Later today, I’ll take the stage with Saskia Steinacker, the global head of digital transformation for Bayer, at Web Summit here in Lisbon. Web Summit is the most prominent tech conference in the world, where I am joined by 70,000+ of my closest friends who are all connected by one thread - technology. Whether using technology to advance any one of hundreds of industries as an entrepreneur, writing about technology - the good and the bad - as a journalist, or otherwise, the energy here makes clear that the future is tech (and that software will eat the world).
Fast forward 8 weeks when I will fly to San Francisco for the JP Morgan Healthcare conference, where a similar number of people from across the life sciences and healthcare industries converge on Union Square (in January, no less), for a week of diving deep to build relationships across healthcare. When I first went to this event, almost five years ago, there was almost no talk about sophisticated computational tools ‘disrupting’ healthcare, outside of perhaps those being used in computational chemistry. Today, nearly every major pharma company has a ‘Chief Digital Officer’ or similar and is driving money and resources, rightly or wrongly, into technology across many layers of their own companies and through partnerships.
As the CEO of one of the most prominent companies working on AI-enabled drug discovery, I (often inelegantly) straddle the worlds of tech and biotech. And so on my way to Lisbon, I read with deep interest David Shaywitz’s new piece (and his first after his transition from being a contributor at Forbes to the Timmerman Report), titled: “Tech VCs and Biotech VCs: Talking Past Each Other on AI Drug Discovery.”
In his usually gripping style, and with real credibility as someone who has studied this space as closely as anyone in biotech, David lays out what those of us operating at the intersection of technology and life sciences have known for a while.
Tech VCs generally believe that any industry in the world can be disrupted, and that sophisticated new computational approaches will eventually do just that. They also believe that large, old incumbents, of which there are many in biopharma, represent an exciting opportunity to create true disruption (and new incumbents). Like Uber and Lyft transformed transportation, Tesla the future of automobiles, Netflix the entertainment industry, and countless companies making space more accessible, tech VCs see the largest pharma companies as akin to large auto manufacturers and cable companies a decade ago, sitting vulnerably and acting slowly. And the reason biopharma hasn’t been disrupted yet? Biology is hard and healthcare is regulated, so it makes sense that this industry is one of the last to fall, but fall it will.
On the other side are the biotech VCs, many of which have first-hand experience with developing drugs and other forays into healthcare. They know how hard it is - a miracle by some accounts - to bring new innovations to patients. They know that the vast majority of drugs fail before they ever even see a patient and they have witnessed momentum and $100s of millions of investment and the hopes of patients in some cases wiped out by the late-stage clinical failure of an experimental new medicine. They have also heard claims of new technology disrupting healthcare before, often to under-deliver in the years and decade after those hopes are first promised. The genomic revolution of the 2000s, which gave us the sequence of the human genome, then 100s of genomes, and today millions of genomes, created a sense that it would change healthcare in a disruptive way. Similarly, the computer aided drug discovery (CADD) promise of the 80s and 90s was hailed as the next industrial revolution by Forbes. Both of these technologies have materially changed the way drugs are discovered and developed, but the rate of change to the industry due to these advances has failed to meet expectations. The risk profile of these investors, and their hesitancy to jump two-footed into the high valuations that correspond to the tech VC's promise of the future, is not surprising (read a perspective from Bruce Booth of Atlas Ventures, one of the most outspoken and well known biotech VCs). All the recent hubbub about massive tech-VC driven companies crashing (a la WeWork) at the hands of crazy founders, and 'disruptive' companies like Uber and Slack seeing shaky IPOs rightly bolsters biotech VC's desire to tread with caution.
As someone who has raised a considerable sum of money from both types of investors, and gotten to know many thoughtful people on each ‘side’ who have passed on an investment, what jumps out to me as the most consistent thread that differentiates tech and life science VCs is their respective willingness to project new data onto an industry and imagine resultant changes happening relatively quickly. Take the DDR1 paper by Insilico, where they discovered a new class of DDR1 kinase inhibitor using Generative Adversarial Networks (GANS) - it was picked apart by those on the biotech side: "this is a well known kinase;" "they had lots of good training data, which isn't usual;" "what they discovered isn’t that distinct from what others have," etc, while folks on the tech side of the fence saw this as a first step, imagined that Insilico is likely to be at this moment working on much more sophisticated projects (for which they likely and rightly will not publish or provide source code), and that in 5 years this technology could be game changing. Just today, a leading company in this space, Atomwise, reportedly completed the largest virtual drug screen in human history, and yet these steps, absent proof of concept data in human patients, is discounted severely. There is no right or wrong here: the life science VCs are probably going to be right on average - many companies in our space will fail to create the revolution they seek to. But betting boldly against technology to change biopharma, let alone any industry, in big ways in the face of the 4th industrial revolution - computation - could mean that life science investors will miss out on one of the biggest shifts their industry will witness in their lifetime. As David describes, the life science investors are asking: “where’s the beef?” before investing, while the tech VCs are willing to invest based on the smell and sounds from the kitchen.
David also talks about the business model (or lack thereof from some perspectives) of these tech upstarts and the bet that tech VCs are making that, as early signs of the ‘beef’ in AI-driven discovery come to light, one visionary and innovation-hungry biopharma CEO might make the plunge and buy one of the most prominent AI-driven drug discovery companies, starting a wave of acquisitions. This in turn will realize both the need of pharma to innovate and the returns of tech investors who bet early.
In my estimation, David only misses one key point, and it is one I and many of my tech investors personally hold dear. That is not to build a company or technology that is unproven but plausibly exciting, create FOMO and then sell to the highest big-pharma bidder at an attractive multiple, but instead a multi-decade commitment to the creation of the first truly tech-enabled independent verticalized biopharma, from target discovery through early clinical development. By tech-enabled, I don't just mean a company using a bit of technology at each stage, as is commonplace today, but the truly integrated development of a soup-to-nut process consisting of multiple integrated technological solutions substantially improving the cost, time or quality required for the discovery of new drugs. This is an important position taken by a handful of companies in our space, and for biopharma and both tech and life-science investors to recognize. As the chefs creating the companies at the intersection of technology and life sciences, we have the closest seat to the smells, sounds and tastes of what we are cooking. And while we may sell the food, who says we will ever sell or share our recipes?
At Recursion we have put multiple drugs into human clinical trials, new uses of which we discovered using our machine learning systems applied to the largest relatable biological image dataset in the world, which we continue to build in-house from scratch. With many more drugs on the way to the clinic, and elements of our machine being applied (and increasingly integrated) to target discovery, drug discovery, computational chemistry and predictive pharmacology, there is a sense that with a lot of work, and with partnerships with the biopharma industry by our side, we really can build in a way that both wasn't possible 5 years ago thanks to advances in computation, automation and data storage costs, but also in a way that large pharma will be hard-pressed to replicate independent of their will to do so.
Our progress and claims would be easy to discount if the entire company was as admittedly inexperienced as I am in drug discovery (after-all, I started the company straight out of graduate school), but we have nearly 1,000 years of true industry-honed drug discovery experience on our team among the more than 50 biologists, pharmacologists and chemists working daily to advance our mission. These experienced drug hunters are working side by side by data scientists and software engineers to create an integrated operating system where all data, positive and negative, created by the company will be relatable and searchable, and deployed on top of that integrated dataset, algorithms are already discovering real, robust and reproducible patterns that no human can see. Don't believe me? Check out our recent Kaggle competition and the forum where several teams sent humans head to head against algorithm and were painfully defeated.
The biotech business model that's 'lacking' according to biotech VCs interviewed by David, isn't all that different from what already exists in the industry: companies like us will discover new drugs or new uses for drugs and get them to patients through a regulated system ourselves or with partners. Using technology, we believe we can eventually discover more drugs, faster, with lower late-stage failure rate (and by design, higher early-stage failure rates) and create a system and process to do it repeatedly. Amazon still sells books to humans who read them - just like bookstores did. But they also expanded that model across new areas and industries, and I am excited to imagine the ways the next generation of technology-enabled biopharma companies and synthetic biology companies will similarly build broadly.
So as we transit the coming years, and as many of the narrowly-focused AI companies, with little biology or chemistry expertise, start hitting the dust, there will be lots of “I told you so” comments coming from biotech VCs. But there are likely to be a few breakout companies aiming not for a quick flip to big-pharma or to create an incrementally impactful company aimed at a single step in the drug discovery process, but instead dedicated to building the next great biopharma company from scratch, alongside revolutionary advances in technology and hundreds of amazing colleagues similarly dedicated and passionate towards that mission and the impact they can have on the world through their success.