Ginkgo Bioworks $dna: AWS for Synthetic Biology?

Thanks to all who are reading this week - criticism is welcome.

It’s hard to overstate what AWS did for the ease of launching a startup. The ability to rent a server after simply pressing a few buttons and entering payment information lowered both the logistical and financial costs of starting a business. It also made venture capital a more compelling asset class. It’s easier to get excited about investing in a seed-stage business when the pitch isn’t only a slide deck asking for money for servers based on an idea! The risk of failure is still high, but if the limited beta-rollout of the product was met with enthusiasm that’s a more compelling signal than just a PowerPoint presentation. Moreover, lower startup costs meant more people could/were willing to try ideas, which probably lead to a greater absolute number of tech startups succeeding, which further heightened excitement about venture, etc.¹

Over the past two decades, costs have not been lowered in the same way for the biotech industry. For a biotech startup, pitching a VC still means asking for money to set up or run a lab, plus money to hire people who are skilled enough to work in that lab effectively. This initial funding challenge is even more damaging for this category than it was for the tech equivalent. To invest in early-stage biotech is, assuming the investment succeeds, to accept that substantial equity dilution will happen along the way (Getting FDA approval is an expensive and lengthy endeavour!) The high initial cost to simply get going is an additional barrier that makes the sector seem even less appealing.

Ginkgo Bioworks is changing this dynamic for companies in the synthetic biology space. Its business can be divided into two parts:

Cell Engineering – this is the company’s bread and butter, and what you’re betting on as an investor. If I have a product that leverages synthetic biology, I go to Ginkgo for help designing promising synthetic DNA sequences and then testing those DNA sequences in prototype cells. During this process, Ginkgo leverages its Codebase and Foundry. The Codebase includes genetic sequence data and experimental data, as well as DNA sequences and host cells that the company’s designed in the past. The Foundry refers to Ginkgo’s lab, which makes heavier use of automation than an in-house equivalent would, resulting in lower R&D costs per unit of work. The term ‘Foundry’ is intentional; it’s not completely inaccurate to think of Ginkgo as a kind of TSMC analog.

Biosecurity – The company’s solutions for identifying, responding to, and preventing biological threats. This segment went on a temporary tear during the pandemic due to Covid testing, but has since come back down to earth; revenue dropped by 48% from 2022 to 2023 and is set to drop even further this year. However, Ginkgo’s biosecurity customers are primarily national governments, which make the segment a great long-term strategic play. If synthetic biology goes mainstream, management will be in a great position to lobby for regulations that benefit its cell engineering business.

The simplest way to explain synthetic biology is to contrast it with how biological changes typically happen. In the natural world, when DNA replicates there are occasional mutations. Some of these mutations are beneficial to an organism, some are not. The organisms with advantageous mutations go on to survive and reproduce.² Rather than waiting for evolution to happen, synthetic biology allows scientists to make alterations to DNA within a lab. These new strands of DNA can then be inserted into a cell, with the hope that it will enable/disable a certain function. The potential applications of synthetic biology are vast, and extend beyond pharmaceuticals. The technology can be used to treat cancer, but also to improve crop yields more effectively than existing chemical fertilizers, and, more entertainingly, make petunias that glow in the dark.³ It’s also expensive and time consuming, which is where Ginkgo’s Cell Engineering arm comes in. Rather than raising money for a lab, it’s more cost effective to outsource some initial R&D to Ginkgo. This has a few potential consequences:

(1)   I can ask investors for less money, and so have an easier time during the fundraising process.

(2)   I can ask investors for the same amount of money and use the additional funds to do more design/experimentation rounds with Ginkgo, increasing my chance at success.

Even if I pick option (1), my chances at success still go up. Ginkgo’s Codebase means the company has access to far more data than I do, and so has a better chance at designing a synthetic DNA strand that works. In the wake of OpenAI’s success, management has been keen to harp on this data advantage. It’s easy to be cynical about this emphasis: the best way for a biotech company that went public via SPAC in late 2021 (and whose stock is down ~90% since then) to rebound is to argue it’s an AI beneficiary. In Ginkgo’s case, this claim is legitimate. The company has 10x the genomic data that public data sets have, a magnitude that’s especially significant as AI seeps into the biotech world! Crucially, management notes that ‘biology did not evolve by end market.’⁴ In principle, there’s no reason why a lesson learned from serving Bayer’s agricultural arm can’t be applied to a pharmaceutical customer.⁵

Gingko’s revenue model on the Cell Engineering side requires some explanation. Customers are charged service fees for its Foundry services, but the company also negotiates downstream payments in the form of equity, royalties, or milestones. That last part is important to understand. Ginkgo works with customers to design and test initial product ideas; if it helps Novo Nordisk develop a drug that later gets FDA approval, Ginkgo doesn’t then scale its manufacturing as that drug goes to market, thus benefiting from additional fees. In other words, the AWS and foundry analogies actually break down quite quickly, as Elliot Hershberg has pointed out:

“The problem is that once a customer has their designed organism, they can use it to grow as much of the programmed output as they want without Ginkgo. Imagine if you could simply grow more servers, for free, once you pay to get set up on AWS.”

So in important aspects Gingko’s business model isn’t AWS’s, nor is it TSMC’s. That said, these downstream payments should ensure the company benefits from the success of its customers. These will take years to materialize in a substantial form, should they materialize at all, but if synthetic biology enjoys massive success Ginkgo should too.⁶ Typically, the company receives equity from early-stage businesses and royalties/milestone payments from more mature companies. These royalties and milestone payments are somewhat more preferable ; one never knows if a drug will get FDA approval, but Pfizer’s less likely to run out of the capital to bring a drug to market than an early-stage biotech startup subject to the cyclicality of the VC cycle. Over the past year revenue growth has started to come more from the big pharma side of things, which is an important validation of what Ginkgo is doing. Novo Nordisk and Merck signing on as customers means synthetic bio is interesting to more than only early-stage companies and VCs!⁷

Within cell engineering, Ginkgo’s service fees are also slightly complex upon a closer look, and have been the subject of at least one short report. The bears focus on two main points. Firstly, the company isn’t always paid its service fees in cash, but instead works with certain early-stage companies in what management terms a ‘Structured Partnership.’ Under the terms of these partnerships, ‘we offer flexible commercial terms on the service fees including the ability to pay a portion or all of such upfront fees in the form of non-cash consideration…. in addition to downstream value share consideration.’⁸ Secondly, concerns have been raised about Ginkgo’s ‘Platform Ventures,’ which entail partnering with investors to fund new synthetic biology companies: “In exchange for an equity position in the venture, we contribute license rights to our proprietary cell programming technology and intellectual property, while our partners contribute relevant industry expertise, other resources and venture funding…… We also provide R&D services for which we receive cash consideration on a fixed-fee or cost-plus basis.”⁹ The worry from short-sellers is that both types of partnerships don’t indicate real demand for Ginkgo’s services. If I’m an early-stage customer offered ‘free’ cell engineering services in exchange for equity in my business, then of course I’ll use Ginkgo rather than build a lab internally. Similarly, if Ginkgo introduces me to funding partners and offers me access to valuable IP, it makes sense to then pay for its cell engineering services. Additionally, if Gingko and a startup share an investor there’s an incentive for that investor to encourage the startup to use Ginkgo. There’s no clearer value-add than landing a portfolio company a customer, and better revenue/customer growth means a better valuation for a company within the portfolio.  

I think these concerns are overstated. The market Ginkgo operates within is in early stages. While it’s true that these partnerships have the potential to overstate demand, trying to grow the market that one provides services to is a perfectly rational objective for a company to pursue!¹⁰ Microsoft offers cloud credits to AI companies it invests in, and while this obscures true demand it would be a stretch to deride it as a ‘dubious shell game.’¹¹ The more AI customers Microsoft has, the more customer feedback the company gets, the better of a product it can build, the more likely Microsoft is to win market share over time, etc. Enticing customers to use your service in the first few innings of an industry can lead to market dominance later on! Ginkgo’s ‘Platform Ventures’ also solve for the temptation platform biotechs face to verticalize once a promising product is found. It’s a historical anomaly that Moderna remained a horizontal mRNA platform after developing a blockbuster vaccine. More typically, a company would’ve jettisoned the rest of the business to go all in on that one success.¹² This historical precedent can lead potential customers to be skeptical of biotech platforms; using AWS as a startup would be a less enticing proposition if you knew you could be cut off from the service at any time. Ginkgo partnering with investors to create new companies solves for this risk while enabling the company to benefit from ideas it might have internally; it would be surprising if a synthetic biology platform didn’t have a few product ideas of its own!

One takeaway from both types of partnerships is that one should really analyze Ginkgo as a blend of both a more typical business and a biotech investor that does an unusual amount of value-add for its portfolio in the early days. Consequently, one should look closely at reported cell engineering revenue numbers to determine the mix of cash and equity. At this point in time, equity revenue is both illiquid, and so doesn’t fund Ginkgo’s operations, and extremely volatile. When the biotech funding market falls of a cliff, as it has in recent years, Ginkgo’s forced to significantly write down the value of equity it’s received from startups. Cell engineering revenue stayed about flat from 2022 to 2023, at ~143mm. However, engineering revenue related to non-cash consideration decreased from ~76mm to ~49mm, a step in the right direction when one is focused on the actual cash the business is receiving. In that same vein, management’s fortunate to have almost a billion in cash on the balance sheet; they’ve spent the last few years cutting spending, acquiring early-stage companies, and acquiring Ginkgo’s main external competitor, Zymergen. If and when the biotech funding market turns around, these opportunistic actions should pay off.

Disclaimer: The information in this post is not intended to be and does not constitute investment or financial advice. You should not make any decision based on the information presented without conducting independent due diligence.

1. There’s of course a point at which too much experimentation is going on, creating a lot of waste in the system.

2. This is a massive oversimplification, as an introductory biology class would illustrate! Plenty of mutations happen in non-coding portions of DNA, and so don’t affect the proteins being transcribed, etc etc.

3. To cite an example from Ginkgo’s most recent earnings call.

4. Page 74 of Ginkgo’s annual report.

5. While biology didn’t evolve by end-market, selling to pharmaceutical companies did, a lesson that Ginkgo’s learned over the past few years: “The conventional wisdom is that specialization and prior expertise in an area is really the only thing that matters and so customers either specialize themselves or seek out specialized companies for R&D partnerships. As a horizontal platform, Ginkgo is often not top of mind when customers are thinking about who to work with in specific areas……..As such, one of our highest priorities for 2023 is doing a better job speaking to customers directly in their market, highlighting the successes we’ve had in their space. We took a small step towards this in 2022, when we defined our “Ginkgo Enzyme Services” product and we see value in replicating this model in other areas. In fact, at our recent Ferment conference, we announced several more specialized services in RNA therapeutics, AAV gene therapy, cell therapy, and microbial engineering.” - Ginkgo’s 2022 Annual Report.

6. This assumes that Ginkgo has customers that participate in this massive success. I think this is reasonable given the broad applicability of synthetic biology, which increases the likelihood that there aren’t only one or two players who benefit from the trend.

7. This doesn’t mean that incumbents will necessarily benefit from synthetic biology. Barnes and Noble announcing it was going to build a website during the dot-com boom didn’t result in the company benefiting from the internet, but it was validation that the internet mattered enough for a behemoth bookstore to direct attention towards it.

8. Annual Report, Pg 76.

9. Annual Report, Pg 75.

10. See Slack’s founder’s piece, We Don’t Sell Saddles Here, for a good explication on this.

11. Page 3 of the short-report linked above.

12. In Moderna’s case it’s very fortunate that management didn’t do this, given how much the outlook for Covid vaccines has changed since 2021.