp-up and for it to grow over the years, but we do expect to see some portion of the growth that we've guided to this year will come from new products, and one of those is the predictive Tx module.

We've talked a little bit about the new module. We've talked about some of the nuances of interpretations of this pivot to ACV. Let's talk about end market growth. To what extent are you -- is your growth levered to new company creation? I know the majority of revenue is not from new company creation or new account creation. But how should we think about if we see a reacceleration of IPO market, VC market, et cetera, how does that flow through to you guys in terms of end-user demand? And what is that for you marginally?

It's -- I'd say we're not banking on that for this year. We're encouraged by the signs in the biotech markets, equity markets. There's a lag between those -- what's observable in fundraising and M&A and the translation of that to acquiring software and doing discovery. So we're not relying on that for this year, but over the course of the 3-year growth forecast that we've given, we are expecting biotech and the rest of life science markets to return to historical levels.

You're also looking at other growth markets outside of drug discovery...

Materials, et cetera.

Where are we in terms of the maturity of the platform for those applications? And how should we think about their contribution to the growth profile?

Yes, that's a really great question. We started that division because it turns out physics is physics, and atoms are atoms. And a lot of the problems in material science, it turned out we could leverage the physics-based these fundamental first principles methods that we had developed for life sciences. Polymers are polymers. So protein is a polymer, but the polymer that coats the airplane wing is also a polymer made up of the same types of organic elements, and we can start to try and understand the properties of those polymers using the same technologies. We have sense develop new technologies that are very specific to material science. Batteries is a good example. The electrochemistry and the -- that's occurring at the interface of electrolyte and electrode in a battery, it turns out there's no exactly biological relevant sort of system. So we have been developing new technologies, in particular, technology around battery chemistry, which requires something called machine learn force fields. Essentially, these are a type of force field that is somewhere between the classical force fields that we use for modeling in drug discovery and quantum mechanics. It's got the accuracy of quantum mechanics and the throughput of classical force fields, new thing, very exciting work. We, for the first time, have been able to simulate that chemistry that's occurring at the electrode, electrolyte interface, which is -- has the potential to allow us to design better batteries for which there's clearly fantastic demand. So I'd say we did a pretty good job getting the business up to a certain level, leveraging existing technologies. But now we're in this innovation stage where we're saying there's some news. And I think you know that was funded by a rather generous grant from the Gates Foundation. We've invested that. That's paid off, but it's just coming online. We're just starting to publish some of the work. So we're really optimistic about the future as we did -- like we did in life sciences. We've been doing so much innovation. We've changed the field. These free energy methods we've developed has been transformative. We think the same thing can happen in materials, but it's earlier days.

I you say earlier days, is the right way to think about the materials side of the business as a growth driver that primarily lives on the other side of the 3-year guidance that you've given? Or should we think of it as a meaningful contributor within the context of this 3-year period?

It's both, actually. I think we will see -- we're expecting that growth in material science business will contribute to the overall growth in this 3-year period. But given what I just said, it's true potential and maybe the real inflection is probably just a little outside that.

Okay. That being the case, what should we think about as the sort of trigger point of turn for that inflection? Is it a technological development? Is it adoption amongst a group of executives who are not used to using these tools? Like what is the event we should look at, okay, this is a sign that we should start modeling more growth from there?

Yes, that's a fantastic question. In pharma and biotech, starting maybe 10 years ago, something like that, was a transition of real acceptance and adoption, I think, driven by us of using computation to design molecules. The material science world is behind. It's behind that. It's being used, but you can tell by the number of computational chemists in material science companies is way lower than it is in pharma. Pharma and biotech, and it really started much longer ago actually, has embraced the idea of using computation. It wasn't working very well, but they embraced it. That's something that hasn't happened yet. That transition has to occur. Now of course, we have to facilitate that. Why would they do that? If there's no technology that's actually useful, why in the world should they invest in computation. But I think that's what has to happen. And it's a chicken and egg problem, right? I mean they're not going to invest in that until they see the technology. But then if they don't start using the technology, they're never going to see the impact. So it's a little iterative process. But -- so I'm not sure I'm answering your question directly, but that's how we will see it. You might not be able to see it. But when we see material science companies embracing computation and the way the pharma industry did a decade ago, that will be the sign that all of a sudden, that field is about to be transformed by computer-aided design just like our drug discovery field has been.

In pharma, a big part of that was people training to become scientists, being early adopters in academia, exposed to academic labs as part of their PhDs, exposing their students to it. That was a fairly long lead time. It's one of those things that it takes forever, takes forever, then it happens all once when it happens. Makes it hard to model the material.

I know. Well, but we are putting that effort in. We have a significant education effort. We put so much work into getting academics using the technology. We actually developed shroding or developed online courses to give to professors to teach computation to students. So yes, it's going to take a generation, not a whole 4, 5 years, right, as students work their way through. But that investment works so well on the life science side. We're convinced it will work on the material science side, too.

Okay. And I think that captures a lot of where we are in that side of the business. Looking forward, past the other side of the 3-year plan, again, nice new guidance. You've given a real clarity on a path to profitability.

That's right. By '28.

Fast forward, we're sitting here in '28. You landed some -- you landed there, maybe a little higher, you're profitable. Sure. Looking forward, how do you think about use of capital in the very long term for Schrödinger? Once you're profitable and growing, obviously, operating margins in this kind of business are pretty attractive. Like what is the right use of incremental capital?

So one of the things that is exciting, at least for me and I think a lot of people in the company is that we're never done when it comes to the platform and innovation. I mean drug discovery is still incredibly hard. There are a huge number of failures. It costs a ridiculous amount of money. It taking 4 years, 5 years to get to a development candidate, that's not okay. It should be taking a year. And the failure rate should be 0 when you get to a certain point. So we will always be I hope this is the goal is to always be the -- we have been. We've been the leader in this space. We've been defining what it means to innovate in this space. And I think we are many, many decades away from saying, "Oh, we're done, everything is all good now. Now drug discovery is as good as it's going to get. And don't forget, you have the whole material science, which, by the way, isn't just one field, material sciences, a huge number of different fields, right, from aerospace to chip design to battery design and countless other things, other types of materials. So that's kind of an exciting thing to be a part of to be able to drive the field forward and keep making it so that we don't have to wait for 15 years for life-saving medicines or materials that change. So that's a big part of it. Now I think the other part of it is one of the things that has been frustrating for us is we've played a key role in generating an unbelievable amount of value for companies that we've been involved in cofounding. But we've owned a very small part of those companies. And I think as we get to a point that you were just describing, there will be an opportunity for us to own more of it. And I think that will also be really great for Schrödinger, for shareholders. So that's another thing that we're looking forward to. That's a little hard right now, obviously, but do you agree?

Yes. Just to quickly expand on that is we spent some time on the Q4 call laying out our portfolio of milestones and royalties. But by that time period, I'd expect to see some of those contributing on a recurring basis at a near 100% margin. So...

Yes.

Awesome. On that note, we are already over time.

Yes, we're a little... That's good.

We look forward to continue the conversation soon.

Thank you so much. Great discussion.

Thank you.