Debating the merits of various biotech business models is in vogue today, in part as a response to the proliferation of experiments around the right form and function of the companies themselves. An often asked question: are we supporting CEOs to build the next Amgen’s or Biogen’s, or are we simply backing glorified project managers? It’s not nearly as simple or as binary, but one’s worldview of what biotech models are viable in today’s capital market certainly shapes the answer to those questions.
This post was triggered in part by a recent question from a potential BD candidate: “I’m not sure what worldview of biotech I subscribe to? Is it Atlas’ asset-centric approach, or XYZ’s approach?” While I agree the former is an interesting question to ask, I cringed when he brought up Atlas as having a single approach to subscribe to. Lots of models can and do work, and while we do like virtual asset-centric biotechs, we’ve also got a much broader “worldview” that just those. To that end, I thought it worthwhile in framing up at least a few flavors on the spectrum.
But before laying out those models, I think it’s worth emphasizing that there are two key elements to our “worldview” of early stage biotech that are essential for any investment we make:
- Innovation is the first critical piece. The scientific basis needs to be compelling and offer a big opportunity for differentiated impact on patients. Big science, novel biology, new modalities, alternative approaches – some component of these must be part of the premise. Incremental innovations around reformulations or other modest tweeks, even if of some value in a disease setting, don’t typically get us excited and haven’t over the past five years.
- Capital efficiency is a second key principle of any company we decide to back. More of the details of capital efficiency in a future post, but in short its about getting the balance right between the amount of equity capital invested and the value created. Back in 2007, Nature Biotech’s “When Less is More” highlighted the theme of capital efficiency, and it is even truer now than it was then. All of the business models we commit to these days have strong theme of capital efficiency running through them.
But the range of business models that can embrace these two key ingredients are large. Some are virtual, which can certainly be “beautiful” as John Carroll of FierceBiotech recently noted. But some aren’t, and frankly wouldn’t be successful if run virtually. Some are focused on single assets rather than product portfolios. But some have a discovery platform-generated portfolio of assets.
Let me frame up two very different genotypes of deals we like, and what I see as a few Critical Success Factors (CSFs) for those models, and then close with a third type that we tend to avoid.
Drug Discovery Product Engines
We’re big believers in the discovery platform model when executed with an efficient deployment of resources, especially when aimed at big biology (e.g., microRNAs at Miragen; long-non-coding RNAs at RaNA), novel biologic modalities (e.g., bispecifics at F-star, Adnectins at Adnexus, cyclic peptides at Bicycle, protein-glycan decoys at Protaffin), or innovative “smart chemistry” approaches (e.g., covalent drugs at Avila, cracking the hard-to-drug challenge at Nimbus). Although all these platforms aim to push innovative science forward, these aren’t “science projects” like many discovery platforms were in the 1990s; importantly, the path to product candidates is generally known at company inception, and the key emphasis initially is on validating that the platform can generate repeatable advances. Novel biology is always full of surprises later on, but the product-oriented direction of these platforms is set during the company’s infancy. Our hope when we’re helping start these types of companies is that on $25-50M of equity capital we can get to a realization of value. In order to do this, one obviously needs a great management team. But beyond that, at least three specific CSFs must typically be met with this model:
- Strategic, iterative target selection must be done early and often, coupled with a willingness to abandon less interesting lead programs if killer applications can’t be identified. The impulse to pick a suboptimal lead horse too soon, or stick to it too long, destroys value in this model as they consume scarce resources. The key is to have a strong sense of the big differentiation angles and how to find programs that fit those spec’s.
- Non-dilutive capital must be secured to offset the equity burn rate of the platform in order to create a robust engine. These non-equity funds typically come from discovery partnerships with Pharma around the platform, but can come through foundations and grant-making bodies. The devil is in the details here: making sure a discovery alliance is truly net-positive is always the rub on these types of relationships. Does the deal generate free cash flow to fund other things? And don’t underestimate the impact on a management team’s bandwidth to do other projects when an all-consuming discovery alliance is in place. In general, I’d argue that getting at or above 1x the amount of equity invested via non-dilutive partnerships is a good aspiration over the first 3-5 years of a startup.
- In-house vs outsourced capabilities must be balanced. Knowing what to build into the core lab competency of these business models is key, and what to strategically outsource. Scaling the fixed infrastructure and employee base too early can create a burn rate that outpaces the value creation, but not investing in the core enough can handicap the platform’s validation and robustness. No platform is the same, so there isn’t a one-size-fits-all answer for this one. But it’s likely that more than a few dozen in-house FTEs is starting to push the limits of efficiency for a discovery-stage startup.
These three CSFs certainly help put these Discovery Product Engine models onto the path to value creation – but in the end, exciting data around the lead product is often the required catalyst for value realization. This can be done through sale of the company, or (rarely today) through equity appreciation in the public markets. An alternative and increasingly interesting structure is to house the platform within a holding company (LLC) where realizations could occur through the sale of individual program-specific subsidiaries (like the structure at Nimbus Discovery LLC).
While this first example model has certain elements of the traditional brick-and-mortar biotech (though I’d argue with greater capital efficiency and value realization flexibility), the second type of business model, below, is very different and is increasingly popular today.
Asset-Centric Development Companies
At the other end of the spectrum, we are also very supportive of single project-focused entities that drive the development of a single innovative therapeutic program from Point “A” to Point “B”. For us, “A” is typically somewhere between a well characterized lead and an IND, and “B” is most often clinical data of some sort and most often Proof-of-Concept in Phase 2a. For example, both Stromedix’s IPF program and Arteaus’ migraine program were both peri-IND stage when we brought them into those startups. These models are almost exclusively virtual in nature; that is, they have no wet labs of their own, and conduct all their science and clinical medicine through partners. The virtues of this asset-centric model have been highlighted previously here in this blog, and more recently in some great blogs by the Index Ventures team. Our aspiration in these models is that the capital intensity for these asset-plays is far lower: $15-25M would be the sweetspot, and occasionally larger if we decide to “go longer” into the clinic. Importantly, although virtual, these company’s often purse novel mechanisms and highly innovative therapeutics, so the depth of the expertise around the in vivo biology and clinical translation needs to be very strong. A few of the CSFs of this model:
- Hiring an experienced senior core team. Seasoned drug hunters/developers who can be broad utility players are key in this model; there’s little room to learn on the job with a lean team, so the core group needs to be experienced, well-versed in a wide range of aspects in drug R&D, and mature enough to deal with the inevitable non-linear nature of these companies. This isn’t fundamentally different than a discovery platform’s leadership team (where great talent is needed too), but the success of an asset-centric play is even more tightly linked to the contributions of the few key hires.
- Accessing the best of the ecosystem. These virtual models rely on a broad network of superb advisors, CRO partners, and academic collaborators, so their selection is critical. A bad CRO is more than a headache, it can be the end of your program. Sometimes the choice of partner will be about cost (e.g., intermediate chemical synthesis), but other times it might be more expensive but of higher quality than hiring an in-house generalist. Making tradeoffs in the choices of the right partner with regard to the project management triad of time, cost, and quality isn’t easy, but a core team with very strong project management expertise certainly helps. Lastly, accessing integrated groups of advisors as real partners (like PharmaAdvisors or PharmaDirections) is incredibly enabling and reduces the entropy of managing lots of individual relationships.
- Balancing BD campaigns with bandwidth. One of the possible drawbacks to this model is that any fire-drill or broad BD campaign is a huge distraction: for example, when the top three people in a 5-person virtual play have to attend lots of partnering or Series B fundraising meetings, it’s a hugely inefficient tax on the company. The best asset-centric companies understand this and have a tight BD strategy around cultivating potential partner/investor options over time. Bringing on superb BD advisors, as well as establishing a board/syndicate of well-networked Directors/investors, is very enabling when management bandwidth is so scarce and degrees of freedom limited.
Most often this single-asset development model realizes its value creation through an acquisition by a larger company, and are often “built for purpose” to that end. Stromedix, PanGenetics, and FerroKin acquisitions are all evidence of this. Furthermore, these single asset plays are amendable to pre-arranged “defined liquidity path” relationships with Pharma, like our Arteaus’ deal with Lilly. Locking in a partner at the beginning helps reduce the financing and exit risk in these companies. We’ve been doing this via our Atlas Venture Development Corp initiative, and others like CMEA’s Velocity and the Lilly Mirror Funds are also experimenting with it.
—
There are many other types of deals that are interesting, but these two frame up the ends of the “biotech worldview” spectrum for us – from product-generating platforms to virtual single asset entities.
Are either of these likely to be the next Amgen? Not in today’s public equity capital markets. But both can retain the option of “going longer” should conditions and the cost-of-capital improve, especially if they haven’t already prematurely scaled. In general its our belief that over-capitalizing young platform or product companies with lots of equity, in the hope that one is building the next Amgen, creates a corporate albatross that is hard to escape – and destroys value over time.
Before closing out, I’ll share one deal genotype that we’re far less sanguine about than many others in the investment community: the big “company-building” product roll-up play. These are created by assembling multiple assets without an underlying discovery platform. The premise underpinning this model is that a broader product portfolio creates diversification (less risk of catastrophic deal failure), creates marketing synergy, and better leverages a management team. These premises may hold in some cases, but more often than not they don’t play out favorably for investors. Michele Ollier of Index Ventures dismantles this business model in her recent post on asset-centric models, as does David Grainger in his description of company-building as the chocolate-flavored poison in biotech. I’d also note that a portfolio of stage-diversified assets in these models often ends up creating the asymmetry of maturity challenge that I’ve written about before, which handicaps the equity efficiency over time (i.e., if the lead works, it drives the value; if the lead fails, it craters the value; this makes any real expenditures on earlier programs irrelevant to value and hence wasted). Beyond those issues, the larger portfolio in these company-building exercises is often associated with high burn rates and little non-dilutive leverage (since there’s no platform to partner). I also think there are other ways to leverage a great management team across multiple asset-centric entities. It’s true that some investors and management teams make this product-consolidating, company-building model work well, and Clovis Oncology may be a great example of a successful deal of this type. But it’s not an approach that has created lots of value for the industry over time, perhaps that’s because there are few management teams like that at Clovis.
In short, lots of biotech worldviews out there – some we certainly like more than others, but we’re not dogmatic that there’s only one route to biotech heaven here.
