A Computational Chemistry IPO From Schrodinger
Table of contents
Elon Musk once said that there’s a one in a billion chance we’re not living in a simulation. Sounds far-fetched until you start digging into simulation theory a bit. It’s backed by theories from leading academics, like Nick Bostrom’s “simulation argument,” which presents three likelihoods, each equally frightening to ponder. To paraphrase Arthur C. Clarke, it is as fascinating to think we’re alone in the universe as it is to think we’re not. What makes simulation theory even more credible are the recent advancements being made in modeling real-world biological entities. Companies like Turbine.ai are actually simulating cancer cells in virtual environments, then testing the accuracy of their efforts by mimicking wet lab tests. It’s similar to computational chemistry, a branch of chemistry that uses computer simulations to assist in solving chemical problems. Today, we’re going to look at a computational chemistry company that just filed for an IPO.
Schrödinger and Computational Chemistry
Firstly, let’s distinguish between “computational drug discovery” and “computational chemistry.” There are now upwards of 120 startups out there working on various aspects of “computational drug discovery.” This means they’re using technologies like machine learning to make improvements across the entire drug discovery process. “Computational chemistry” also falls under this label, but specifically refers to the use of computational methods to optimize a molecule once a target has been identified. What Schrodinger has developed is a physics-based computational platform that is capable of predicting critical properties of molecules with a high degree of accuracy. It also uses artificial intelligence to explore up to billions of molecules within a few days.
It would be extremely difficult to realize a competitive advantage in a drug discovery program by using a platform exclusively based on machine learning or deep learning. Instead, Schrodinger has developed an approach to integrate physics-based and machine-learning-based scoring methodologies that allows the machine learning model to interactively prioritize additional molecules for physics-based analyses, known as active learning.
Similar to Turbine, Schrodinger is able to run these simulations with such a high degree of accuracy that they’re approaching that of a lab experiment.
In a recent peer-reviewed study including approximately 3,000 molecules across approximately 90 distinct projects, FEP+ exhibited an error profile that indicates its affinity predictions approach the accuracy of running a laboratory experiment.
The strength of this platform is even more evident in its users and how dependent they are on it.
Selling Software
In looking at the S-1 filing for Schrodinger, they largely install their software on location for their clients which means the revenue is recorded differently. In other words, it’s not a Software-as-a–Service (SaaS) business model that’s easy to understand. Still, the principles of selling software remain the same regardless of how the revenue is being recognized. Investors in such businesses like to see high retention rates (subscribers continue to renew their subscriptions), flagship customers on board (reference clients who can help you bring on new clients), and increasing customer expenditures as time goes on (something Schrodinger refers to as annual contract value or ACV). Schrodinger checks the boxes in all three cases.
The ability for Schrodinger to increase run rate – or ACV – within their customer base can be seen by the increasing number of customers they have with an ACV in excess of $100,000.
- 2016 – 87
- 2017 – 103
- 2018 – 122
In addition to growing their ACV, retention rate for existing customers with an ACV over $100,000 was over 96% for the year ended December 31, 2018, and for each of the previous five fiscal years as well. There’s also no shortage of reference clients using the platform. In 2018, all of the top 20 pharmaceutical companies, measured by revenue, used Schrodinger’s software, accounting for $22.0 million, or 33%, of their 2018 revenue, and these companies have been their customers for an average of 15 years. If you’re an investor looking at these numbers, understanding what the software does is completely irrelevant. This is a product that isn’t just a “nice to have”, it’s something that companies have to have. As long as the retention rates remain at that level and the ACV consistently increases over time, this is a healthy business to own. And the subscription offering is just one piece of the company’s potential.
Internal Drug Development
If you develop software that helps other companies bring drugs to market, at some point you may think about doing some of that work yourself. In mid-2018, Schrodinger began conducting their own internal drug discovery efforts which have led to the below pipeline:
There are several trains of thought here as to whether or not they should develop their own drugs on their platform – or as some like to say, eat your own dog food. On one hand, they do have the expertise on staff with over half of their 400 employees having PhDs along with “an in-house drug discovery group comprised of a multidisciplinary team of approximately 70 experts in protein science, biochemistry, biophysics, medicinal and computational chemistry, and discovery scientists with expertise in preclinical and early clinical development.” On the other hand, perhaps they ought to stick with what they do best – developing software. After all, drug development is an expensive process and prone to lots of unpredictability which leads to lots of share price volatility.
Collaborative Drug Development
Maybe the best way to move forward is to have some skin in the game while letting someone else take all the risks. Schrodinger is currently collaborating on more than 25 drug discovery programs with more than ten different biopharmaceutical companies, including a number of companies they co-founded. Here’s a list of all the companies where they presently have some meaningful skin in the game.
The nice thing about these equity stakes is that if they never pan out, no harm done. For example, Nimbus Therapeutics was looking quite promising in 2016 when they sold their drug program to Gilead in 2016 for $1.2 billion. (Schrodinger actually co-founded Nimbus back in 2009.) Then, things went south last month when the drug’s primary endpoint wasn’t met. Still, Schrodinger managed to score $46 million in cash off the sale of Nimbus with another $46 million coming to them – maybe. Then, you have companies like Morphic which Schrodinger founded in 2014 and which is publicly traded now under the ticket MORF. All of these equity stakes provide Schrodinger with an opportunity for future payouts based on the success of these drug programs. Lots of upsides and no real downside. (This is what Intrexon was trying to do with their ECC business model before they pivoted into the lucrative business of selling sliced apples.)
Conclusion
A great way to build brand awareness is by getting your product in the hands of future customers. More than 1,250 academic institutions across the world used Schrodinger’s software in 2018. The company has also extended their computational platform to materials science applications in fields such as aerospace, energy, semiconductors, and electronic displays. A growing number of materials science customers are now using the platform and Schrodinger believes “the materials science industry is only beginning to recognize the potential of computational methods.” Beyond drug discovery, there’s an entire swath of industries that might find the platform equally useful. Since they already have some experience in setting up companies, maybe the next step will be for Schrodinger to setup industry verticals which can diversify revenue streams so they’re not overly reliant on the pharma industry.
If the IPO happens as planned, Schrodinger plans to list under the ticker “SDRG” on the Nasdaq exchange.
Schrodinger has a solid value proposition and a convincing business model. Did we decide to invest in them? Find out about the complete list of disruptive tech stocks and ETFs we’re holding in the “Nanalyze Disruptive Tech Portfolio Report,” now available for all Nanalyze Premium annual subscribers.
Share
The company’s software solves Schrödinger equation using certain approximations available in the scientific literature or developed by the company.
Not sure about that but we’re seeing zee Germans did in an academic setting, or came close to it, or something like that. Sometimes news like this is a bit of a cry wolf.
Another interesting company having some similarity to Schrödinger: XtalPi.
XtalPi is an AI drug R&D company that uses computational physics, quantum
chemistry, and cloud computing to provide intelligent drug development for
pharmaceutical companies. The company’s ID4 (Intelligent Digital Drug
Discovery and Development) platform can accurately predict the important
characteristics for new drugs by measuring various chemical indicators,
enhancing drug R&D efficiency. ID4 has enabled hundreds of pharmaceutical
companies to accelerate their research pipelines, leading to dozens of new
discoveries.
Yes, they are an interesting firm that just took a $400 million round a few months back: http://nanalyze.donebox.hu/2021/04/companies-ai-drug-discovery/
Credit: Analytics India Magazine
Article: AI Helps Solve Schrödinger’s Equation — What Does This Mean For The Future?
An AI model helped solve Schrödinger’s equation with great accuracy, which opens up the potential for a lot of future applications
Scientists at the Freie Universität Berlin have come up with an AI-based solution for calculating the ground state of the Schrödinger equation in quantum chemistry.
The Schrödinger’s equation is primarily used to predict the chemical and physical properties of a molecule based on the arrangement of its atoms. The equation helps determine where the electrons and nuclei of a molecule are and under a given set of conditions what their energies are.
In principle, the Schrödinger’s equation can be solved to predict the exact location of atoms or subatomic particles in a molecule, but in practice, this is extremely difficult since it involves a lot of approximation.
Central to the equation is a mathematical object, a wave function that specifies electrons’ behaviour in a molecule. But the high dimensionality of the wave function makes it extremely difficult to find out how electrons affect each other. Thus the most you get from the mathematical representations is a probabilistic account of it and not exact answers.
Cheers for the interesting news!