On Ginkgo Bioworks
This piece represents ARK’s newest form of written content: “Works in Progress (WIP)”. WIP content is a snapshot of our internal research process. These pieces are unfinished and generally more technical than ARK’s blog content. The goal of WIP content is to open our thinking as it evolves such that we can gather feedback and learn quickly if we are wrong. Please see ARK’s general social media disclosures, which also apply to WIP content: https://ark-invest.com/terms/#twitter. Note ARK’s Adviser disclosures at the end of this piece as well. Finally, please note that certain of ARK’s clients are shareholders of Ginkgo Bioworks (DNA) at the time of publication (10.22.2021).
By: Simon Barnett and Pierce Jamieson
INTRODUCTION
Ginkgo Bioworks (DNA) operates at the interface between some of the most powerful technological shifts in history — artificial intelligence, DNA sequencing, DNA synthesis, genetic engineering, collaborative robotics, and more. Collectively, these technologies are empowering the field of synthetic biology (synbio).
We believe that synbio companies taking on diverse projects with flexible, high-throughput workflows could be among the first to generate critical institutional knowledge about genetic engineering in a production environment.
Whether one likes or dislikes Ginkgo — we think it’s hard to argue that the company isn’t forging a new category for itself. Therefore, it’s not surprising to see investors struggling to grok Ginkgo’s technology stack and business model which have hallmarks of biotech, venture capital, and software engineering, amongst other things. That synbio already is a loosely defined, esoteric industry surely exacerbates the confusion. For these reasons, we haven’t been shocked to see an uptick in polarization on social media — especially since Ginkgo’s foray into the public markets.
Importantly, this is not a bull case for Ginkgo, nor is it an argument for fair value. We do not compare Ginkgo to other synbio companies. Indeed, we are continuing our work on substantially all companies in the sector. This also is not a point-for-point rebuttal to any other work.
We believe that Ginkgo Bioworks is an archetypal synbio company — one whose platform infrastructure could be a boon to the entire synbio ecosystem over the coming years and decades. If you’re curious about Ginkgo Bioworks, its technology, or its platform business model — read on.
GINKGO’S HORIZONTAL BUSINESS MODEL
Since its founding in 2009, Ginkgo has been steadfastly and openly committed to a horizontal (platform) business model. Crudely, we can divide the synbio value chain into three main components:
- Discover ~ Identify a molecule, process, or strain with commercial value.
- Engineer ~ Reprogram an organism to fit the goals outlined in Step 1.
- Manufacture ~ Optimize and scale the engineered organism from Step 2.
Ginkgo has chosen to focus solely on its core competencies — discovery and engineering. It partners with both related and unrelated customers who onboard cell programs to Ginkgo’s platform. While it does optimize bioprocess parameters for manufacturing its engineered strains (as shown below) — Ginkgo doesn’t actually operate manufacturing facilities. The company works with contract manufacturing organizations (CMOs) to handle this step. Keep in mind that we are not advocating for a platform-oriented approach over a product-oriented approach in this piece.
We think a unique aspect of this business model is how it enriches Ginkgo’s technology stack (Foundries + Codebase) in a virtuous cycle. As a reminder, Ginkgo’s Foundries are mostly automated, high-throughput laboratories where the company conducts physical experiments in parallel.
Owing to its concerted, decade-long effort to lower strain test costs (as shown below), the average cell program now is fully paid for by Ginkgo’s customers. We believe Ginkgo has rightly chosen to pass these savings onto customers rather than extracting a margin from its Foundries. In this way, cell program customers are effectively subsidizing the development of Ginkgo’s Codebase. This is true even for failed programs because of Ginkgo’s focus on experimental forensics — why did the program fail and how can they prevent a similar failure mode in the future? In a more traditional product-oriented approach, companies pay their own R&D expenses and may need to underwrite discovery and commercialization risks that Ginkgo doesn’t.
Ginkgo’s Customers
In the following sections, we’ll deal with issues relating to platform monetization and customer mix.
Platform Ventures
For the past several years, Ginkgo has been upfront with its strategy to form, advise, and spin-out companies in different commercial verticals such as Joyn Bio, Motif FoodWorks, Verb Biotics, Ayana Bio,Allonia, and Kalo Ingredients. Collectively, these groups are called Platform Ventures. In its S-1 filing (pg. 79–80) from 08/11/2021, Ginkgo clearly disclosed each of its Platform Ventures.
Why does Ginkgo create Platform Ventures?
We can think of a handful of reasons. Firstly, recall that many of these ventures were created when it was more expensive to use Ginkgo’s platform than to operate a synbio lab internally. Back then, Ginkgo had fewer proof statements about the power of its platform. Simply put, the overall value proposition for customers was lower.
Given its horizontal model, Ginkgo likely didn’t want to dilute its intellectual capital by developing assets internally. Spinning out Platform Ventures means Ginkgo avoids competing with potential customers. Keep in mind that the Codebase assets developed through Platform Venture programs are evergreen and repurposable by Ginkgo.
By helping capitalize Platform Ventures, through both direct and third-party investment, Ginkgo is able to navigate the factors we’ve just laid out. In exchange for Foundry services, Ginkgo receives a combination of cash and non-cash (e.g., equity) compensation. As of today, there are only six Platform Ventures, so it’s not taxing to discern the deal structures. In its S-1, Ginkgo provided deal overviews (pg. 79–80) as well as detailed compensation breakdowns beginning on page F-123. On page F-124, you can see how Ginkgo booked deferred revenue from Kalo Ingredients. Simply, Ginkgo is under contract to provide Foundry services to Kalo with defined, revenue-triggering checkpoints laid out in the companies’ technical development plan(s) (TDP).
Structured Partnerships
We’ll briefly touch on Structured Partnerships. These are similar to Platform Ventures with the key distinction that they are not spun out of Ginkgo. Instead, these are existing companies with programs that Ginkgo has deemed valuable enough to enter into a more formal collaboration. Currently, there are only two Structured Partnerships — Genomatica and Synlogic (SYBX). Once again, the details are codified in Ginkgo’s S-1 on page 80.
Recall that Ginkgo may choose to receive low-cost-basis equity as a form of prepayment for R&D services, which is explained in Ginkgo’s S-1 on page 83 under the section entitled Foundry Revenue. As an example, you can see how Ginkgo adjusted for its non-cash (equity) compensation from Synlogic on slide 51 of Ginkgo’s May Investor Day presentation filed on 05/11/2021.
Related Parties and Customer Mix Shift
Together, we refer to Structured Partnerships and Platform Ventures as related parties. Ginkgo’s current level of related-party revenue concentration is a widely known fact — and an important one to consider.
We’re sympathetic to investors who are concerned about customer concentration. Indeed, this was a concern three years ago regarding Twist Bioscience (TWST) whose largest customer at the time was, ironically enough, Ginkgo. Twist has since greatly diversified its revenue mix across many applications of synthetic oligonucleotides — especially within next-generation DNA sequencing (NGS), as shown below. We plan to pay close attention to customer mix shift in the coming quarters and years. Based on our research, we think the bulk of cell programs will shift towards unrelated parties. We’ll discuss this in the next section.
Unrelated Parties
We believe that Ginkgo’s cell program composition will become increasingly weighted towards unrelated parties. Why do we think this?
As we’ve shown above, the pricepoints Ginkgo is able to offer potential customers have reached a critical threshold where it’s now cheaper to partner with Ginkgo than for a startup to build and operate its own lab. Still, less-capitalized startups might need to part with equity to afford Ginkgo’s services. This is a friction point, but one that we think is being ameliorated by Foundry fees that are getting inexorably smaller. If a customer cannot afford the platform on its desired terms, it’s likely that they’ll just need a year or two for costs to continue falling. Moreover, as customer successes accumulate, startups may be more willing to work with Ginkgo. Over the past 18 months, Ginkgo has announced many new partnerships (some of them being multi-program), such as:
- Agenus ~ Vaccine Adjuvants
- Tantu ~ Gastrointestinal Diseases
- Cambium ~ Industrial Biomaterials
- Givaudan ~ Fragrances, Cosmetics, and Consumer Goods
- Huue ~ Organic Textile Dyes
- Antheia ~ Plant-Derived Pharmaceuticals
- Bolt Threads ~ b-Silk Protein
- Sumimoto Chemical ~ (Undisclosed)
- Biogen ~ AAV Production for Gene Therapy
- Corteva ~ Sustainable Agriculture
We believe many of these partners are unrelated, but look forward to seeing the details published in time.
Some have argued that some of these ventures must be fake, citing a lack of web design, a dearth of personnel, or brief corporate history. We think these statements inadvertently highlight a wonderful aspect of Ginkgo’s platform and business model — that even small, inchoate companies can take a shot at entering the synbio industry. We are certainly not claiming that the majority of these ventures will prove successful. That is unrealistic. However, recall that even failed programs inject valuable information into Ginkgo’s Codebase that can be recycled and reapplied elsewhere. By lowering the barriers to getting on Ginkgo’s platform, customers will get more shots-on-goal (cell programs), further increasing the likelihood of upside on equity interests or royalties.
We’d be remiss not to mention that many of these new customers are large, established entities that generally compensate Ginkgo with upfront cash payments and product royalties.
Given the historical obstacles associated with company creation in the synbio space (such as initial CapEx), Ginkgo has chosen a flexible model that allows it to monetize its platform through a combination of cash and non-cash components such as equity and royalties. Established, well-capitalized companies are more likely to pay cash for Foundry services and/or give Ginkgo a small royalty for successfully commercialized products. Smaller companies may offer Ginkgo a split of cash and equity in their business in exchange for Foundry services.
We’ve also seen questions relating to revenue round-tripping from Ginkgo’s existing investors. Frankly, we’d be more concerned if Ginkgo’s investors didn’t support products developed on the Platform. We’d also be concerned if Ginkgo’s investors were the only ones supporting Platform Ventures. But they’re not. For example, Motif FoodWorks’ recent $226 million Series B round was led by new, unaffiliated investors such as the Ontario Teachers’ Pension Plan Board, AiiM Partners, Wittington Ventures, Rethink Food, Rage Capital, Rellevant Partners, and funds managed by BlackRock.
As we’ve mentioned above, 2021 has seen many unrelated, marquee customers flock to Ginkgo’s platform. We believe this trend will continue as customer successes accumulate and Foundry fees continue to fall. Anecdotally, this year has been one where Ginkgo’s sales pipeline has taken on inbound sales from companies wishing to get on the platform. This also explains one reason why Ginkgo went public when it did — to have the capacity to scale alongside increasing inbound demand.
Successful Programs, Processes, and Strains
We have been surprised to see it claimed that Ginkgo’s platform hasn’t generated successful products or strains. It clearly has. Here are three examples.
- Motif FoodWorks ~ Ginkgo engineered yeast strains to maximize protein production for use in Motif’s products. As we outlined previously, Ginkgo exceeded Motif’s project specifications by 70% in a timeframe shorter than originally estimated. These product(s) are set for commercial launch in 2H’21 and 1H’22 — supported by a $226 million Series B round.
- Cronos Group ~ Ginkgo achieved a key productivity milestone by successfully transferring several cannabinoid biosynthesis pathways from cannabis plants into yeast and bacteria — enabling commercial-scale production of CBG molecules. Cronos launched this first-in-kind 2:1 THC|CBG gummy on 10/20/2021 in Canada.
- Aldevron (Danaher) ~ In less than one year, Ginkgo helped Aldevron increase the production yield of vaccinia capping enzyme (VCE) by 10X. VCE is a critical component in mRNA-based therapies and vaccines.
We’d also like to take a moment to reference a great Twitter thread by @Varro_Analytics who has shed light on Ginkgo’s work with Aldevron. Recall that Moderna’s (MRNA) COVID-19 vaccine utilizes VCE from Aldevron. According to Ginkgo’s management, the royalties from this project may begin flowing by the end of 2021. These royalties are 100% contribution margin for Ginkgo.
GINKGO’S TECHNOLOGY PLATFORM
Together, Ginkgo’s Codebase and Foundries (collectively, the Platform) constitute a substantial amount of protected IP and institutional know-how. We believe these capabilities coupled with Ginkgo’s platform business model will result in a powerful and enduring competitive moat.
Foundries: Adaptive Laboratory Evolution
A single Google patent search yields 100+ patents assigned to Ginkgo Bioworks (including through acquisitions). Broadly, these patents protect an array of inventions enhancing Ginkgo’s ability to perform adaptive laboratory evolution (ALE) — one of the most central methods in synthetic biology (synbio). But what is ALE?
On a high level, in ALE, an experimenter applies evolutionary pressure to a population of cells to change them in a desirable way. Instead of making targeted changes to an organism’s genome, one can link the organism’s fitness (survival and reproduction) to a desired set of traits and then allow Nature to do the heavy lifting. With a carefully controlled experiment, cells harboring genetic mutations that confer the desired traits should begin to dominate the population. This is survival of the fittest after all.
ALE experiments typically are iterative and yield minor, incremental improvements with each cycle. Occasionally, the culture’s fitness can break out of a shallow improvement trend into an exponential curve, as shown in the chart below.
We argue that Ginkgo’s platform (Foundries + Codebase) maximizes the probability that such an exponential improvement will occur. In later sections, we also outline how the platform optimizes for experimental speed, throughput, and internal unit economics.
Foundries: Cells Beyond Yeast
Within its legacy Foundry capabilities, Ginkgo has created and patented several cell lines capable of utilizing alternative sources of nitrogen (N), phosphorous (P), and sulfur (S)—all critical-to-life nutrients. The company’s cellular repertoire extends beyond yeasts to include other fungi (e.g. N. crassa, S. cerevisiae), bacteria (e.g. E. coli), and mammals (e.g. CHO). By cleverly linking the genetic elements they are optimizing with those that enable alternative N, P, and S source utilization, Ginkgo’s scientists can finely tune their ALE experiments across these cells.
Visually, you can think of fitness in an ALE experiment like a rugged landscape with peaks and valleys, as shown below. The horizontal axes correspond to DNA sequence changes that occur inside the organism’s genome. Imagine we begin with a ball resting comfortably at Point 1 — representing an unmodified strain. With each ALE iteration, a force with random direction and magnitude acts on the ball. This force corresponds to mutations (or modifications) to the organism’s genome.
The scientists’ goal is to help the ball get from Point 1 (an unmodified strain) to Point 2 (the global fitness peak) — the point of optimal fitness. Unfortunately, ALE experiments often feature several vexing scenarios. If the ball is caught in a local minimum (a valley), any change can appear beneficial to fitness, making it difficult to know which ascent leads to the highest possible peak. Alternatively, if the ball is sitting on a local maximum (a peak), it may be tempting to claim victory as any change would detract from fitness. And yet, an even higher peak (global maximum) could be hidden away above the clouds. We argue that Ginkgo’s platform IP not only gives them an advantageous starting Point 1 (custom cell lines) but also the ability to chart an optimal path towards Point 2.
Let’s use one of Ginkgo’s customers, Motif FoodWorks, as an example.
Motif asked Ginkgo to help them optimize a yeast-based platform capable of producing plant and meat-derived proteins at a discount relative to conventional methods. By linking the yeast’s ability to harness alternative nutrients with its capacity to synthesize the desired protein products, Ginkgo was able to deliver yeast strains that exceeded Motif’s specifications by >70%. Ginkgo’s strains produced high-quality proteins in amounts 5–20X higher than previously thought possible. Motif expects to launch its first ingredient from this yeast-based platform in 2021, with a second ingredient slated for commercialization in early 2022. Indeed, this successful ALE example required much more IP beyond Ginkgo’s ability to harness alternative nutrient utilization.
Foundries: Biosynthesis and Automation
In addition to optimizing protein expression, Ginkgo has patented novel methods for the biosynthesis of lipids (e.g., fats, oils, waxes), terpenoids, cannabinoids, and various other biomolecules like nucleic acids and plasmid vectors.
Beyond its arsenal of cell strains and synthesis methods, Ginkgo’s myriad wet-lab patents reveal more about how the company can quickly iterate towards optimal strain fitness. Contrary to a misconception, Ginkgo employs both off-the-shelf and proprietary liquid handling techniques. For example, the Foundries utilize a clever method to transfer minute amounts of liquid reagents under positive pressure using rotary valves. Ginkgo also employs automation technology that it co-developed with Transcriptic (now Strateos). Together, these capabilities allow Ginkgo to combine traditionally disparate lab processes together into the same unit operations — improving its per-experiment economics across multiple cell types. You can see this evidenced below by the fact that strain test throughput is growing faster than the number of daily lab operations.
In 2017, Ginkgo purchased a DNA synthesis company called Gen9. This corpus of IP enables Ginkgo to synthesize large (>10 kb) genetic constructs de novo that contain many genes up to entire metabolic pathways. Using this technology, Ginkgo can explore unique regions of sequence space during ALE (see the above ALE landscape image) — regions that may just harbor fitness peaks. For those interested, you can learn more about Ginkgo’s biosynthesis capabilities here, or see the image below.
By insourcing large-scale DNA synthesis, Ginkgo obviates the need to stitch smaller fragments together using conventional molecular cloning methods — an oft error-prone process that can take weeks. Instead, Ginkgo can directly synthesize large constructs rapidly and to their exact specifications. Perhaps more importantly, vertically integrating through DNA synthesis gives Ginkgo pricing leverage and supply chain independence over a key input for its experiments.
Foundries: Hardware and Measurement Tools
Equally vital to its ability to perform high-throughput experiments is Ginkgo’s ability to measure the results of those experiments.
The company has a protected set of technologies to tag genetic sequences inserted into cells, enabling Ginkgo to precisely monitor how steps within a metabolic pathway are changing the desired outputs. These capabilities are augmented by custom functional assays for cell characterization that Ginkgo has developed in collaboration with Berkeley Lights (BLI).
For those unaware, Berkeley Lights’ platforms utilize a modified form of optical tweezers — a breakthrough technology that was awarded the Nobel Prize in Physics in 2018. Berkeley’s systems move single cells using perturbations in an electric field guided by incident laser light — a technique that slashes platform energy requirements by 100,000X over standard optical tweezers.
Moreover, in a recent preprint, Ginkgo and Synlogic described a novel assay methodology enabling automated, high-throughput measurement of enzyme activity in E. coli cell lysates. The researchers used this approach to screen natural variants of the same enzyme(s) from >1,000 different plant, fungal, and bacterial species (i.e., homologs). With this technique, we think Ginkgo can better winnow down enzymatic starting points for further optimization.
Codebase
We believe Ginkgo’s Codebase and associated software stack are substantially differentiated from resources in the public domain. As referenced previously, these capabilities enable Ginkgo to accelerate strain iteration, lower Foundry costs, and increase the probability of program success.
Before we dive into the details, we’d like to broadly address how life sciences companies interact with open-source data repositories.
Let’s turn to the molecular diagnostics industry for a moment. The drastic cost declines associated with next-generation DNA sequencing (NGS) have caused an exponential increase in publicly available information linking DNA mutations to disease phenotypes. Many of these so-called variant interpretations are housed in free databases such as ClinVar and gnomAD. Nearly all companies at the vanguard of the molecular diagnostics industry leverage these resources in their variant interpretation pipelines. And yet, despite what would appear to be a level playing field in this regard — this industry is fiercely competitive. Why?
- Biology still is mostly unexplored. Especially at genome-scale, most variants in a patient’s genome won’t have an associated truth label in a database (i.e., pathogenic or benign). This highlights the importance of creating systems that assist with the scalable, on-the-fly interpretation of variants and their downstream functional consequences — such as Jungla’s Functional Modeling Platform (FMP).
- Biology is complex and not fully deterministic. It requires a systems-based approach. Much like Ginkgo, companies in the diagnostic industry are layering in orthogonal information such as transcriptomics and proteomics to understand nuanced biology at scale.
- Public databases oftentimes are fragmented, poorly annotated, and are of lower quality. As such, there is a ceiling to their usefulness. Companies need to create bespoke systems to ingest, structure, and discount lower quality data.
The molecular diagnostics analogy extends to the hardware layer in Ginkgo’s Foundries. Virtually all leading molecular diagnostics companies leverage Hamilton robotics during sample preparation as well as Illumina (ILMN) sequencers to generate data. Does that take away from their differentiation? Of course not.
While new, off-the-shelf equipment is present at cutting-edge firms, it certainly isn’t enough to generate a competitive moat. It requires both wet and dry-lab IP, proprietary methods, and institutional expertise weaving around capital equipment to create differentiation. This is just as true for molecular diagnostics companies as it is for Ginkgo.
Now let’s get into the Codebase.
Codebase: A Library of Reusable Biological Parts
On a high level, Ginkgo’s Codebase contains a vast library of reusable pieces of genetic code (indeed, these are DNA sequences) and engineered cell lines (we touched on this already). Importantly, the Codebase also contains exceptionally detailed performance data showing how these reusable pieces of code behave in the context of each of Ginkgo’s various cell lines. Code that runs one way in E. coli may not run the same way inside B. subtilis.
What are we referring to when we say ‘reusable pieces of genetic code’? These are DNA sequences of variable size that (a) have been fully characterized in Ginkgo’s Foundries and (b) accomplish a specific task. These parts could include things like gene clusters encoding for large swaths of a metabolic pathway, elements that can regulate levels of gene expression, or sequences that generate peptide tags with specific mass/charge (m/z) ratios detectable during mass spectrometry to assist with cell state quantification, as shown in the graphic below.
Returning briefly to the Foundries’ hardware stack, we can appreciate why even Ginkgo’s sequence data is unique. Ginkgo’s sequencing cluster makes extensive use of long-read sequencing (LRS) technologies from both Pacific Biosciences (PACB) and Oxford Nanopore (ONT.L). Why?
The vast majority of sequence data in open-source repositories was generated using Illumina (ILMN) short-read sequencing (SRS). However, many microbial genomes (e.g., bacteria) are highly repetitive and are composed of numerous tandem repeat elements — these are notoriously difficult to analyze with SRS technology. By incorporating both flavors of LRS, Ginkgo can generate higher quality, more complete genome assemblies spanning many of the hardest-to-sequence organisms.
LRS platforms are tremendous at analyzing RNA isoforms, going far beyond what SRS-derived gene expression data can yield. Extending this, we think Oxford Nanopore’s technology is great at capturing native epigenetic information, assembling bacterial genomes, and linking the behavior of regulatory elements such as distal enhancers and promoters. Meanwhile, PacBio’s accuracy can be leveraged to faithfully reconstruct diploid (e.g., mammalian) genomes, assess viral vectors, capture ultra-low-frequency mutations, and sequence through complex alleles.
Codebase: Data Structures and Genome Mining
Recently, we wrote about how accurate, LRS technology can be used to construct highly compact and computationally efficient data structures (e.g., pangenome graphs). In the related paper, the authors used their method (mdBG) to scour >660,000 bacterial genomes in search of genes conferring antimicrobial resistance (AMR) using minimal computing resources.
We note that Ginkgo has deep expertise and proprietary methods at the interface of metagenomic bioinformatics and AMR. Specifically, Ginkgo acquired these capabilities when it purchased Warp Drive Bio’s genome mining platform, which is purpose-built to uncover sequences conferring AMR (amongst other things) that wouldn’t be detectable with conventional methods— further differentiating Ginkgo’s sequence data from publicly available resources. We believe the integration of Warp Drive’s mining capabilities greatly enhance Ginkgo’s metabolic engineering efficiency, allowing them to focus on the best enzyme(s) for pathway tweaking.
To that end, a recent preprint between Ginkgo and Synlogic hints that the genome mining approach has applicability beyond AMR and novel antibiotics. In this scenario, the authors examined pathway and enzyme homologies between several species. Ultimately, they transformed a custom strain of E. coli using components from multiple organisms. Though early, this led to a substantial improvement in non-human primates’ ability to metabolize leucine (a critical-to-life amino acid), as shown below.
Given its newest Foundry focused on mammalian cell engineering, as well as a strong secular shift towards cell and gene therapies, we believe Ginkgo will push further into the biopharmaceutical landscape in the coming quarters and years.
Codebase: Functional Data
Beyond the sequence data that encodes for its library of reusable biological, building blocks, Ginkgo’s Codebase is highly annotated with functional data. What does this mean exactly?
Recall biology’s central dogma, that DNA is transcribed into RNA, which is then translated into proteins — the primary functional actors of an organism, as shown in the cartoon below. One might conclude that DNA sequence changes could therefore be used to predict a cell or organism’s traits. The reality is rarely so simple.
In most cases, DNA sequence data alone is insufficient to make definitive predictions about changes to an organism’s traits. That is to say that DNA sequence data is not fully deterministic of phenotype. Between each canonical pillar of the central dogma (DNA, RNA, protein), there are complex processes that biologists still are struggling to fully grasp — like alternative splicing, post-translational modifications, subcellular localization, and many more.
These complexities once again highlight the importance of Ginkgo’s state-of-the-art hardware layer. Without this, Ginkgo wouldn’t be able to fully characterize the downstream, functional consequences of the reusable parts and experimental parameters (selection pressures) also contained in its Codebase.
In summary, Ginkgo’s Codebase is much more than raw, low-quality DNA sequence data. As we’ve discussed, the Foundries contain both off-the-shelf and proprietary analysis methods that ensure the information being fed into the Codebase (e.g., sequence data, functional assays) is unique. With its biosynthesis acumen, Ginkgo can turn vetted Codebase sequences into real biological constructs for use in the Foundries. Altogether, we believe Ginkgo’s platform (Foundries + Codebase) helps its scientists chart optimal solution paths during ALE and other directed-evolution experiments — lessening the probability of program failure, decreasing iteration time, and improving operating efficiency.
It should go without saying that many of the references in this section may represent Ginkgo’s technology stack as of a few years ago. Companies in this space often have a rolling body of new IP that augments or supplants older approaches. Finally, we’d also like to point out that this has just been a casual walk through Ginkgo’s IP forest — it’s not an exhaustive list.
Miscellania
There are a handful of misguided opinions we feel are clouding the discussion around Ginkgo. While these didn’t cleanly fit in the body of this piece, we still felt it necessary to include them. We plan to update this section ad hoc in the future.
Is a price-to-sales (PS) multiple the best way to think about Ginkgo?
We’re not here to argue for what Ginkgo’s PS multiple should be. Instead, we contend that a PS multiple is a poor, standalone metric to appraise Ginkgo’s value. This ratio only takes into account Foundry revenue, not downstream value creation through product royalties and/or equity interests.
Consider for a moment how one values a (pre-commercial) therapy company.
You can take a sum-of-parts approach, which is one mechanism we use to value therapeutics pipelines. Crudely, you must parse the indication for use, patient population, actual science/mechanism, and probability for commercialization based on transition statistics that depend on the current phase of development (e.g., Phase 2). By discounting backward and summing the net-present value (NPV), you can attempt to value an entire therapeutics pipeline. While this method doesn’t map 1-to-1 to Ginkgo, we think it’s a reasonable framework for assigning NPV to cell programs.
Of course, we don’t take a fully reductionist view, either. While therapeutic modeling heuristics can be helpful — they’re not the end-all-be-all for valuing Ginkgo.
Is Ginkgo similar to Berkeley Lights or to Zymergen (ZY)?
Despite scurrilous attempts to pull Zymergen’s woes onto companies like Berkeley Lights or Ginkgo — we contend these comparisons are ill-informed.
Berkeley Lights unequivocally is a life science tools company. They make machines, not organisms.
Ginkgo also is not similar to Zymergen. As we’ve mentioned, Zymergen is a product-oriented company that seeks to internally discover, engineer, and manufacture assets. Ginkgo is a platform-oriented company that focuses mostly on engineering. In the body of this piece, we’ve showcased how synbio business models and technology stacks are intertwined, further demonstrating what is different about these companies. With a platform model, the failure (or success) of any given product shouldn’t matter to the same extent that it would in a product model.
Lastly, we’d like to address disappointing instances of ad hominem attacks and agism directed towards Ginkgo.
While we may agree, in the most general sense, that experience and age are correlated — we wholeheartedly disagree that youthfulness in synbio is a bad thing. In fact, we think that Ginkgo’s unrivaled connectivity with MIT and the broader academic ecosystem in Boston is a significant advantage. Moreover, Ginkgo’s founders, management team, board of directors are steeped with academic and industry veterans like Tom Knight (considered the father of synbio), Marijn Dekkers (CEO; Bayer, CEO; Thermo Fisher Scientific), Christian Henry (CEO; Pacific Biosciences, CFO/COO Illumina), Shyam Sankar (COO; Palantir), Reshma Kewalramani (CEO; Vertex Pharmaceuticals), Arie Belldegrun (Founder of Allogene Therapeutics and Kite Pharma), amongst others.
FIN & FUTURE
Ginkgo believes that we should apply rigorous engineering principles to help tame the randomness embedded in biological systems. That we shouldn’t wait idly by for the next quantum leap in genetic engineering (e.g., CRISPR) to make biology that much easier to program. That genetic information and manipulation techniques are more powerful when shared and built into a lingua franca.
Ginkgo believes in a future where we can grow almost anything. And we believe that too.
We’ve tried to position this piece as a balanced starting point for our continued work on Ginkgo. We’re not naïve to the myriad obstacles facing Ginkgo and synbio companies more generally — manufacturing handoffs, scaling bioprocessing, related-party revenue concentration, the uncertainty surrounding a new business model, the nascency of the field overall, just to name a few.
While we’re still deciding the best course of action, we may update this piece with bull and bear arguments, a valuation framework, etc.
© 2021, ARK Investment Management LLC. No part of this material may be reproduced in any form, or referred to in any other publication, without the express written permission of ARK Investment Management LLC (“ARK”).
The information provided is for informational purposes only and is subject to change without notice. This WIP piece does not constitute, either explicitly or implicitly, any provision of services or products by ARK, and investors should determine for themselves whether a particular investment management service is suitable for their investment needs. All statements made regarding companies or securities are strictly beliefs and points of view held by ARK and are not endorsements by ARK of any company or security or recommendations by ARK to buy, sell or hold any security. Historical results are not indications of future results.
Certain of the statements contained in this WIP piece may be statements of future expectations and other forward-looking statements that are based on ARK’s current views and assumptions and involve known and unknown risks and uncertainties that could cause actual results, performance, or events to differ materially from those expressed or implied in such statements. The matters discussed in this presentation may also involve risks and uncertainties described from time to time in ARK’s filings with the U.S. Securities and Exchange Commission. ARK assumes no obligation to update any forward-looking information contained in this WIP piece. ARK and its clients as well as its related persons may (but do not necessarily) have financial interests in securities or issuers that are discussed. Certain information was obtained from sources that ARK believes to be reliable; however, ARK does not guarantee the accuracy or completeness of any information obtained from any third party.
