Hi, everyone. We're going to get started here with the next presentation. Thank you all for joining us this afternoon at J.P. Morgan's 43rd Healthcare Investment Banking Conference. My name is Harry Pearson. I'm an associate with the Healthcare Investment Banking team. It's my pleasure to introduce Dr. William Feehery and John Gallagher, Certara's CEO and CFO, respectively. We're going to do a presentation for about 20 minutes, followed by some open Q&A. Bill, why don't you take it away?
Great. Thanks very much. I'm William Feehery. I'm the CEO of Certara and very pleased to be here today. Starting out with our standard disclaimer chart, which I won't read. But if I move to just talk a little bit about the company, so Certara is focused on what's called model-informed drug development. This is a process during drug development where we seek to incorporate what's known about how drugs work in the human body, not just the statistical data that we get from clinical trials, but what can we know about the systems biology, what can we know about the drug from other similar drugs that have been developed. And the process is one in which we're always seeking to look at what data we've collected to make a prediction about the next step.
By doing that, we can help our clients make better decisions during clinical trials, and ideally, they change the odds and change the costs of drug development. We're not a new company. We've been around in one form or another for almost 20 years, and as a result, we've had some time to get this going, and we have quite a significant client list. We have a very significant group of scientists that work at the company that specialize in biosimulation and model-informed drug development, and they're very visible in the scientific literature and how we participate in that. And I think most importantly, we have proven results. You can look at plenty of recently approved drugs. You can look at label claims. You can look at the applications.
You can talk to clients about how our software has changed the course of the drug development in a positive way and how regulators have accepted that, and I think that's kind of the story here around how we're changing the face of drug development. Just a few numbers to kind of give you an idea of how big we are and where we are. We serve about 2,400 clients across the global pharmaceutical industry. Of them, 50% of our business, approximately, is in what we call tier one. Those are the top 50 or so pharmaceutical companies. About 30% of our business is in what we call tier three. Those are the pre-revenue biotech companies and smaller companies that are out there. So you can see we have quite a range in who we help and how we help them.
Our company is 1,600 people scattered across about 30 countries. Why are we so scattered? And the answer is, well, because we need to be where the global pharmaceutical industry is, and it is truly global. We have a very highly educated workforce, as you might expect in biosimulation, with about a quarter of us having PhDs or advanced degrees. And I think one of the stories here about biosimulation has to do with the acceptance of regulatory agencies. So our software is used by 23 regulatory agencies around the world, the most important of which for drug development are the FDA, the European EMA, Japanese PMDA. And that acceptance really is quite important because no one, and you'll hear me say this again, probably, but no one in drug development will do anything that regulators won't accept and ideally would like to foster. We fall in that category.
As a result, if you look at the results, we can look across over 100 drugs, hundreds and hundreds of label claims where we can see the use of our software where we're not only mentioned, but when we're mentioned in a label claim, that's a specific example of where our clients were able to avoid doing a clinical study. They avoided the cost. They avoided the time. They avoided the risk to the humans that would have been in those studies. And the regulators accepted that as they approved it. And that's really what drives a lot of the interest in our business. Now, why do we do this? Everybody in this industry, I assume, is in the pharmaceutical industry, so I might not tell you.
I might tell you something you don't know, but it's worth reminding you, which is the process of drug development is abysmal in the odds. So even after you spend a lot of time doing molecule selection and optimization, once you get to a clinic, there's still almost a 90% chance of failure. Now, that reverberates throughout the industry and drives up the cost of drug development to between $2 and $3 billion if you want to amortize all of the costs of the drugs that don't make it along the way to make one that actually does get FDA approval. And this has been true for a long time. We can look back decades. We can see papers published that show numbers like this. And you can find papers published that suggest that it's actually getting worse over time.
So what we're really about is, to some extent, you can think about it as we want to change the odds at each phase here. So what we're doing in biosimulation is at each phase, we're working with our clients to look at the available data that they've collected on their drug. It might be an early stage. It might just be lab data. As you move into preclinical, you have some lab data, maybe some animal data. And what we're trying to help them do is predict what's based on all of the known data about this drug, all of the known scientific data that's relevant, what's likely to happen in the next stage based on what we know, based on what's likely to happen in the next stage. Would you do anything different? Would you change the design of your trial?
Would you change the populations that are in it? Would you change the dosing? Would you stop, perhaps? Right? And applying that sort of disciplined thinking to this changes the path of drug development and ideally changes the odds that we see on this chart and the overall economics of the industry. So how do we do this? As I said, one of the tricks here in biosimulation is we work across all the phases of drug development. Now, obviously, you can't have one piece of software and one question across all phases of drug development. Questions you ask in drug development change as you move forward. But we kind of think about it as we focus on the really important strategic questions that a drug developer needs to really think about because it really drives whether the drug's successful and how much money you're going to spend getting there.
For example, in discovery, what are the important questions? One is, what's the best target to go after, and what's the best candidate to go after if you think about what's likely to be approved? As you move into preclinical, we get involved in translating data that you might have obtained from animals into the crucial dosing in your first- in-human studies. We get involved in looking at toxicity and risk profiles. As you move into early clinical, we tend to be involved a lot in the dosing strategy, and we start to get involved in thinking about different populations of people that are going to be exposed to the drug. If you think about different populations of people that are exposed to the drug, not all human populations respond equally.
People are different by not just age, weight, and gender, but by genetic profile, by comorbidities with other diseases you might have, and in other ways. So we want to think about what happens when this gets into a wider population. Maybe we want to change dosing. Maybe we want to monitor patients differently. Maybe we want to exclude some patients from the studies of this drug. And then as you get into late clinical and things are really starting to accelerate, a lot of money tends to be spent here. Now, we can help our clients as they start to think about drug-drug interactions. Do we really need to do those studies? Can we avoid some of them? Impact on populations.
We can think about things like translational medicine where we're looking at adult clinical data and starting to think about how do we translate this into pediatrics, for example. So there's many, many questions we can answer along the way with the software and the data that I'll show you. Now, how important is this? We spent some time thinking a little bit about this, and since we're modelers, we decided to put together a little model to show how we think about it. On the left is basically just some input data.
We pulled this from some standard published data just in terms of what's the failure rate of drugs at each phase of development, and then how much money is spent on the ones that fail at that phase, right, as opposed to so if you had 100% approval rate at each phase, you would avoid that cost. On the right, we've modeled this on the red lines. What's the cost that gets spent in each phase to develop one approved drug? Right? So we're starting with hundreds. It's being winnowed down over time to get one approved drug, and that adds up in our model to about $2 billion, which seems in line with what a lot of people have reported, then we decided to play a thought experiment.
We said, well, what if using biosimulation when we think biosimulation can be really impactful, but what if it's just a little impactful and we can just change the odds by a little bit? So we went back and we said, well, if we're going to just change the odds by 3% at each phase. So on the left, in lead optimization, instead of a 94% failure rate, we said biosimulation changes this to 91% failure rate. Still huge failure rate, not a big change. What happens as you flow that through the model? And if you compare the red and the blue lines on the right, you save about $300 million per developed drug. That is a big driver, so what's the message here? Pretty small changes in the odds lead to really large changes in the cost spent per drug.
That changes that can flow through the industry and really how drugs are developed through the world. Now, we played this forward. A lot of people ask us about our TAM. You can kind of use this to think about that as well. What we did here is we took that $300 million in savings. We said, well, as a company, we'd have to share that with our clients. So we'd only capture a portion of that, of course. We looked at how many new drugs tend to be approved in a year, multiplied that out. You can come up easily with an estimate of the TAM for us in the $4 billion range. Again, if we're affected by more than 3%, that'll climb. What's the overall point? Is the 3% a scientifically chosen number?
But the point is really, we only have to change things by a small amount to have a big TAM, to have a big return for our customers, and something that we can be proud to share in as we build our company, and you can see that tacitly, the industry has been accepting this. If we look at sort of data that's out there, you look at the scientific literature, which is kind of a measurement of what are the drug industry interested in, what are they working on. Since we've been active in this, there's been over 20,000 publications in things like in silico modeling and biosimulation, and then on the right side, I think, is a really interesting one. We've looked at the number of FDA guidances that specifically foster the use of different uses of biosimulation.
You can see over the years, they've been increasing as the FDA accepts more and more use cases for biosimulation as they approve drugs. And we expect that we're always in discussion with the FDA and other regulatory agencies as we develop more models and we push the level of acceptance. So what we're doing here in Certara, we've been around a while. We have a number of different pieces of software. And what we've been doing is knitting them together into an overall biosimulation platform. And the idea here is we're active across all the phases of development. The data that gets attached to drugs during development sort of starts small and it grows over time, but it follows the development stages. And bringing different software capabilities to the client at different stages, you can help answer those critical questions that I mentioned earlier.
Certara really has four software platforms, which I'll talk a little bit about what they do, that we're overall working to integrate. Underlying all of this is we have a tremendous amount of biosimulation models and data. And we have hundreds of people with scientific experience that are both developing our software and also available to work on our clients' drug development projects. And I'll talk in a minute about how we're inserting generative AI into this to enhance it. So just to talk a little bit about the products we have, I think the two ones that we have that kind of get classified in biosimulation are called one of them is called Simcyp and one of them is called Phoenix. Simcyp is a very large model of the drug kinetics in a human body.
We predict where the drug goes in the body, how it gets metabolized and excreted. The model incorporates all of the different organs of the body, and it's designed to vary for different populations of people. So you can use this model to take the available data and incorporate a lot of the known science of human systems biology to predict quite a lot about how a drug will perform in a clinic as you go forward. But it's really a loop between prediction and then do some experiments and then analyze the data and then go back into predictions for the next stage. Our Phoenix platform is more on the analysis side. So now we've done our prediction. We've decided to go forward and do an experiment or a clinical trial. We've collected the data.
Phoenix is used to analyze that data and then go back into the next loop, so these two software together is our biosimulation platform. Certara Simcyp is used, I guess, by a smaller group of experts. There's probably thousands of copies of Simcyp out there, and there's tens of thousands of copies of Phoenix out there. But it has to do with just the size of groups that are available as you're predicting versus you're analyzing clinical trial data. Our most recent acquisition was called Chemaxon. It was in the beginning of the fourth quarter. We made that very deliberately because we want to bring Certara more fully into the discovery phase. Why do we want to be in the discovery phase, well, because in the discovery phase of drug development, typically people think about it in terms of the design, make, test, and analyze phase.
A lot of decisions, a lot of really crucial decisions are being made about what drug to bring forward. What we intend to do here is, number one, we want to enhance the software available to our clients in the drug discovery phase. More importantly, we want to integrate this, as I said, into a platform. We're trying to bring the predictive capability we have in the clinical phase to the drug candidate selection process that we have in the discovery phase. Then lastly, probably we've talked the least about this in the past, but it's not unimportant. We have a product called Pinnacle 21, which is the underlying data product we have under our clinical trial data collection for biosimulation. Pinnacle 21 is a data validation product.
It's used to validate clinical data that, for example, that a pharma company might buy from a CRO or that you might use, for example, when you're submitting to the FDA, and we've added to this platform so that the problem we're trying to solve here is if you're a large pharma company, you have hundreds of CROs, you have hundreds of labs. To a large extent, your function is to take all of that data, bring it in, standardize it, validate it, make some sense of it, and that process is both expensive, time-consuming, and it gets in the way of doing really good science in terms of biosimulation, so our software is being built upon this as our data collection and validation platform. Now, like most companies probably have presented here, AI is being recognized as an important development for us as we go forward.
We got started pretty early in AI because we bought a company right before the really exciting announcements came out from OpenAI that's got people excited around the language processing capabilities of AI. And we've had some time to think a little bit about not just what can you do with AI, but what can we do with AI that will really change the company and would demonstrate that we really have something. So we thought about this in three ways. One is we're a biosimulation company. The key thing for biosimulation is we have very high-end experts that spend months and months of time developing these models from scientific literature.
We're developing AI tools to basically enhance the productivity of our experts, enhance the time that it takes to produce new models, which will enable Certara to expand the reach and to penetrate a lot of the white space that we haven't gotten into. The second one is we listed here as submissions, but think about this as every scientist has to write reports. Those reports might range from a page to, if it's a new drug application, might be millions of pages. The language models that are out there are very good at generating text, but if you just pull up ChatGPT and try to produce an NDA, I guarantee you won't be successful. There's a little bit more to it. We launched a product last summer called CoAuthor, which produces the clinical study report.
Our internal studies have shown a 30% savings in cost and time to produce clinical study reports, and we plan to expand that to more and more types of reports. That product has been launched to really good reception in just a couple of months. Let's just say we've gotten that to the sales in the millions of dollars, which is pretty good for a product launch in the pharmaceutical industry, and we expect good things from it as we go through into 2025. The point being, we actually have demonstrated a good commercial use of AI that's both useful to our customers and revenue-generating to our company.
And then lastly, as we get into discovery, there's plenty of opportunities for us here to tap into some of the data sources that we have available to create new predictors for the types of molecules that our customers would want to take forward in drug development. So we have one product out there. We have several features that have been added to our existing ones. And as we go forward in 2025 and 2026, you'll see a steady stream of new products coming out here in Certara involving our AI investments. Briefly, this is a case study. It's a little bit complicated chart, but I took it from one of our customers' public presentations. And the point here is we worked on Scemblix with our customer. And it kind of shows the journey, right? So we worked on this drug for about a decade.
We started small, a pretty simple model, helping them think about their trials. This particular drug had some complications to it. It had some interactions with some foods. It had some nonlinearities in the PK. It had to do with transporters, which changed the dosing and needed to be explained in the drug application, and we worked with this drug for, as I said, for 10 years. As they got more data, we refined our models, improved it, and the end result is the drug was approved. It was approved with all the biosimulation that we did. Our clients credited it with eliminating 15 studies that they would have otherwise done, so that's a direct savings to them, not just in money, but also in time.
I think it was then it went on to do even more as we got into doing translational studies for the pediatric studies that came after this, right? So it kind of illustrates drug development is a long process. As we get involved, we tend to start small in terms of revenues and modeling. But as this thing gets bigger, it can really add up in terms of the impact we have for our clients. All right. So our growth strategy really boils down to three things. One is there's still white space for biosimulation. There's still places we haven't modeled. There's still profitable places where we can apply biosimulation to help the drug industry make more decisions. And we're going to continue to invest in expanding our technology further. The second one is, as I said, there's an opportunity here to create a bigger platform.
And then the third one is embedding AI across all of this and even enhancing biosimulation further. Now, today, this morning, we announced some results of how we did last year. And you can see in the fourth quarter, we grew 13% in revenues and 22% in bookings. So we were pleased with our execution at the end of the year. Now, that does include a contribution from our most recent acquisition, Chemaxon, which performed well, we think, in the first quarter after the acquisition. For the full year, we grew 8% and 11%, which I think was probably less than we expected to be honest in the beginning of the year, but not bad considering it was a pretty lumpy year for pharma R&D in general, given some of the changes that were going on in both biotech and in big pharma.
So we're pleased with the performance. I would say our expectation as we go forward in 2025 is that the market would kind of remain similar to 2024. But because we're launching a steady stream of new products, we've made a good investment in our commercial team. And there's kind of a tailwind in biosimulation in general from our regulators. We think that we're well positioned as we move into the new year. This just shows over time what's our growth rate look like. We're basically in double digits growth rate over the last few years in both revenue and bookings. Now, we announced that our regulatory services business. We're looking at strategic options for that business.
So for the purposes of this chart, I said, "Well, what if we look at our business X regulatory services?" And you can see our core biosimulation services have also been growing in the double-digit range throughout that period. So we have a healthy business, plenty of opportunities to move forward, and a lot of demand for the new products that we've been investing in over the last couple of years. If you look at our mix between services and software, we have been very deliberately over the last couple of years investing organically and inorganically to shift that mix. 2021, we were 70/30. If you flash forward to 2024 at the end of this year, if we were to take out regulatory services, it's going to be more like 50/50. Now, we're not embarrassed about our biosimulation services by any means. Those are both high-margin, sticky business.
And more importantly, it's really necessary for the type of business we're in, given that there's plenty of clients out there that wouldn't be able to fully utilize the capabilities of biosimulation because they don't have all the expertise. So it's good that we have the ability to help them, and they recognize that with the growth. But really, at the end of the day, this is a software-focused company. By having the software, we get the right to add the services on top of it, is the way we kind of think about it. Now, we are continuing to invest in that software business. Over the last couple of years, we've been gradually increasing the amount of R&D we spend on software. Obviously, in 2024, that's a three-month figure. So it's also increasing in 2024. And we plan on continuing to increase it in 2025.
Now, why are we doing that? One is because we have a good track record here of building a high-growth, high-margin software business. Second is we believe that there's still plenty of good white space opportunities to continue to invest in things like cloud-based infrastructure, AI, and then this next-generation product development pipeline that we've created. And then a piece of the story in Certara has been the inorganic growth. We have been acquisitive over the years. This is just kind of a selection of the larger deals we've done over the last couple of years and how we thought about scoring ourselves and whether they were successful or not. So we think about acquisitions from the standpoint of strategically, does it make sense? Does the price we're going to pay make sense for our shareholders compared with, for example, developing it organically ourselves?
And then lastly, we spend a lot of time thinking about integrating the teams and keeping the teams. Largely in all of the cases here, that's been the case. We have somewhat different situations on why we buy some companies. For example, Pinnacle 21, we bought that. If you kind of look at it, we bought it three years ago. It's more than twice the size today. If you look down the list here, Vyasa was our AI engine. Applied BioMath was a very significant Quantitative Systems Pharmacology services group, which combined with ours has created, I think, the largest and most sophisticated group in the industry. So they've also had important strategic benefits for our shareholders.
But I think over time, with the size and the type of deal we've done and the integration capabilities we've had in the company, I think we would say that this has been a very successful and accretive program for our shareholders. All right. So my last chart is just to talk a little bit about the vision of where we're going. So we do already lead the industry in biosimulation. I think we're the largest ones that you can find independently. And if you look at scientifically, we're really pushing things forward. We're a trusted partner to a lot of companies out there that really depend on this software. And our investment opportunity here to integrate this software platform is significant.
Now, I would say just to talk a little bit about for the people in Certara and the culture here, a lot of us really see this as an opportunity not just to build a really great company, but to change the direction of an industry that's basically critical to everyone around the world, and so that really drives a lot of how we think about the world and what we can do for our clients, so we're really proud. We talk about transferring drug development from molecule to market. We've had a big impact. We haven't by any means tapped out the opportunity, and we're looking forward to continuing to grow and become a more important part of the pharmaceutical story as we go forward. Thank you very much.
Great. Thank you so much, Bill. I wanted to start, maybe kick off questions. I think you touched on the ongoing regulatory strategic review and helpful data that you had showed of the acquisition and divestiture kind of backward-looking numbers. What would it look like on a go-forward basis in terms of revenue growth and margins? And how do you think about that?
You want to start that one, John, or?
Oh, sure.
Yeah.
It's just how do you think about revenue growth and EBITDA on a go-forward basis as you think about as you're going through a strategic review?
Oh, I see what you're saying. Look, I think the strategic review, we own this, we actually own a very good regulatory services business that does medical writing. I think the issue is if you look at that business and you look at what we're doing in terms of biosimulation and leading the industry and what's the investment story of why you want to invest in Certara, it's probably not because we own a really great regulatory services business. We owned it in the beginning because it was important to kind of get ourselves credibility in biosimulation as that's become established. That's become less important to us. And that business has been not accretive to our growth story, right? So on the other side, it's a good business. It has margins in the 20%-30%. It's about $50 million.
We believe that there's value there and there'll be a good home for it and that we can redeploy the capital as well. And also not just redeploy the capital, but just sort of streamline the story and the investment strategy that we're pursuing across the biosimulation platform.
Great. And you provided a really helpful example or illustration of how you all think about building up the TAM for the business. What do you see as the most underpenetrated areas of the market right now that are exciting to you?
Well, it's a great question. We've looked at the difference between sort of the penetration between our best customer and ours. So we work with all of the large pharma companies, but there's really big differences in terms of how much they've embraced this. So some of this has to do with just biosimulation is pretty complicated. Even the experts sometimes don't realize all of the places that this can be applied. And so there's kind of an education and services component here that we believe that we can just penetrate the existing drugs we work on to a significant amount. There are also plenty of new technologies out there where we have been investing. So I'll give you a couple of examples. Over the last couple of years, we've been investing heavily in our neuro models because those have been important in the development of Alzheimer's drugs.
We've been investing in modeling the immune system because that's important as you select biological molecules. That's important as you think about CAR-T. It's important as you think about vaccines, for example, so as we develop those types of models, it's opened up all of those markets, all of those customers that are working with us. We have groups that are working on weight loss drugs. We have groups that are just working on cell therapies in the company, and those are kind of like very quickly moving areas where we're developing models, but also our pharma companies, our pharma clients are moving very quickly, and so we're right there with them, and those have been, I think, important things, and they'll continue to grow, and our penetration will continue as we go forward there.
Great. Before we move on, I wanted to see if there's any questions in the audience. All right. Well, I know it's late in the afternoon, but maybe can I squeeze in one more?
Sure.
Great. I think could you talk a little bit more about the company's strategy to integrate software and the capabilities within Certara and how Chemaxon fits into that strategy?
Yeah. Thanks for the question. It's a good one, so one of the issues that we thought about strategically for the company is that biosimulation to date has been focused on really the clinical studies. So we tended to get involved to some extent in preclinical, and then our involvement grew as clinical studies got bigger. Now, the problem is that if a company's already selected a drug, there's not a lot we can do other than tell you that that was the wrong choice. And A, sometimes that's not so easy to do. And it's sort of unsatisfying in general in terms of what I said about how do we change the odds of the industry. You really want to make a better choice early on so you can cut out all the spending on all the drugs that don't make it.
So we knew we had to get into drug development. The development area is very fragmented from the number of companies that are out there. There's lots of little software companies out there. And so what we thought about here is what we bought in Chemaxon is really a couple of kind of key capabilities. So one is the ability to search large databases of chemicals. So they have what's called a chemical cartridge. There's only a couple of them out there. It gets embedded in software all through discovery. And it's kind of a key piece if you're going to help someone select a drug. We need to be able to search through big databases there because it's a lot of data access. And then the second piece is they have software around doing product design.
Now, there are other companies out there that are doing things like AI drug discovery. That's not really in competition with us. If companies are doing that, they can plug into this. But the bigger picture here is we've got a solid platform in drug discovery. And then we want to integrate this whole thing. So as you're picking your molecule, we want to use our biosimulation technology to predict how that molecule will do in the clinic, not just against, let's say, a target. And we're providing additional information that you can use to screen molecules and make a better choice. So sorry, long explanation, but that's revolutionary in the industry. Nobody's really able to do it right now. And if we're able to do that with Chemaxon and some of the investments we're making, I think there'll be a pretty excited group of customers for it.
No, great. I appreciate the answer. Maybe one last one. You had touched a little bit on CoAuthor and what you all have been able to accomplish there integrating AI. What's the path forward in terms of that you see using AI to enhance internal development and product capabilities? How is that shaping your view of the product roadmap ahead?
I think we've done a lot of thinking about how do you have real impact for AI. For the core biosimulation technology we have, the reality of our market is that when you go to the FDA, you can't have a black box. The whole point here is to have an explanation about why a drug works and then based on your model, where did you get that model? Where did you get the equations? What did you consider? Did you backtest that model against other? They want to know all that, right? And so that's not going to lend itself to machine learning, for example. On the other hand, the process of producing those models involves basically inputting vast quantities of scientific literature that is the slowest, most expensive part of what Certara does, and it's really our bottleneck.
So we can enhance the productivity of our experts. So these very high-end experts, if we can make them three times as if we can get a lot of the busy work out of their way and set this up so they can develop models faster, that's good. And it can also expand our market because really what happens for our drug development clients is often they'll come to us, and if we have a model available, they'll absolutely want to look at it and play with it and ask questions before they go into clinical development. But if we don't have a model available, they won't sit around for six or nine months waiting for it to develop. They're just going to go do some clinical trials, right? And so the speed to producing models, I think, also is a factor in the size of our market.
So that's a big piece of this. And then the other piece of it is there's a lot of data that we don't take into account in traditional biosimulation because you might know it's important, but you don't know why. So a classic example is, let's say, genomics data, right? You know the gene is important to the disease, but that doesn't necessarily mean that anybody's figured out the exact chemistry of how it all works, right? So in the past, we would say, well, we don't use that data. But now that we have AI, there's ways of incorporating that data in the biosimulation software that we're delivering. And the more data that you can bring to those critical decisions, the better decisions you can make, and the odds will change, right?
Fantastic. Just give one more chance. And otherwise, I think thank you so much, both of you. And yeah, everyone, have a great rest of the day. Thanks.
Great. Thank you.