Revolutionizing Drug and Materials Discovery with Machine Learning & Computational Physics
Drugs and materials are a foundational cornerstone for the further advancement of human health and technology. Drugs play a crucial role in managing diseases, lengthening human life spans, and improving the quality of life for humans across the globe. Billions benefit from the advancements our society has made in developing new medications, but the process of developing these drugs is neither short, cheap, nor simple. From the initial concept of a drug, to identifying the appropriate components necessary to manufacture it, to introducing a safe and approved drug that is available to the millions of consumers whose lives could be improved by it, the entire process can take up to a decade and cost billions of dollars.
Most of this is spent in the ‘discovery phase’, the most complex and costly part of the process by far; during this part of the process, researchers determine the molecules or compounds capable of creating the drug, then test them in a series of different scenarios. The majority of potential molecules are determined to be unusable as researchers slowly refine the pool of candidates throughout testing.
The process is very similar for materials sciences, an equally important industry in human society for discovering and optimizing polymers, silicon hardware components, and the metals in our vehicles. Uncovering new materials will continue to play a huge role in allowing technological leaps and opening doors to innovations that reshape industries.
Given the importance of both drugs and materials in benefiting industries and the wellbeing of society as a whole, uncovering new possibilities in both of these spaces should be a top priority. Unfortunately, the discovery phase is so fraught with challenges and the potential combinations of molecules is so vast that traditional lab-based testing methods can only explore a fraction of the potential outcomes. This means that most potential drugs or materials will never see the light of day - the process is simply too time-consuming and costly to test all the potential combinations. But the missed opportunity to build a better world, one could argue, is far more costly to the entire human race.
Enter Schrödinger (Ticker: SDGR), a company leading the way towards a quicker, more efficient way to navigate the discovery phase. The company has developed a highly complex computational software that predicts how molecules will behave and combine with other molecules, in turn allowing researchers to simulate and analyze molecular interactions in a virtual environment. This reduces the need for costly and time-consuming physical experiments while achieving the same results, creating a faster, cheaper, and more efficient drug and material discovery phase. Schrödinger’s platform is paving the way for a future of innovation in some of the world’s most important and society-shaping industries.
**Disclaimer: I’m not a financial advisor, analyst, planner, or anything else to do with finance (besides a nerd). This article is intended solely as a foundation for your own research and decision-making processes. Please don’t make any financial decisions based solely off what you read.**
As of Writing - SDGR 0.00%↑
Market Cap: $2.5bn
Gross Margins: 60%
“Somewhere, something incredible is waiting to be known.”
Schrödinger was founded in 1990 by Richard Friesner and William Goddard III, and was named after the famous Nobel Prize-winning physicist Erwin Schrödinger (if you’re thinking about a cat in a box right now, you’re on the right track). The two founders, both distinguished scientists in their own right, were very cognizant of the chokepoint in the discovery phase that was so crucial to uncovering and commercializing new medicines. They set about creating a solution to that problem by blending two different approaches to predicting the properties of molecules - one was machine learning, which was efficient and scalable, though limited on its own, and the other was a computational, physics-based method that reduced the limitations of machine learning but was slow, cumbersome, and unscalable. While both of these approaches had been attempted in an effort to make drug discovery more efficient, neither worked on its own. Goddard and Friesner’s vision for blending molecular physics and technology into a single, efficient solution was the foundation for creating Schrödinger.
Over the following decades, the company enlisted the help of industry leaders, scientific experts, and software engineers to build out a complex platform that was capable of simulating molecular interactions. Their cutting-edge technology has allowed them to build strong partnerships with some of the most well-recognized academic institutions and biotech giants, collaborate with some of the world’s largest pharmaceutical companies, and more recently branch into their own drug discovery pipelines. And while the company has been around for a significant amount of time, more recent advances in computing capacity, AI, and machine learning has allowed Schrödinger’s platform to take a step closer to its full potential.
The company came public via a good old-fashioned IPO in early 2020, at an opening price of ~$26 per share. The company followed the broader stock market to eye-watering valuations and a subsequent nosedive over the coming years, and as of writing sits at $35.82, good for a market cap of about $2.5bn and a total return of 38%.
Schrödinger continues to operate at a loss, but it presents unique potential for investors with a healthy appetite for risk and a long-term investment horizon. With any company, though, and especially with very risky growth names like Schrödinger, it’s important to understand the business thoroughly before making a decision on whether this company has a place in your portfolio. So, let’s dive into Schrödinger’s business, management, and some of the red/green flags that investors interested in the stock ought to be aware of.
The basic business model for Schrödinger is relatively simple; they sell their software on an annual contract basis for customers within materials science and drug development industries. Though they were previously focused exclusively on expanding and improving their software platform, Schrödinger has started to transition towards being more of a bio & tech hybrid by collaborating with pharmaceutical companies to advance the discovery process. Their revenues now come from a combination of sources within these collaborations, as well as the annual revenues from licensing their software. While there are several revenue streams, Schrödinger broadly categorizes them into two segments: software and drug discovery.
“How drug discovery is traditionally done is trial and error. Schrödinger is transforming the way scientists are designing drugs by allowing scientists to do more experiments on the computer, and fewer in the lab.”
-Schrödinger CEO, Ramy Farid
Schrödinger’s software platform helps biopharmaceutical and materials science companies predict the properties of molecules through accurate, physics-based methods in combination with fast and scalable machine learning. Compared to traditional methods that are only able to analyze approximately 1000 molecules per year, Schrödinger’s software is able to analyze billions of molecules every day. This allows a massive increase in the probability of finding a molecule with the desired properties (one study found up to an 8x increase in desired molecules identified) at far greater speeds and levels of efficiency. In plain English, the software does more, does it better, and does it faster.
The team at Schrödinger recently built and added an application to the platform that allows for communication and knowledge-sharing between interdisciplinary teams with different responsibilities, as chemists often work with teams containing experts from a variety of fields (which I’ll touch on a little more later).
The software is provided to biopharmaceutical companies and businesses in the materials science spaces. This is the primary revenue generator for the business and contributes 75% of total revenues. The software offers attractive 78% gross margins (+1% YoY) and is still growing fast, with 20% YoY growth reported in the company’s 2022 annual filing. Within the software segment, the company also generates revenues from software maintenance (updates, technical support, etc.) and professional services, which include training, setup & installation, and assisting customers with tasks.
However, much of the software (not all, though; more on this later) is licensed out based on annual or multi-year contracts, meaning that growth in this segment can be very lumpy. Looking at the last couple years of software revenues…
… we can see that these are signed mostly in Q4 and Q1 before tapering off afterwards. Management also cited a large burst in multi-year contract signings at the beginning of 2022 as the primary reason that the software segment recorded its first reported YoY decline in revenues for this most recent quarter (Q2 ‘23), as the revenue recognized for this segment is entirely due to the timing, size, and number of contracts that are up for renewal in a quarter. Management is currently in talks to sign several $1m+ contracts in Q4 ‘23, with some of these being multi-million dollar, multi-year contracts that are likely to give a nice boost to shares in Q4, according to management.
The software has decent penetration into the pharmaceutical market as well, with Schrödinger reporting that all of the top 20 pharmaceutical companies (by revenue) licensed out the software in 2022 to enhance their discovery processes, demonstrating the value of the software and Schrödinger’s success in catering to these clients, which accounted for 32% of total software revenues for the year.
As for materials sciences, Schrödinger stated on the most recent 10-K filing that their penetration into this market is only just beginning; the underlying physics are identical between materials science and drug discovery processes, so the platform transfers easily towards this application. Where drug discovery (without the use of software) can take anywhere from 6-10 years, however, materials science is far more demanding, with new materials requiring anything from 10-20 years to bring to market, which would suggest to me that there is even more value to be had from Schrödinger’s platform for materials science companies. These companies can use the software to find new materials for a variety of applications: electronic displays (OLED), polymers and composite materials for aerospace and defense, microelectronics for semiconductors, and more sustainable substances for batteries.
Schrödinger has disclosed two current agreements to license out software for materials science applications; Bill Gates’ own Gates Ventures, which uses it to test different simulations with the ultimate goal of improving battery performance, and Eonix, which uses it to design materials for safer and more energy-dense lithium batteries. Schrödinger received a small equity stake in Eonix as part of this agreement and is elibile to receive more if Eonix is successful in hitting some agreed upon milestones in the discovery process.
My final point on Schrödinger’s software here is the computing requirement; there is a ton of computing power that goes into powering the software, and the networking infrastructure it requires is massive. Therefore, Schrödinger offers its software platform via two methods 1) on-premise software, where the customer pays an annual fee and is granted a license for the software and hosts it on their own infrastructure, either physical or cloud-based 2) hosted software based on a subscription fee (recognized throughout the year, rather than a single, annual payment), where Schrödinger hosts the license and customers have access to the cloud-based platform.
For customers with inadequate networking infrastructure to host computationally demanding software, it’s far cheaper to opt for the hosted software; I think this was a wise move by Schrödinger to offer this additional hosting method, as the investment in computing power would be a barrier to adoption for companies with fewer resources, and it provides more stable and recurring revenues that are less lumpy than annual contracts. As for Schrödinger’s own computing capacity, the company entered into a partnership with Google Cloud in 2020 to expand the capacity and speed of their platform through what essentially amounts to a supercomputer, with thousands of Google GPUs helping to accelerate platform speed and machine learning processes, as well as host the software for their customers.
“High-performance computing and access to compute power has really allowed us to scale the number of calculations, the number of simulations we can do, in trying to find the right molecule, for the right protein, for the right disease.”
-Karen Akinsanya, Chief Biomedical Scientist at Schrödinger
On top of licensing out their software to pharmaceutical and materials science companies, Schrödinger also uses their platform and team of experts in drug discovery collaborations and partnerships with pharmaceutical companies. These typically involve Schrödinger researching and identifying drug candidates or specific molecules for pharma companies, in which they become an active part of the development process. There are a number of advantages provided to customers who enter into collaborative agreements with Schrödinger, including:
Schrödinger uses the software for the customer, allowing them immediate access to the full suite of benefits the software provides without having to spend time and money training their own staff to use the platform.
This has the added benefit of not requiring customers to make a significant infrastructure investment to run the platform themselves, which requires some pretty significant computational capacity that Schrödinger already has (through their partnership with Google Cloud to run their software).
Access to Schrödinger’s team of industry experts and scientists, as well as early access to any new tools or functionalities that Schrödinger rolls out on the platform.
These partnerships are mutually beneficial; firstly, successes in these partnerships and with larger pharma companies are a huge validation of Schrödinger’s platform. Schrödinger is also in turn able to generate revenues from these partnerships through upfront payments, research funding, milestones throughout the drug discovery, development, and commercialization processes, and potentially through royalties and/or equity stakes in companies, depending on the contract. They have equity stakes in a number of companies:
The fruits of these equity stakes were demonstrated in February of 2023, when a branch of Nimbus Therapeutics was sold to another pharmaceutical company. Because of Schrödinger’s minority ownership in Nimbus, they received a total of $147m in cash from the closing of the acquisition. In an industry where consolidation is the norm, more one-off deals like these can likely be expected. Furthermore, this is just one way that Schrödinger can generate revenues from these equity stakes; obviously, if any of these companies are successful, Schrödinger will in turn benefit. However, this is, again, a lumpy revenue generator that makes it difficult to predict Schrödinger’s future revenues.
As for non-equity partnerships, Schrödinger’s most important by far is with Bristol-Meyers Squibb (BMS), a global pharmaceutical titan (~$130bn market cap). They entered into an agreement in 2020 for Schrödinger and BMS to collaborate on research and identify a series of molecules. Schrödinger received $55m as an upfront payment when the agreement was made, with the potential to receive up to an additional $2.7bn from BMS for discovery milestones across the molecule targets. If any of the molecules that Schrödinger discovers are successfully commercialized by BMS, they can receive royalty payments on all sales of the product in the range from mid-single digits to low-double digits (probably around 5-12%).
This partnership with Bristol-Meyers Squibb is going to be an immensely important test of Schrödinger’s capabilities, and I believe that their success in this collaboration will largely determine the success of Schrödinger and subsequently the share price over the coming years; if they are successful, their software will be proven as effective and value-additive to pharmaceutical companies (given that Schrödinger itself will be using the software in discovering the molecules) and their value as a partner/collaborator with one of the largest pharmaceutical companies in the world will be demonstrated. While I believe that the company can continue to ramp revenues and drug discovery revenues over the coming years, I believe the current valuation on the stock is pricing in success with the BMS partnership - if they are not successful, this assumption will change, and they will likely face some multiple compression that hurts shareholders.
As a final note on the drug discovery front, Schrödinger launched their own in-house discovery efforts in 2018, wherein they are attempting to develop their own targeted therapeutics for commercialization by using their software platform (this is where their move to become more like a pharmaceutical company than a pure biotech business comes in). Schrödinger has enlisted a team of 150 scientists for this business segment, including experts in biochemistry, biophysics, medicine, and computational chemistry fields, as well as employees with experience in preclinical and clinical development. With this team, the company has ramped up its in-house drug discovery efforts over the last 5 years, and is now in the process of developing therapeutics for cancers, immunology, and neurology.
This part of Schrödinger’s business is very much unproven; to date, they have not commercialized a single product or recognized a single cent of revenue from their own drug development efforts, though they do have one drug in Phase 1 clinical trials at the moment; called SGR-1505, this drug is a MALT1 inhibitor, meaning that it seeks to target the MALT1 protein in our body and inhibit its functions - if the MALT1 protein is out of whack, it can result in a lymphoma cancer. So SGR-1505, if successful, will serve as a ‘brake’ on the protein, helping to control or stop the spread of cancer from the MALT1 protein. They also plan to start another Phase 1 clinical trial on their SGR-2921 drug, which will seek to inhibit the CDC7 protein that’s responsible for cell division and creation - if not functioning properly, it, too, can result in cancer. This trial is set to start in the second half of 2023, but neither of the products in clinical trials can be expected to hit commercial markets anytime soon, if at all. This segment of the company is very much an unproven story, one that requires an immense amount of patience from investors as the company continues to realize operating losses on it.
In summary, the Drug Discovery segment of the business is still very much in its early phases and operates at a gross profit loss, though it’s trending closer to gross profitability, with growth of drug discovery revenues (84%) far outweighing the growth in cost of revenues (10%). The segment recognizes revenues from upfront payments for collaborations, research funding, and drug discovery/development milestones. They can also receive royalties from partners that are able to successfully commercialize drugs using Schrödinger’s software, though this is dependent on the contract of the collaboration.
Fairly decent results for Schrödinger on the Glassdoor front, with a 4/5 star rating on the overall employee experience at Schrödinger and a nearly 80% approval for CEO Ramy Farid. I didn’t find anything that really gave me cause for a deep second look on either the positive or negative sides here, so let’s dive into the CEO and the rest of the management team.
CEO - Ramy Farid
Ramy Farid is undoubtedly a smart guy. I should, I suppose, call him Dr. Farid, because he achieved a post-doc in chemistry at Caltech in 1991, and joined Schrödinger fairly early on, beginning as a scientific advisor for the software platform in 2002, and slowly working his way up the ranks before taking over the head honcho job in 2017. He oversaw the 2020 IPO, and has been involved with expanding the company’s portfolio of joint ventures.
Farid also brings a lot of experience with the business side of the industry, serving on the board of directors for a long list of different therapeutics and pharma companies, many of which are directly involved with Schrödinger today through collaboration programs. Beyond these facts, unfortunately, there was little else that I could uncover on Ramy Farid. Being the CEO of a smaller company that isn’t typically covered within investing circles, and a freshie at that, there simply isn’t a lot to find on the guy. While I didn’t find anything negative about him, I also didn’t find anything terribly positive about him. As I believe management to be such a key factor in a company’s success, I’m going to lean towards this being more of a detractor than a neutral rating when I build out Schrödinger’s report card.
The original founders, Richard Friesner and William Goddard III, are still involved with Schrödinger today. Both of them are scientific advisors, while Friesner also serves on the board of directors. Neither of them hold c-suite roles today, but reading their resumes, it’s not surprising they were able to found such a scientifically advanced company. Both Friesner and Goddard have received pretty significant praise within the scientific community and a number of awards. In fact, the founding of Schrödinger seems to be a footnote for their total list of accomplishments, from what I could tell during my research.
Nevertheless, it was difficult to find a ton of information on them pertaining to their actual roles and responsibilities at Schrödinger. While I view it as an encouraging sign that Friesner is still involved with planning the strategic direction of the company as a member of the Board of Directors, neither is involved in day-to-day operations from what I could tell. I generally give major bonus points for founder-led companies, as I like the shareholder alignment that one typically finds with these companies, so I have to view the founder’s roles as more of a detractor than a neutral point as well.
Normally, I don’t look too closely at the employees beyond their overall rating of the company on Glassdoor. With Schrödinger, however, the employees and scientific teams are an important part of the story. Firstly, this is a highly competitive field with lots of demand for experts - with Schrödinger’s broad array of needs for high-level scientists in biophysics, software, computational chemistry, medicine, and lord only knows what else, it’s immensely important that they are able to retain talent. Secondly, I believe the employees and the expertise they bring help to build out Schrödinger’s competitive advantage over businesses offering similar products, as well as providing a barrier to newer entrants hoping to enter the market. Thankfully, they are doing exceptionally well on this front; the company has an employee retention rate of 93% as of the end of 2022, with 344 of their 787 total employees holding PhD degrees (nearly 44% of their workforce). I will get more into the competitive advantage of such a highly skilled workforce later in the article, but I definitely view this as a strong positive for the company.
This is a new segment that I’ll be focusing on more for future articles. How a company allocates its capital directly influences the success of the company and the return for shareholders, so it’s an important thing to research.
As for Schrödinger’s capital allocation, they’re focusing on two primary areas of investment; the software/platform, and their own drug discovery portfolio. For the software side of things, this is exactly what I want management to be putting their capital towards. This money is primarily going into R&D to improve the underlying science of the platform and its accuracy/capabilities, as well as hiring software engineers and programmers.
As for their own drug portfolio, I have to say that I’m still a little skeptical about Schrödinger’s move to wholly-owned drug discovery programs. It’s a highly risky and extremely costly move by them, and one that so far hasn’t even been proven on the gross profit front. What I most dislike about this move is that it turns their customers into their competitors; they are now competing with the people that they’re trying to sell their software to by developing their own therapeutics. I personally like businesses that create win-win situations for both themselves and their customers, and their expansion into in-house drug discovery doesn’t follow this mindset.
Put yourself into the shoes of a pharma company that is producing a MALT1 inhibitor drug, similar to Schrödinger’s lymphoma treatment; you still need computational software to assist in the drug discovery process, but now buying it from Schrödinger is also providing revenues to a direct competitor, essentially assisting them in possibly developing a more effective drug that would render your last 2-6 years of development useless. You’re probably not super stoked on this and may even consider going to one of Schrödinger’s competitors for their software and swallowing the switching costs just to avoid this. Pharma is a highly competitive industry, and this is almost certainly the mindset of at least some of their customers as Schrödinger begins its own discovery portfolio.
Furthermore, it’s highly risky; Schrödinger has no experience with drug development, the different phases of clinical trials, or commercializing products. This may be a classic case of diversification for diversification’s sake, or diworseification, to sustain high revenue growth figures and prop up share prices for a while. Overall, I don’t like the move. I’m a big supporter of the move to collaborations and partnerships, and I think that was a strategic move, but this particular strategy has the potential to alienate their customers while simultaneously falling flat on its face, never earning a dime, and rapidly ballooning operating costs for no good reason. And if all that wasn’t enough, there’s a pretty significant opportunity cost involved if this doesn’t pay off; diverting resources from the core software offering may slow Schrödinger’s capacity to further build and improve their platform, which may give competitors an opportunity to swoop in and steal some market share for their most valuable product.
At a Glance:
YoY Rev Growth (Q2 ‘23): -8.5%
Gross Margin (Q2 ‘23): 55.8% (+8.2% YoY)
3-Year Revenue Growth: 18.7% CAGR
Growth in Outstanding Shares: +2.86% since 2020 IPO
Negative free cash flows, EBITDA, EPS, and returns on investments
Now, it’s time to be honest with ourselves. An investment in Schrödinger today is an investment in the story, which to date has been mostly unproven, and the hope that they will be able to achieve their goals in the future. The company is still early on in an expensive growth phase, and as such, there is little to like and lots to dislike about their balance sheet. We’ll get into the negatives here before I cover the few positive indicators of their financials.
Balance Sheet - The Bad & The Ugly
SG&A Expenses Growing Faster than Revs
Selling General & Admin Expenses are grew nearly 41% YoY, while revenues grew only 31% over that same period. While it’s completely normal for a growth company to need to spend a significant portion of their revenues on selling their revenues, the general hope is that the product is able to reach an inflection point, wherein the sales and marketing expenses have paid off and revenues begin to ramp faster than SG&A expenses. Schrödinger hasn’t gotten there yet, likely due to intense competition within their market that forces a larger spend on selling and marketing activities.
Given that Schrödinger is still in the growth phase, profitability is not really expected of them yet, but being EBITDA positive is a nice indicator that a young company is heading towards profitability. Unfortunately, Schrödinger’s EBITDA losses have only continued to grow, ballooning to -$161.2m on a trailing twelve-month basis. Management sees no immediate end to these growing losses either, which leads into the next negative.
No Timeline for Profitability
Management doesn’t see any near-term realization of profitability. Though they did turn a positive profit for Q1 ‘23, this was largely due to the afore-mentioned one-off sale of a branch of Nimbus Therapeutics, which Schrödinger has an equity stake in. As for steadily maintaining this profitability, the business is nowhere near that point, and management has stated time and time again that they don’t expect this to change anytime soon, if at all, and that operating losses will continue to grow significantly in the coming years. They plan to continue pouring investments into software engineers and programmers, pre-clinical trials, and drug discovery processes for their in-house pipeline of drug development, research and development, and protection of their intellectual property. So, there’s no delusion here - this is a very long-term hold for Schrödinger shareholders, and extreme volatility and growing losses must be expected.
Balance Sheet - The Good
Schrödinger has very little debt on the balance sheet, with only $0.38 of debt for every dollar of equity. Now, you can take this one of two ways. I’ve chosen to label it under the good aspects of the balance sheet for a few reasons; because of all the other risks, I’m not sure that riskier long-term debt is necessarily another risk that needs to be added to the pile. Furthermore, a low debt-to-equity ratio will likely make lenders more willing to provide Schrödinger with money if they need it and can significantly reduce interest expenses, especially in a volatile interest rate environment like we’re currently in. Given that Schrödinger’s revenues can be very cyclical, I also believe a low D/E ratio is best for them so as to avoid getting caught with their pants down if they record a quarter or two with very few contract signings or drug discovery milestones aren’t hit.
However, you could argue, and I would find it very difficult to disagree with you, that financing a company on equity rather than debt can be more expensive, as shareholders generally want a higher rate of return than debt issuers for the added risk, possibly leading to a higher cost of capital. Also, if the company does require some financing and chooses to do it by issuing shares, it would lead to shareholder dilution. Personally, given the lumpiness of Schrödinger’s revenues and an already high risk profile, I’m going to say their lack of significant debt is a good thing here.
Not a Ton of Shareholder Dilution
Since Schrödinger’s 2020 IPO, the total number of shares outstanding has increased by only 2.9%. For a company dealing in software, which usually has exorbitant stock issuance for SBC (stock-based compensation), this is a nice bonus for shareholders. Stock-based compensation expenses hover around $10m in each quarter, which isn’t egregious and probably necessary to maintain Schrödinger’s high employee retention rates, and the fact that they haven’t had to issue a ton of extra stock to do this is encouraging.
Declining Cost of Revenues (Maybe)
On a quarterly basis, Schrödinger’s cost of revenues declined YoY in Q1, perhaps indicating that they’re beginning to hit some operating leverage. However, the decline was very small (-8%) and definitely needs to be monitored to see if this is a long-term trend or just a quarterly fluke; that same figure was up 5% sequentially from Q4 ‘22.
Key Performance Indicators (KPIs)
Annual Contract Value (ACV)
This metric determines the value of software contracts that Schrödinger realizes in a fiscal year. It’s not very different from annual recurring revenue, which is more typically the metric used for SaaS companies. However, because a portion of Schrödinger’s clients choose to host the software themselves and subsequently pay an upfront fee for a year or multi-year contract, ACV is a more accurate figure to follow.
Given that the majority of Schrödinger’s revenues are from the software, this is definitely one of the most important metrics to follow. As of the 2022 annual report, Schrödinger’s ACV was $141m, which they’ve grown at a 16% CAGR over the last decade. More importantly, perhaps, is the number of customers contributing over $100K in ACV. This is important to monitor due to the risky nature of the industry that Schrödinger and their clients operate in; a greater spend indicates the customer has a greater amount of spending capacity, which in turn indicates that they are a more stable and established player. The number of +$100K ACV customers is 227, a number they’ve managed to grow at a 12% CAGR since 2013.
Number of Customers
Also important is the total number of customers. Building out the software and further penetrating the market is crucial to Schrödinger’s success, so keep an eye on this metric to ensure they aren’t starting to lose customers or stagnate in growth. It also demonstrates the value of their offering over competitors if they can continue to grow the total number of customers consistently. As can be seen in the chart above, Schrödinger has 1748 total customers (they don’t count the peasants contributing <$1000), up from 742 in 2013.
This chart is slightly outdated, but I unfortunately couldn’t find an updated one with decent resolution. Still, it gives you a rough visualization of the steady growth Schrödinger has achieved in several of the KPIs.
Customer Retention Rate
While growing customers is important, retaining them is even more so. If the company is consistently losing customers, it would indicate to me that the value customers are getting from the software isn’t worth the investment, or that competitors were able to offer something better. Thankfully, and this is one of the strongest arguments for Schrödinger’s offering, their customer retention rate is a strong 96%. While this is down 2% from 2021, I think the most likely cause is simply that it’s a tougher macro-economic environment that has probably forced some clients out of continuing to purchase Schrödinger’s software or out of business entirely. However, if you choose to make an investment in the company, keep a very close eye on this metric (it is hands down the most important) and make sure that the decline doesn’t continue.
A steady slide on this customer retention rate would be a good sign to get the hell out and run for the hills. Given the huge value-add of the software, most pharma companies are going to spend the money on some type of software, so if the retention rates consistently fell, it would likely mean their customers were going to another provider . The significant switching costs would mean that a competitor would have to offer a significant improvement over Schrödinger’s platform to justify their clients leaving, and if that were the case, Schrödinger has been out-competed. If so, run - run away, and don’t look back.
Defense & Offense
Expanding ACV with Existing Customers - Offense
Deepening the value derived from existing customers is a runway for growth that Schrödinger has targeted to ramp up revenues and continue growing. Given that these customer relationships already exist, Schrödinger can benefit from this more organic growth without a significant extra investment in sales and marketing, with the added benefit that they can also skip the long sales cycle (anywhere from 9-12 months) by selling more license seats to customers they already have relationships with. Schrödinger has two main paths to help increase their ACV with existing customers: 1) making the platform more usable for interdisciplinary teams, ie expanding the use cases, so that a great number of people within an organization are able to use it and benefit from it 2) making the platform more valuable, so that companies want to buy more licenses/seats for the software.
While there’s no specific metric to measure their success with this, keeping an eye on the number of customers with higher ACV ($1m +) could give a good sense of it. Most customers are unlikely to spend more than a million in their first year with a product - they want to test it out and verify the value they can get from it before ramping up their spend only if it’s value-additive. So, if the number of customers with an ACV of $1m continues to grow, this is a good indicator that customers are satisfied with their first run with the product and are ramping up their ACV. Schrödinger had 18 such customers in 2022, up from 15 in 2021. Management also commented on this in the most recent quarter’s earnings call, stating that they were in talks to increase existing contracts with more than a dozen biopharma companies across the globe.
Improve Software & Foundational Science - Defense
Software is the core pillar of this business and will remain so as long as their drug development efforts are still in their early phases. Even if those efforts are successful, the software will remain a steady, non-cyclical revenue generator for Schrödinger. As such, continuing to improve the foundational science behind the physics-based platform will be crucial to offering a more competitive product to the market. Improvements to date have been: increasing the number and types of targets that Schrödinger’s platform is able to identify, increasing hit rates (essentially the percentage of compounds with the desired properties that the software can identify) in discovery, and enhancing predictive accuracy. All of this, in turn, helps to build a barrier against competitors and increases the value added for customers.
AI has taken a significant boost in the couple years, and in the last year most especially. While this has provided a big boost to the capabilities of Schrödinger’s platform, it’s also resulted in a significant improvement to open-source software and the ability of pharma companies to build their own in-house versions of the software. So maintaining an improved product over open-source offerings and those built by other pharma companies on the market will be crucial. The most important way to continue distinguishing their product is to ensure the algorithm at the core of their software and the science that underlies it all is accurate and constantly improving. This is their main form of defense, and it’s crucial that they maintain it and continue to build it if they are to find success as a company.
Complexity & Academic Employee Base - Defense
The pharmaceutical industry is incredibly complex. The underlying algorithm and science behind Schrödinger’s platform is incredibly complex. And, if you’re anything like me, science as a whole is pretty complex. Maintaining a competitive position in this field requires both attracting and retaining top talent and expertise in these complex fields, and Schrödinger has done a good job on both fronts. With 344 PhD-level employees and a 93% retention rate on their staff, this is a solid form of defense for Schrödinger. Competitors can’t just whip up a group of highly skilled workers like this on a whim, and creating a company culture that inspires them and keeps them around is even more difficult. So the sheer talent levels on display at Schrödinger will help to build barriers against competition, and will certainly serve as a high hurdle for newer entrants in the market.
Switching Costs - Defense
In addition to the complexity of the actual industry, the software is no walk in the park either. I don’t think I need to expound too heavily on this - it’s simply not a platform you can just pick up and start using. It requires a significant training and onboarding process for customers, and there’s a definite ramp-up period while the software is integrated into a company’s workflow. Once the time and money has been invested in a software, it would take a significantly better product to convince a company to basically flush that initial investment down the loo and start all over. This is likely a large contributor to the high retention rates for Schrödinger. However, switching costs are a shaky defense, as if Schrödinger gets complacent and there’s suddenly something much better available on the market, companies will make the investment to switch. So it’s key that Schrödinger continues to innovate, improve their products, and offer a serviceably competitive platform to protect their competitive position.
Flywheel - Offense & Defense
For businesses, a flywheel is a loop wherein activities feed into and strengthen each other, driving a self-reinforcing, upwards spiral that drives growth in continuous circles. As one part improves, it positively impacts the other parts of the flywheel and makes the entire system more effective over time. Flywheels can provide enormous organic growth for a business, and Schrödinger has a flywheel at the heart of its software and drug discovery collaborations.
As more molecules are correctly identified, this data can be fed back into machine-learning processes to improve the capabilities and accuracy of the software. Their collaborations with pharma companies can also help to improve the usability of the software and the underlying algorithms of the platform; by working in real-time with clients to identify areas of improvement for the software (as well as collecting more yet more data), Schrödinger can further upgrade the software to meet customers where their needs are. Here’s where the flywheel kicks in; as they participate in partnerships and license out their software to more companies, their base products get better. With better products, they can sell more software and enter more partnerships, and then improve products still further.
It’s a beautiful flywheel, and in such a complex field, small iterative improvements to their software is all the difference between their products and those of their competitors. Their flywheel can serve as a bit of offense and defense, helping to grow their competitive advantage over other businesses offering discovery software while expanding the number of customers and collaborations that Schrödinger caters to.
Schrödinger operates in a massive and steadily growing, high-demand market, and provides value-additive solutions to a set of extremely large and wealthy pharmaceutical companies, in addition to their materials science applications. Given the abundant value of Schrödinger’s customers and the market they serve, there’s some pretty intense competition for providing these solutions. Let’s dive first into just how big this market is before getting into the competition Schrödinger is up against.
While Schrödinger itself doesn’t release figures on the size of its total addressable market, the market for biosimulation technology serves as a decent proxy. Though only $2.4bn today, the industry is expected to grow at a nearly 16% CAGR to reach $7.6bn by 2030. I also think it’s handy to take a closer look at the wider industries in which Schrödinger operates:
Precedence Research estimates the drug discovery market to be worth over $60bn this year, and expects it to grow at a decent CAGR of ~9.2% through to 2032, which would be pretty stellar growth and a great secular tailwind for Schrödinger to benefit from. However, this isn’t their only market, so I also brought in data on the materials science discovery market, or Materials Informatics, as Precedence calls it.
This is definitely a much smaller, but also faster-growing market. Estimated at just short of $200m this year, they predict a CAGR of more than 26% through to 2030, in part due to tailwinds brought on by platforms like Schrödinger’s.
Finally, though I’m immensely reluctant to provide this, I’ll show the TAM for targeted therapeutics, which are the type of drug that Schrödinger is looking to develop with their own drug discovery portfolio. The reason I don’t really want to provide this is that they haven’t proven any ability to make money in this market yet. Still, for the sake of thorough research, here it is:
Pretty impressive market here as well, though the growth in this market is much slower, with just a 2.1% CAGR. If they are able to eventually take advantage of this market, this would be a pretty significant boost to their total market, but again, this is as yet unproven.
While Schrödinger, founded in 1990, has a fair bit of experience under their belt, they are definitely not the only ones in their space. It’s a pretty fierce competitive landscape for providing this computational discovery software, making it all the more important that Schrödinger continues to reinvest in expanding the capabilities of their platform. While management believes, and they think the wider industry does as well, that they are the “gold standard” in the space, it’s still important to understand the competition and the threat they pose. Due to the number of competitors, I won’t go super in-depth on each one but provide just a quick overview of who they all are.
Publicly Traded Companies:
BIOVIA - Dassault Systemes
Cadence Design Systems
Chemical Computing Group (US) Inc.
Cresset Biomolecular Discovery Limited
Cyrus Biotechnology, Inc.
Insilico Medicine, Inc.
Simulations Plus, Inc.
Out of these companies, there are a few different strains. There are some trying to deploy an exclusively machine-learning based approach, and there are university programs that are maintained by grad and post-doc students. Others are massive businesses with strong balance sheets and ample resources, both financially and in human capital, to build out more advanced software platforms. Simply put, these larger businesses have the capacity to outspend Schrödinger to build the best product on the market. Schrödinger also faces competition from pharma companies that build out their own in-house software platforms, and companies that focus on smaller market niches. Between all of these, Schrödinger is facing a lot of competition. Unfortunately, I don’t see this as a market where multiple companies can succeed - the best product will win out by providing the highest accuracy while saving the most time and money. There is simply no other reason to go for a lesser product when the return on investment is higher with a better platform.
There will inevitably be a winner in the space, but it all comes down to who can employ the most resources towards building the most capable platform with the largest array of use cases, from end-to-end, and who has the most advanced underlying science to combine with machine learning. For this last reason, I don’t believe many of the smaller companies trying to deploy exclusively AI/ML-based solutions are going to pose much of a threat, but larger companies that have the scientific resources to compete with Schrödinger’s highly skilled team definitely pose a significant threat.
Schrödinger is an incredibly difficult company to model, even for my back-of-the-napkin valuation methods. I see it going one of two ways, and have built projections off both possibilities. One is that the drug discovery segment of the business begins to pay immense dividends for top-line growth and profitability, and several of the molecules Schrödinger designs in partnership with Bristol-Meyers Squibb are commercially viable - demonstrating both the immense power of the software, earning a ton of revenues, and expanding brand awareness within the industry. The other takes a gloomier outlook; the drug discovery segment never really takes off to the level that management is hoping - it continues to be a drag on total gross margins, and growth is limited to the software’s industry penetration.
The Bull Model
I stayed relatively conservative with the success of the BMS partnership for this model and assumed that the collaboration only resulted in a couple commercially viable novel molecules being identified, and royalties only starting to kick in for 2027. The sales growth is quite bullish based off an assumption that software penetration and capabilities will continue down their significant runway, and that other drug discovery milestones will contribute to top-line revs.
I assume a 24% sales CAGR through to 2028, declining very slightly from 2023-2025, then ramping up marginally as the BMS royalties kick in. This would see the company making ~$680m by 2028.
I assume an exit EV/S of 8x (it’s 11x at time of writing), and assume that the market is favourably viewing the levels of growth and the success of the Bristol-Meyers Squibb collaboration, and that Schrödinger is able to turn a small profit at this point, or are at least EBITDA positive.
The end result is a $5.4bn enterprise value, representing a 19.7% CAGR. Schrödinger operates on a very frugal balance sheet, with no long-term debt on the balance sheet - if nothing happens to change that, then the market cap would actually be slightly higher than the enterprise value (meaning slightly better returns for shareholders).
This far outweighs my hurdle rate of 10-15%, but it definitely isn’t the only path this company can take.
The Bear Model
I assume a 22% sales CAGR that would have the company earning $623m by 2028, based off a fairly steady 2% decline in revenue growth every year.
I assume an exit EV/S of 4x, based off slowing top-line growth and the failure of the drug discovery segment to meaningfully contribute to continued expansion for Schrödinger.
Under this model, Schrödinger has an enterprise value of $2.5bn, representing approximately 1.8% CAGR through to 2028. This would be a dog investment under this scenario, further demonstrating the need for this company to do well on their collaboration with Bristol-Meyers Squibb.
Both the bear and bull cases here are possible, I believe. Undoubtedly, my exact figures are dreadfully wrong, so please don’t consider them if you are making any decisions on the company. I simply provide them for a rough visualization of the different directions the business could go. It truly all comes down to their success with BMS - if it goes well, they will earn a significant amount of revenues, boost their brand awareness, and likely receive some tailwinds on their software sales, as well as likely enter into many more drug collaborations. If their collaboration doesn’t go fantastic with BMS, however, this investment will likely trade flat. Software sales will likely continue to grow, as will revenue growth as a whole, but the drug discovery will fail to really ramp up to any considerably profitable level, and the market will punish Schrödinger for the subsequently prolonged operating losses.
The ultra bear case here is that one of Schrödinger’s competitors completely out-competes Schrödinger, and revenue growth falls off a cliff. If that were the case, this stock is likely done for.
Not looking bad for Schrödinger on the analyst front; based on data from Nasdaq, analysts are viewing the stock as a ‘Strong Buy’. Here are the potential returns based on the 1-year price targets:
Low: ~ -12%
Average: ~ +61%
High: ~ +126%
Now, these are just 1-year targets, and I’ll make my usual disclaimer that analysts can be just as wrong as the next human. It’s also not unanimously positive for the stock, with a short interest of 11%. This is a pretty significant amount of betting on a continued downturn in the share price, but shorts are also not very aligned with long-term holds, which is definitely the mindset that investors in Schrödinger need to have.
Risks to Share Performance
Phew. This is going to be a pretty long list, so I’m going to give it to you in bullet form instead, at the risk of blabbing on to you for the next 5481 words about everything that could go wrong and hurt the share price.
Drug discovery partnerships are never commercialized or don’t hit milestones, limiting growth in this segment
Timing on these revenues could also be delayed, which could lead to more short-term
May never realize returns on investments for in-house drug portfolio, or inexperience in clinical development shows through
Fierce competition in the space begins to commoditize the product or outdate Schrödinger’s software
Schrödinger’s rich valuation could get hammered if they fail to live up to the seemingly perfect expectations that are priced in
They don’t find success in their partnerships/collaborations, which damages their reputation and subsequently the number of customers that want to sign contracts
Founders not as involved as I’d like to see
Not much evidence of operating leverage
Questionable capital allocation decisions
Super competitive space, with many larger companies (with more resources) building out platforms
Significant investment from Bill Gates and collaboration with Gates Ventures
High retention rates for both employees & customers
Continued investment into most crucial part of business (software) and a focus on maintaining a competitive edge in this segment
The Short Story
Software like Schrödinger’s is, without a doubt, the future for the drug and materials discovery processes. The value that software like this adds, the time and money that it saves, the side effects it negates, and the increased likelihood of finding the right molecule it drives; it all adds up to an entirely changed discovery process. The increased capability of AI and machine learning will only accelerate this advancement and its subsequent adoption. I see no world where software that is able to leverage physics, chemistry, and machine learning is not widely adopted across every company undergoing discovery phases for new drugs and materials. The importance of continuing to develop new products in these markets, too, cannot be understated.
I remain steadfastly bullish on the future of this brand of technology: on the good that it will allow in the world, on the advanced and environmentally-friendly materials it can help to build, and on the life-altering medicines that it will help to develop. These products will continue to improve the world around us, and these kinds of technologies have the chance to be the value-additive pillar at the foundation of them all.
But that doesn’t necessarily mean Schrödinger will be the company to do it. Or that, if they are the ones to do it, the company will make for a good investment. Many good businesses makes for POS investments. And Schrödinger has a long way to go before they can prove to investors that they are a safe play - if they can at all. Schrödinger is still very much in a risky growth phase, and the 2018 addition of their own drug discovery process throws another risk-oriented wrench into what was otherwise a mostly steady, reliable stream of revenues from their software. They also operate in a very risky and cyclical industry, and many of the drug candidates from their partnerships and from companies they have equity stakes in are likely to fail, resulting in completely sunken costs that will make it difficult for this company to realize solid returns on investments. Finally, the technology and science at the core of their offerings is incredibly complex, making it very difficult for most investors to get a good sense of what their competitive edge is; any investors that aren’t experienced physicists themselves would likely fail to appreciate the significance of an improved algorithm from a competitor, and would only recognize the disruption when revenues and share prices were already in decline. All told, Schrödinger has not demonstrated its ability to execute on its lofty goals and attractive-sounding story, is difficult to understand, and yet is trading at lofty valuations despite all this.
So, make no mistake; Schrödinger is a very, very risky play. It could represent a huge payout for investors who are willing to stay patient and handle huge amounts of volatility. It could also fall flat on its face and lose investors all their money, so anyone interested in investing in Schrödinger has to be completely aware of that possibility. If you can’t handle volatility or wildly swinging stocks that will keep you up at night, this stock isn’t for you. If you like stable, compounding businesses with fortress balance sheets that are going to reliably produce a buck for you, this company isn’t for you.
Ultimately, I think investing in Schrödinger comes completely down to your risk tolerance, desired returns, and investment strategy. If you are interested in investing, I would recommend Q2 or Q3 as a nice entry point, as these are historically their lower quarters, but to be frank, this isn’t a company most of the great investors would ever even consider. I follow most closely the Peter Lynch style of investing, and I can tell you for a fact he wouldn’t touch this stock with a ten-foot pole. It has many of the hallmarks of a terrible investment, though a lot of promising features and a tempting story as well.
With that said, let’s hand out Schrödinger’s final report card:
Final Grade: B-
And that’s all for Schrödinger! Let me know what you think about the company in the comments section, and we’ll be back with another article before you know it! We will be covering WSP Global next, a homegrown (Canadian) rollup of engineering and consulting firms that’s producing steady and reliable results while organically growing their acquired operations at impressive clips.
This article on Schrödinger is one of my last articles, for now at least, on exciting but unprofitable companies with enormous upside potential. It is time to turn towards my basket of Peter Lynch-esque stocks (boring, underappreciated by Wall Street, and growing steadily) that are far more inline with my own investment strategy, and WSP will be the first of these. Subscribe below if you want that report sent straight to your inbox, as well as podcasts on investing topics and weekly newsletters to boot!
If you made it this far, I appreciate you! But if you pump your email in below and punch that subscribe button, I’ll appreciate you even more!