The OSF Playbook: A Journey to Investing in Future Sci-Tech

Sizing Up Synthetic Biology Risks

When we started, we didn’t know what the company’s greatest strengths or weaknesses might be. After a few rounds of discussion, we found that the company was well positioned to handle the technical challenges of creating engineered organisms at small, lab-scale volume. The greatest technical risk was whether it could successfully scale up manufacturing, which we found to be a key failure point for many synthetic biology companies.

Working with domain experts, we established the company had solid data on strains it had created, and had developed a rich toolset and an organized, repeatable design process. We were surprised to find that unlike many science-based companies, the technical risk was relatively low because of the strength of their early data combined with competitors who had already demonstrated design, scale-up, and profitable sales of a plant extract replacement product. While the company had filed patents, it was not clear how much advantage this offers because it remains to be seen which key patents may dominate the synthetic biology space and which patents may offer little advantage due to potential workarounds.

During the diligence process, we updated the value of technical risk factors based on new facts we learned and expert input. For example, expert due diligence helped explain how the company constrained its development by choosing a well-studied organism, which lowered technical risk and increased our confidence in their ability to create the product. Each modification changed the outcome scenario ERM values (Figure 3), which helped direct the focus of our next questions to company and domain experts.

We learned more about the company and industry by modeling not only the company itself but also its most significant competitor. After comparing their approaches, we found that common difficulties in this evolving domain were a) creating a commercial product that can be produced at scale, and b) creating significant technological advantage over and economic moats blocking competition.

We gained further insights about the industry context by looking at the bigger-picture questions shown in Sidebar 3. It seems likely that the synthetic biology chemicals industry may settle into an oligopolistic market structure with a few dominant firms and many smaller niche players due to similarities with the traditional specialty chemicals and biotech industries which have such a market structure. Many synthetic biology companies are looking to manufacture similar chemical families, and most competitors are optimizing their processes for a certain parent biochemical pathway.

A winner-take-all dynamic seems unlikely in this segment of synthetic biology given the variety of players, low technical differentiation, narrow moats, modest switching costs, and potential for IP workarounds.

Gauging the Upside

Ultimately, the model produces a single number, the ERM, which measures the overall attractiveness of the investment and is the probability-weighted average of all possible scenarios.  It makes it possible to assess the relative impact of factors that have different units. ERM is a next-generation update to the rules of thumb VCs have been using for decades to try to maximize returns over a portfolio.

For example, there is the oft-quoted heuristic to choose companies with a ten-fold or greater return based on the assumption that one-third will fail, one-third will break even, and one-third will be winners.  This guideline is easy to use, but turns out to be wrong much of the time.1 Instead of fixing assumptions about probability of success, decision modeling estimates the chance of success for each startup individually. When performed across a portfolio, the model offers a tailored, integrated view of potential returns.

Under ideal circumstances — with a perfectly chosen model structure and perfect base estimates for model factors — the ERM would give us the actual expected value of the investment.  Yet ERM depends on the assumptions behind the model structure and model inputs. We may be highly uncertain about some of these assumptions; the model can help us grapple with this uncertainty by allowing us to develop an intuitive feel for the system by changing input values and model structures and observing their effects on the value of the ERM. Thus, observing the distribution of ERM outcomes across many different sets of assumptions may be more useful than trying to develop a single perfect base case.

In the model structure, we can accommodate different possible outcome scenarios, market scenarios, and exit scenarios that affect potential returns. We considered:

  1. Using base case assumptions, is the expected return high or low?  What is the ERM in an optimistic scenario?  Are these numbers in line with typical values for this industry? If not, why not? Are the model assumptions suspect, or is the company truly an outlier?
  2. Try changing the assumptions based on new insights or considered alternatives. Should the base case assumptions be changed? Why or why not?  
  3. Perform a sensitivity analysis based on the ERM to identify which factors have the highest impact on returns. What are the risk factors that might cause failure? If returns are low, what are assumptions that, if changed, might make the investment viable?

While ERM summarizes the expected value of the investment, the downside of a single value output is that you lose information about the likelihood of different scenarios. Thus, looking at the probabilities and rewards across scenarios is still important (see Figure 3), especially for VCs who may be investing in small numbers of companies.

  1.   Seth Levine, “Venture Outcomes are Even More Skewed Than You Think”, http://www.sethlevine.com/archives/2014/08/venture-outcomes-are-even-more-skewed-than-you-think.html