The Creation Of Biotech Startups: Evolution Not Revolution

Posted August 15th, 2019 in Biotech financing, Biotech investment themes, Capital efficiency, Drug discovery, Talent

The startup paradigm for the creation and funding of new biotech companies has evolved enormously over the past two decades. Recently there’s been a pair of articles from Tech VCs about applying alternative models of company creation (here, here) to the world of biotech, so thought it was worth reflecting on the prevailing venture formation landscape and offering up some counterpoints.

Back in 2015, I shared my perspective on five changes to the biotech venture model since 2005 – and to a large extent those same dynamics remain at work today. I’ll start with a quick review for context.

When I first started in the venture business fifteen years ago, founding biotech entrepreneurs would often prepare pitch decks and then shop them around to different VCs as part of the typical “dog and pony” show. Sometimes those were academic professors or post-docs, sometimes they were entrepreneurs leaving existing biopharma jobs to do the startup thing. They’d often propose to build or lease a standalone bricks-and-mortar lab, hire 15-20 scientists, and advance an academic observation into bona fide drug discovery during the Series A round. Sometimes it would take $15M or more just to figure out whether the initial academic insight was reproducible and generalizable as a therapeutic approach. Often the initial premise wouldn’t survive these killer experiments, but because “real” capital had already been invested these companies would frequently try to “pivot” to other programs or approaches. Sometimes, albeit rarely, these pivots worked in biotech. But with fixed infrastructure and sunk cost, it was hard to just walk away. Returns suffered in this model.

Our 2006 vintage “prove-build-scale” model of biotech investing emerged from this world and was an attempt to change the paradigm. During the “prove” seed phase, the aim was to demonstrate the scientific robustness of the founding concept as cheaply as possible. This model was largely enabled by the emerging “virtualization” of the ecosystem. Virtual drug discovery startups that leverage CROs and partners from around the world, assembling expertise on an as needed basis, became a realistic operating model about a decade ago. With the rise of biotech wet-lab launchpads like LabCentral, hybrid models with both in-house labs and heavy virtual outsourcing emerged, and are a common feature today. For novel areas of biology, some internal lab footprint is often critical.

Enabling technologies certainly helped (and continue to help) create the virtual infrastructure that powers up many biotech startups: computational drug design with partners like Schrodinger, in vitro safety screening, cheaper DNA sequencing and bioinformatics, lab automation, etc… Even virtual team management platforms like Slack for global remote project efficiency have helped.

I’d also argue the post-merger dismantling of Big Pharma’s R&D footprints has had as much to do with the rise of virtualization as these enabling technologies: when large and small R&D sites around the world were shut down to reduce R&D footprints, many of these teams/facilities formed into new specialty CROs or were absorbed into existing ones. Europe and the Midwest have a huge number ex-Pharma sites that became CROs. Further, the massive expansion of offshore medicinal chemistry services as China and India liberalized their business approaches also played a key role.

The venture-backed biotech ecosystem was embracing this virtually-enabled drug discovery model a decade ago, and that trend has just accelerated since – and will likely continue to.

We’ve evolved our strategy at Atlas over the past 15 years to what we call our seed-led venture creation model. I’ve written much about it in the past (here, here) so will quickly sum it up. During the setup and creation “prove” phase, we partner with our scientific and business co-founders to create new companies. We generally write small checks initially (from $100Ks to a few $1Ms) to do the seed-stage validation work, deliver the killer experiments, or assemble the right science/IP. But we also look for other signals – like whether the entrepreneurs involved in the startup have the savvy execution and strategic thinking required to launch and scale a young drug company (i.e., a “talent signal”). Another important variant of that talent signal is if experienced skilled entrepreneurs raise their hand to want to put their careers behind the idea during the seed phase. We also socialize the concepts with downstream Pharma and future funding partners during the seed to characterize the market interest in the approach (i.e., a “market signal”). This seed-led approach, and integrating these various signals, helps to bend the risk curve in our favor by weeding out the ideas/startups earlier with a high degree of signal stringency.

This in-house venture creation model helps bring the right elements together that were hard to assemble in traditional “pitch-deck” venture formation: rigorous and objective exploration of the founding science, enough capital to do the key early work, and, most importantly, the ability to attract and retain the right talent. Experienced drug R&D veterans and senior business talent can often better stomach the career risk of jumping out of bigger companies by joining not just one isolated high-risk startup in a random office park, but by joining an in-house startup community of other entrepreneurial executives within an incubator-like model.  Most are mid-career by definition – in order to have assembled the requisite experience set. To align interests and motivations, founder-EIRs know that if the scientific premise doesn’t reproduce or validate, that we’ll do well by them – often by recruiting them to other projects. As company creators, we often shared a piece of the founding stock pool, which financially aligns us with the entrepreneurial team (as our ownership goes down, not up, with larger capital raises and lower valuations).

Today there are many flavors of this in-house venture creation model at different firms: some more artisanal and bespoke, some more systematic and formulaic. There used to be only a few firms doing this type of work 10 years ago – but the venture creation model has delivered stellar returns in the last decade for founders, employees, investors, and patients. We went from tombstones hailing the death of life science venture capital in 2011 to being one of the star-performing sectors in the venture capital industry. And with that success, today we’re seeing lots of VC firms talking about doing their version of in-house company creation. And seeing lots of tech VCs move into biotech. As they say, imitation is the sincerest form of flattery.

But as the biotech venture ecosystem continues to change, we will need to continue to evolve our model, of course.

A big change today is that the world is awash in capital. The tsunami of funding into private biotech firms in the last few years has led to the rise of the Series A mega-round: go big or go home. Some of these have been successful, but many others won’t be. Along with more funding at the company level, VC fund sizes have also gotten bigger across the board, which means more capital has to be put to work per deal, worsening the venture capital math problem. Series A sizes are way up, as are later rounds, and it seems like every week there’s another $100M+ mega-round. There’s also just more capital going into the same number of startups, in general. According to Pitchbook’s data, the number of therapeutic biotech firms getting their first financing remains largely flat for the past 5+ years despite overall venture funding numbers being 2-3x greater.  I’ve written on this paradox many times in the past (here is one example), which differs greatly from the tech space where the number of startups exploded with more fund flows. This concentrated allocation of capital is potentially troubling, as it could presage that smart de-risking decisions aren’t being done with discipline. Over-funding and crowding in ‘hot’ spaces like I/O and gene modification will likely dampen returns. I worry a great deal that in a world of capital abundance we will lose stringency as a sector. Two truisms come to mind today: more startups have died of indigestion than starvation; and, the average fitness of the herd goes down with abundance. We are at risk for this in biotech today.

At Atlas, we’ve tried to stick to our capital efficient startup model of seed-led venture creation, augmented by what we think is an optimal fund size configuration, without being dogmatic or formulaic about it. Some science deserves to be in an asset-centric focused startup, and other science requires a full and expansive platform build. We added an Opportunity Fund to power up our more capital intensive companies, and to secure additional positions in follow-on financings. But in general we’re still starting companies that have a “prove” seed phase: whether it’s a standalone seed investment, or a first “seed-tranche” of a more significant Series A, we remain firm believers that de-risking should be done before a large amount of capital gets deployed. It’s the essence of the value-creating, positive aspects of equity capital efficiency.

In the past month, proponents of a “new” model of biotech startup creation have opined on the subject: tech entrepreneur Jared Friedman of Y-Combinator (here) and Jorge Conde of Andreessen Horowitz (here). They both propose to apply elements of the tech VC approach to the world of biotech.

As a matter of first principles, I welcome their engagement in the biotech ecosystem (and that of other tech investors) as I’m sure there are things we can learn from alternative approaches, and vice versa. And I truly hope they are successful – which means more innovative drugs will make it to patients, and that’s a good thing.

Having been in a diversified venture fund for nearly a decade (before we went biotech-only in 2014), I appreciate the cultural value of the cross-fertilization of ideas. Prove-Build-Scale was a mantra both the tech and biotech side of Atlas embraced for years. But I also appreciate how vastly different our ecosystems are and caution against naïve assumptions to the contrary.

Let me start with what I agree with in Jared and Jorge’s approach: I fully embrace the value of seed-stage de-risking of novel therapeutics companies (the essence of our 15+ year biotech strategy) and how the ecosystem and enabling technologies are continuing to support virtual and efficient biotech operating models. This is certainly true, and is at the heart of our investment model.

But I’d like to share a few counterpoints.

  1. The cost curve of drug R&D hasn’t meaningfully changed with virtualization or new enabling technologies. Drug R&D is hard, and the typical program costs a lot to bring it through initial hit finding, hit-to-lead, lead optimization, preclinical, and early clinical testing. Those phases – what I would call drug discovery and translational research – cost on the order of $25-50M per program, take 5+ years, and face high failure rates (with expensive false positives accruing spend over time). Sadly, more virtual R&D models, even those enabled by computation, in our experience aren’t fundamentally cheaper or faster over this entire multi-year translational period. However, they are often more strategically flexible to respond to changes in the program (the benefit of variable vs fixed costs), and that flexibility allows for less expensive “down time” as well as earlier kill decisions on programs (and companies) than in prior biotech operating models. Unfortunately, as I’ve written before, translating novel biology into real drugs is complex and still consumes everything else in the R&D process; it’s the biology that takes so long and costs so much to unravel. While the cost to get a startup formed and initial experiments going may be cheaper than it was 15 years ago (when you had to build your lab), that’s generally not the big cost over the 5+ years of moving from taking a “hit” into patients. And, frankly, we’ve been doing this model for a while: it’s not a novel idea to do lean seed-stage experimentation in the process of starting biotech companies.
  2. Seed-stage capital in biotech doesn’t bring a product forward, it just starts the long journey. In the tech world, a seed stage investment can help a startup assess product-market fit questions and perhaps even get early revenue traction with consumers/users; that’s just not the case with therapeutics. This model may fit with other areas of “biotech” like research tools, synbio service companies, or B2B life science businesses, but making drugs for patients isn’t weeks or months but many many years away from revenues. Given the timelines, and the reality that a biotech will burn cash without any top line for 10+ years, you can’t “just” be a seed investor. You have to keep investing capital until real inflection points happen, typically human proof-of-concept in the clinic (leading to an M&A event or an IPO). If you are only a seed investor, you will undoubtedly be massively diluted in the process. That’s why we, and most other biotech investors, tend to participate in all the private financings of our companies – you typically have to be there for the entire journey to make returns in biotech.
  3. Disruptions in regulated businesses where lives are at stake necessitate deliberate evolution not revolution. Mark Zuckerberg’s idea that success requires startups to “move fast and break things” doesn’t apply to therapeutics companies. We are making bioactive substances to put into patients, maybe even permanently engineer their genes. This isn’t to be taken lightly or haphazardly. Regulators need to be brought along with new innovations. Iteration cycles are long. Safety and toxicology studies need to be thorough and take a certain immutable amount of time. Patients need to understand the risks, as people die because of drugs in clinical trials. Further, as an animal lover, think about the hundreds of animals we expose in our preclinical experiments in each drug program: the idea of animal models as things to “break” by moving programs faster doesn’t seem right. Doing well-controlled, scientifically-justified experiments is not only smart business, but it’s the right moral thing to do. Drug R&D, unlike many of the areas that tech VCs invest in, isn’t the place for rapid iteration where it’s ok to have a sloppy beta-test product hit the market (or go into the clinic, by analogy). Deliberate, thoughtful advancement of optimized drug candidates that could go the full distance to approval is not really a choice but an axiom. The therapeutics R&D model has evolved over the past few decades, and I’m sure it will continue to do so. But it won’t change overnight as “do no harm” (i.e. “don’t break things”) is an important guiding principle in bringing new medicines to market.
  4. Experienced executive talent is critical to success in therapeutics. CEO inexperience, naivety, and poor judgment have destroyed a ton of value over the past 40 years in biotech. In light of this history, we and other veteran VCs generally favor executives who have honed their skills running R&D programs or closing BD deals, and already demonstrated some level of judgment, credibility, and, frankly in light of Theranos, business ethics. I’ve written and tweeted on this extensively, so won’t rehash it here – but the idea that a recent PhD or post-doc who has studied a narrow piece of biology for 5+ years is ready to run a multi-disciplinary drug R&D organization powered with $10Ms in their first job out of academia seems far-fetched. We do, however, back first-time CxO’s all the time (most of our CEOs are first-timers), including lots of folks who’ve never “made money” for themselves or investors before. We are constantly adding talent to our network from biotech, pharma, and academia. We’ve recruited folks from all over the country, and our executive ranks are full of immigrants from all over the world. We pride ourselves on a portfolio that’s not a “club” of only insiders, but are open to working with great folks from around the ecosystem. Our in-house venture creation model does generally bias us towards folks who are willing to live in or commute to Boston, but even that’s not always the case.
  5. Founders, entrepreneurs, and early stage company-creation VCs frequently have the same interests. As company formation focused investor-entrepreneurs, we are hurt by onerous VC terms too. In general, we try to increase the alignment between our interests and that of our other co-founders – and most VC financial terms don’t tend to do this. Large liquidation preferences sitting on top of the cap table affect my early stakes much like a founding management team. I often have proposed “call-able” future tranches of funding if milestones are met, rather than “put-able” future tranches (the latter enable investors to buy up at their discretion at pre-agreed lower prices vs keeping options open for better cost-of-capital outside financings). One big difference in alignment arises with anti-dilution: generally early founder-investors don’t get reloaded with fresh stock or options the way management teams do (without investing more capital). I’m very sensitive to teams that are indifferent to dilution, just as most founders are. As for “control” protections and the most important decision of a Board – which is the hiring and firing of a CEO – I do think investors who’ve significantly financed the enterprise should have a big say in this decision. Further, as a general observation, biotech VC terms have actually become more startup-friendly in recent years (here).

Those counterpoints aside, I eagerly await the verdict on whether these new tech-oriented, young founder models for biotech venture formation deliver value for their teams, investors, and patients.

Unfortunately, given the nature of drug R&D, we won’t know that verdict for a long, long time.

This entry was posted in Biotech financing, Biotech investment themes, Capital efficiency, Drug discovery, Talent and tagged , , , . Bookmark the permalink.