All right. Good morning, everyone. I'm Tejas Savant. I cover the life sciences space here at Morgan Stanley. Before we get started, for important disclosures, please see the Morgan Stanley Research Disclosure website at morganstanley.com/researchdisclosures. And if you have any questions, do reach out to your sales rep. So it's my pleasure to host Ginkgo Bioworks today, and from the company, we have Anna Marie Wagner, Senior Vice President of Corporate Development. So thank you for joining us, Anna Marie.
Happy to be here.
Maybe, you know, just to kick things off, can you share some of Ginkgo's sort of key accomplishments over the course of the past year, and what are you most excited about as we head into 2024?
Sure. So I'd probably highlight a couple of real growth levers for the company that are really taking shape right now. The first, and we've talked about this quite extensively publicly before, is just our ability now to very credibly sell into biopharma.
Mm-hmm.
I think many, many folks who think about a life sciences platform would expect that our customer base would be effectively all pharma. But in fact, for us, it was the last market that we really tried to enter, and we're seeing just a huge amount of momentum in that space. And for us, it's a real leading indicator for the quality of the platform that we've built. And seeing these very technically sophisticated customers choosing to trust Ginkgo with their hardest R&D challenges is a really important indicator for us. The second is really around AI.
This is an area where Ginkgo really hasn't spoken too much about what we do in AI historically, because I think really until the last year or so, the conversation around AI in the life sciences was focused on the algorithm.
Mm-hmm.
It was focused on the technologies that were then being applied to public databases and small specialized datasets inside certain companies. Our view was always that the algorithms were not what was going to differentiate-
Mm-hmm
... computational advances in the life sciences. That the missing variable was really on the data side and not on the tool side. And so I think what's changed in the last year is the broader ecosystem, I think, has recognized the power of that data. And so if you think about what Ginkgo has built, it is this combination of very large metagenomic databases that we've acquired and built over time, that are the perfect substrate for training foundation models in biology.
Mm-hmm.
As well as a Foundry, which, if you think about it, is effectively a reinforcement learning tool. It's a, it's a labeled data maker. And so as we think about AI going forward, our ability to both really advance the foundation models that are available in biology, but also, and perhaps more importantly for our customers, develop these fine-tuned applications that answer commercially relevant questions-
Mm-hmm
... I think a real opportunity for the field broadly, and an opportunity for Ginkgo to really start showcasing the value of the data that we've been building, and that we're able to generate, to do the reinforcement learning for these models.
Got it. And so obviously, I mean, that's a great segue into, you know, your partnership with Google Cloud. Can you just talk about, you know, A, I mean, what, what do you bring to the table? What does Google bring to the table? I think there were two components to it, right? I mean, there was the cloud computing piece, and then they invested, I think, $56 million in non-dilutive financing. So just in terms of, you know, the cloud computing piece to start with, walk us through, you know, how that works. Are there any minimum purchase commitments associated with the contract?
Yeah. So to make foundation models, you need three things.
Mm-hmm.
You need talent-
Mm-hmm
... you need data-
Mm-hmm
... and you need a lot of compute.
Mm-hmm.
The new architectures for AI have really opened and shown the world how valuable running these models at scale is. And that's very different, I think, from the way that the life sciences industry has approached AI historically, where it's been sort of marginal regression on very specialized datasets. And so as I'm sure everybody in this room has heard, accessing compute is not a given anymore. As OpenAI has really shown the world what is possible with artificial intelligence, you've got every industry really figuring out, "What is my AI strategy, and how do I-
Mm-hmm
... how do I lean into this?" And so compute has become a scarce resource. So the first part of our partnership with Google is, as you alluded, our securing of next-generation compute capacity. So, we are committing over $250 million over the next five years to Google Cloud in order to get access to their next-gen compute, which they call TPUs, at, you know, attractive prices, so that we can make that infrastructure available to our customers. In the same way that we've built the rest of our platform, you know, our Foundry is a, you know, scale investment in the kind of infrastructure that allows us to then run very low marginal cost experimentation.
Mm-hmm.
Right? And AI is just the next generation of that. I think as Google looks at Ginkgo, we're a pretty unique partner for them because, you know, for them, it's strategically important that they get the life sciences industry using computational methods. And Ginkgo sits at an interesting place in that ecosystem, where we can actually draw the rest of the industry onto their platform by drawing them into Ginkgo.
Mm-hmm.
As we're building foundation models, as we're building applications, and whether we're running those internally or we're exposing them on, on Google's marketplace, that is helping draw additional customers into that ecosystem-
Mm-hmm
... and obviously driving long-term cloud usage. Then the other, the last piece I would say is, and one of the reasons Google really leaned in and sort of saw Ginkgo as a strategic partner, is their understanding of the importance of the Foundry-
Mm-hmm
... in supporting the development of AI tools. We are at the very beginnings of AI and life sciences, and so, like, if you talk to their DeepMind team, which is an incredible team, and these are the folks that created AlphaFold and, and, and what we think of as the leading biological AI tools today, what that team lacks, they've got the talent, they've got compute. What that team doesn't have is a wet lab that can test the 1,000 different predictions that their model just spit out. And that's what Ginkgo's Foundry is, is a way to test hypotheses. And so again, our ability to very quickly improve these models.
Mm-hmm.
through physical experimentation in the Foundry was an important realization, I think, for Google, and helps them lean into that $50 million engineering funding.
Got it. I mean, it's early days still, you know, since the announcement, but has it changed the tenor of any customer conversations?
It absolutely has. Yeah. I think it's... Already, even before this announcement, a lot of customers were looking at Ginkgo as their AI play-
Mm-hmm.
Simply because of the Codebase that we've amassed. Sometimes Codebase can be valuable because it's obvious we've had success in a similar area. Sometimes Codebase is valuable simply because it helps us get to the right hypotheses faster.
Mm-hmm.
So just that large corpus of data. And so this just accelerated that understanding. And, you know, all of these big companies, again, are trying to figure out their strategies. If you think back to those three things you need: talent, data, compute, Ginkgo clearly now has the compute.
Mm-hmm.
Ginkgo clearly now has the data, and from a talent perspective, Ginkgo's been working in AI, in biology specifically, for many, many, many years. And so for our customers who are now just trying to figure out their strategies here, that's an important set of components for us to help accelerate them.
Got it. Makes sense. Switching gears to cell engineering, you know, one of the key points of focus for investors has been the change in your contracting strategy lately. How much of that was a proactive choice versus a reactive one, reflecting sort of current market conditions?
Yeah. So the success-based pricing was very much proactive on our part.
Mm-hmm.
As the analogy that we talk about is the evolution that we've seen in the advertising industry, where, you know, there is much more value to be captured if you can demonstrate a high ROI.
Mm-hmm.
In the life sciences, and as in really any industry, the thing that our customer cares about is probability of success. The reason they're coming to Ginkgo across all of our programs is because they believe we can deliver a higher and faster chance of success. What success-based pricing does is it sort of answers that question for them on day one.
Mm-hmm.
Like, you are not actually taking the risk of success. We are... If you are paying for this, you're getting success, and that's a very different value proposition than they can get anywhere else. It's a very different value proposition than funding it internally with their own team or than anyone else is able to offer. And so the question for us was really, at what point can we offer this type of a differentiated value proposition? And we were able to find in certain areas of the work that we do, that we have a high enough success rate and a predictable enough success rate in areas that are relatively efficient for us to run-
Mm-hmm
... where we could make those kinds of bets. And you know, to our last conversation about AI, it is not a coincidence that enzyme engineering, which is the area where we've offered success-based pricing-
Uh-huh
- is our most computationally driven set of work today. That is, that is driven by our proprietary AI tools today already, and so we are, we are able to both predictably and efficiently deliver those, those projects.
Got it. One of the questions that... Yeah, it's a tough question to answer, but that I've gotten is, if, if you guys had stuck to the cost plus model, what would those 66 program adds you have in the back half of the guide look like?
Yeah, so maybe put another way, how much do we see the success-based pricing sort of,
Yeah
... impacting our program mix?
Mm-hmm.
We certainly see it growing. It's a valuable offering in the market, and what it's done is it's helped us reengage with some customers-
Mm-hmm
... that, you know, maybe we lost because they, they were underwriting a lower probability of success than we were, and so the ROI didn't make sense for them initially.
Mm-hmm.
And so it certainly has been important. I don't think it's, you know, we said on our earnings call, you know, that it, it's not gonna be the bulk of our programs by any means. You know, it's still a relatively small portion of our overall programs, but we are seeing growth there, and I think that's sort of reinforcing the value that this is, this is adding to the market. One thing I will just highlight is, again, our view here is that this is not a pricing change.
Mm-hmm.
That this is a different, really timing model, if you will, on success-based pricing. If you think about what happened in the advertising industry, like when Google introduced pay per click rather than just pay per view, they were able to charge much more for that. And so if you think about the net present value of those two offerings, if the service provider is better-
Mm-hmm
... at providing that service, if we have a higher probability of success, if Google better targets advertising, then there is value there to be captured because the customer is underwriting a very different probability of success. That's what we see in something like enzyme engineering, where we actually believe that there is the opportunity for us to improve the value of our programs at Ginkgo, while providing the customer with a differentiated value proposition.
Got it. And Anna Marie, when you talk about success-based payment, do you include the fixed and milestone-based contract structures also under that umbrella, or is that just the payment at the end, if there is success?
Yeah, no, success-based payments are really just a payment upon, and this is important, technical success. So those programs-
Got it
... would still have the traditional economics of, you know, upfront services value and sort of downstream economics in the form of commercial milestones or royalties. But the technical payment-
Mm-hmm
... that services fee is, is delayed until the technical success has been reached. For our, for our traditional programs, there is a payment for the upfront services, regardless of technical success. The way that those can be structured is as a sort of firm fixed price, and, and then you recognize revenue as you progress, you know, the percent completion of the program, or cost plus, where it's, it's a little bit more open-ended. And then that can have some revenue recognition implications, but not otherwise material.
Got it. Makes sense. You sort of alluded to this in terms of, you know, starting with enzyme services and protein production, because that's where you see the most predictable success rates. How do you see that evolving over time, and what could be next to, you know, go through the same sort of contracting change?
Yeah. So there's nothing immediately on the horizon.
Mm-hmm.
To be clear, I would suggest that, again, that framework of you need to have a high enough probability of success-
Yeah
... but even more importantly than that is the predictability-
Right
... of the success, right? So if we can get very good at predicting, okay, we know that some percentage of these programs are gonna succeed, that allows us to price appropriately. So that predictability to me is the most important driver.
Mm-hmm.
I would expect that to the extent that there are additional services where we're able to offer that, it will probably be in our more traditional microbial engineering pathway, simply because those are our more mature Foundry workflows, which leverage the sophisticated automation as well as our AI tools. It will probably be longer before we're able to offer those types of models in mammalian cell engineering, simply because those are more bespoke projects today.
Got it. And is that an internal dashboard that, you know, you or Mark look at, where, you know, you have an internal red line in terms of how much at-risk work you'll take on? Or does this really go on a project-by-project basis? Like, what I'm thinking is an enterprise-wide sort of risk management metric, where, you know, you're focused on, you know, this much work at a minimum being cost plus.
Yeah. So, again, today, it's not a big enough-
Right
... event to really worry about.
Right.
I think certainly if we started seeing that our underwriting and success rates or something was not accurate-
Mm-hmm
... that would become more relevant.
Mm-hmm.
Again, the only places that we've launched this are in areas that are highly predictive and very high probability of success, so-
Got it
... I don't consider those overly at risk. But yes, certainly we do look at the mix of our programs, and we like to keep a balance between those archetypes.
Got it. Another sort of investor inbound has been in light of, you know, the decoupling, so to speak, between program adds and sort of NTM revenue growth-
Mm-hmm
... would you sort of disclose, you know, not just the program adds, but also the fraction of those that come with, you know, next 12 months' revenue?
I haven't heard that specific question before. I can think about that. I do think that there is value in helping people understand the different models-
Mm-hmm
... that make up our program mix.
Right.
And we have, over time, we've provided more details on what the mix of programs looks like. Most recently, we disclosed how the downstream economics of our programs-
Mm-hmm
... are shaped, whether it's equity, milestones, royalties, some combination of those. So I think, you know, the real ask here is how many of our programs that we signed were success-based pricing, and that feels like a very reasonable disclosure-
Got it
... to start making.
Got it. You know, on the last earnings call, you also talked about, you know, the ops team finding that the Foundry could support, you know, 2-3X the demand with relatively low amount of incremental investments. What are some of the levers identified to improve, you know, overall equipment and people effectiveness?
Yeah. So, I mean, this industrial engineering team literally went around the Foundry, you know, with clipboards and would watch a machine for a few hours, and we now... or even days, like, how often is this machine running? How much of this person's time is being spent on sort of productive engineering tasks? So it really was a very bottoms-up build across every piece of equipment, every function within the Foundry.
Mm-hmm.
I'd say, honestly, the biggest one is. So, there are these two components, equipment utilization and people utilization.
Mm-hmm.
On the equipment side, it really is just utilization, and what enables that is really more on the software side. And so just making sure that we are more efficiently scheduling-
Mm-hmm
... it becomes a scheduling problem more than anything else, scheduling our experimentation such that we can leverage our existing capacity more efficiently, rather than just, oh, we'll add another, you know, liquid transferring robot because-
Right
... it helps, you know, debottleneck this step. You know, we can think more, more carefully about, well, how do we debottleneck that step without adding the robot? And, and there is a lot of low-hanging fruit there-
Mm-hmm
... that's been identified. On the people side, some of that is around standardization of our programs, so creating a little bit less bespoke work that people need to manage, and being able to therefore leverage more of the shared infrastructure that we can take advantage of. And then some of it is just giving people the tools to operate more efficiently and take away some of the busy work, the administrative work, and handle that more efficiently.
Got it. One more question, and then we'll switch to industrial biotech. You know, as you factor in these productivity enhancements with the new sort of contract structures, I mean, that's obviously gonna come with, you know, lower average revenue per program in the near term. What does that do to your cash flow, particularly, you know, not so much this year, where, you know, you guys have reiterated your cash burn target, but for next year?
Yeah. So, again, well, I'll, maybe I'll tell you about how we manage cash flow-
Mm-hmm
... in particular, and then I, I don't think that these recent, sort of changes really impact that overly much. But the way that we manage the business is, if, if you think about sort of sources of, of value, right? We get some of this value upfront for providing services, and we get downstream value.
Mm-hmm.
The way that we manage the business is, let's not rely on the fact that we think downstream value is gonna show up-
Mm-hmm.
- and it's gonna be very profitable. Let's manage the core services operations of the business to reduce burn, such that we always have multiple years of runway. Now, the reality is that it's entirely possible that a downstream opportunity hits-
Mm-hmm
... and pays off in a real way, and we may be able to adjust our pace of investment. But for now, we're really focusing on even putting aside the potential value on downstream. We're reducing burn in the core business-
Got it.
to extend runway, you know, as, as much as needed-
Got it.
until we reach that, cross that chasm.
Got it. In terms of, you know, industrial biotech, you know, end market, you talked about that being a little bit soft. How much is your industrial biotech exposure versus pharma and ag-related biotech?
Yeah. I would... It's oversimplification-
Yeah
... but at least historically, it's been about, in recent quarters, it's been about a third, a third, a third.
Okay.
And really the growth has come. We've seen growth in all segments, but the growth has really come from pharma, which just a couple of years ago was-
Right
... very little of our business. But we are seeing growth across all segments.
Mm-hmm.
I think in industrial biotech, you know, if I think about the growth drivers of those industries, in pharma and in ag, you have most of the R&D work is being done in-house, and they are already quite biological industries, if you think about it that way. And so the driver there is really, are those companies willing to outsource that R&D to Ginkgo? In industrial biotech, you have lots of very large companies that have great R&D departments, but that R&D is not focused on biology today.
Mm-hmm.
And so there, it's much more a question of biological penetration rather than the willingness to outsource. The willingness to outsource is sort of there, that they don't have the in-house solution. And to me, that's more a question of, at what point can we engineer really competitive products with biology-
Mm-hmm
... that compete with traditional extractive, technologies? And so you, you've seen these cycles of, oil prices are high, and then the industrial biotech industry booms for a little while. Oil prices drop, and suddenly those products are no longer cost competitive. We know that biology is capable of being very cost competitive. We just, as an industry, don't have the tools yet to make those products. And so I think as the tools mature, as our capabilities advance, that will be the natural driver of that industry.
Got it. And in terms of just, you know, the impact from higher rates and, and VC funding, et cetera, any kind of, like, bottoming out you're seeing in industrial biotech or, or too soon?
I'm not a timing the market person. But no, we certainly have seen capital to industrial biotech has dried up. And so you know, that startup community in industrial biotech is having a tough time, for sure. I do think, again, in the long run for Ginkgo, I think it presents an opportunity where we are the variable cost in what has historically been a fixed cost industry. And when you have unpredictable environments or challenged environments, finding those efficient and flexible-
Mm-hmm
... providers is valuable. And so while it's certainly a painful period for this sector, I do expect it to drive some longer term benefits to our model.
Fair enough. You know, in terms of the collaborations that you've announced with Merck and Novo, can you talk a little bit about how those came about, and how did you sort of demonstrate the value add versus, you know, the pretty high bar for their in-house capabilities?
Absolutely. You know, the reality is, there are many different R&D sort of priorities and focus areas for these companies. It really just takes one part of the business to start sort of seeing the value. So what we've tended to see with these very large and technically sophisticated customers is we'll get in the door in one place. So, for example, with Merck, it was in biocatalysis.
Mm-hmm.
We've done enzyme engineering across all of our, you know, all of our sectors. We're very, very good at that. And so our first program with Merck was in enzyme engineering, effectively for a biocatalysis program. Well, you know, that established Ginkgo as sort of a trusted partner to Merck and a trusted service provider. And as we demonstrate success in that early work, it becomes very easy for those initial believers inside Merck to go march us around and introduce us to a lot of other R&D leaders. And so with the larger pharma and the ag companies, for that matter, it's really a matter of establishing credibility in one niche and then expanding throughout the rest of the organization from there. And so with Merck, you see us adding several more programs-
Mm-hmm.
with them this year, following that initial work. You've seen the recent announcements with Novo around our kind of initial success with them and the expansion of that program. So, great, great progress thus far with that segment, which again, to me, is the most important leading indicator for our business.
Just from a balance sheet/top line standpoint, which one of these is the most impactful needle moving, and how are the sort of upfront versus milestone structured?
... for Merck and Novo specifically?
Yeah.
Yeah, I can't comment on the specific economics of those transactions. You know, our big pharma collaborations in general do tend to follow the traditional model there, with a healthy upfront and then, you know, especially if we're developing a drug with them, real development and commercial milestones-
Mm-hmm.
And royalties thereafter.
Got it.
Leave it at that.
All right. Switching to biosecurity, you know, can you talk a little bit about the conversions between the cell engineering and biosecurity business? I mean, that's been a point of emphasis in a lot of conversations for Jason and on the last few earnings calls. And what can you point to in terms of the early wins there, on the synergy sort of front?
Absolutely. So we're in this sort of transition right now from kind of wartime to peacetime, if you will, in the biosecurity business. The pandemic was that sort of response focus, and it was all hands on deck, figuring out how to test, how to isolate, how to create vaccines and therapeutics. We're now in this really interesting moment where we can think about what is the right infrastructure to build for pandemic awareness and prevention going forward. And the types of questions that you wanna be able to answer to do that have a lot of overlap with the types of questions that we answer every day in our cell engineering business. And so the biosecurity component of that is really around endpoint detection, right?
So how do we get around the world and start collecting data, have these really important nodes so that we can create the map and the radar system, if you will, for biology-
Mm.
around the world.
Mm-hmm.
But then the types of questions you're asking are things like: All right, for this piece of DNA that I found in the world, what does it do? Is it something I should worry about? Is it, is it something totally harmless, or is it a little scary? If it's scary, where did it come from? Was it engineered, or is this just evolution throwing a curveball at us? All right, if it's something I need to be worried about, how do I address it, and how do I make vaccines that prevent against this? What are the therapies that I can create? And, and that last set of questions, those are all the questions that we deal with in, in cell engineering.
Mm.
So I get really excited because I look at biosecurity, the biosecurity business, and that endpoint detection element of that business as a generator of huge amounts of data-
Mm.
-that then can feed into our overall platform, and especially as we start leaning into AI, feed our ability to then also answer some of those downstream questions as well.
Got it.
which we've already done, by the way. You know, we've had a collaboration with IARPA around answering some of those-
Yeah
... questions around, you know, is this sequence engineered? So we are seeing that convergence. And I think, you know, we're sort of in that early phase of designing biosecurity programs with countries around the globe. And I think, you know, I'm looking forward to the coming years, where we, I think, switch into that implementation mode and really-
Mm
Start collecting that information.
Was the sequence engineered?
Sorry? I... No comment, no comment.
How much of biosecurity revenues are K -12 testing services related, and do you expect any upside from the recent spike in COVID case counts at all?
So if you look at the first half of the year, the bulk of that is-
Yeah
... is still, K-12 testing. You know, we expect the second half of the year, and schools just started, you know, the public health emergency is declared over. We do not expect there to be significant K-12 testing in the back half of the year.
Mm.
So what we see in the back half of the year, we expect will be more that the beginnings of that sort of infrastructure building phase. In terms of the recent COVID spike, we're obviously still working with the US CDC on variant tracking, and I think the real question will be, you know, are we still in a mode where we can treat COVID a little bit like the flu?
Mm-hmm
... and, you know, folks are gonna do all right? Or are we starting to see variants of real concern that are both dropping out of the, out of the vaccines, are not getting picked up on the existing tests? And, and do we need to really lean into a more concentrated monitoring effort there as well?
Got it.
But too early to tell.
Got it. And then in terms of just the gross margin implications with the more recurring portion of biosecurity revenue, how are you thinking about that on a go-forward basis? And the new international contracts being signed on surveillance, are those more service revenue or product revenue?
Honestly, I think it's probably a little bit too early to tell.
Mm
... just given the scale of that business. I would expect that, again, if you use the peacetime/wartime analogy, I would expect that infrastructure, that peacetime business, because it's very stable and it's very recurring, can sustainably operate at a bit of a lower gross margin.
Mm-hmm.
And then you'll, you'll likely see higher margins in the rapid response elements, to the extent that something does need to be.
Got it. Got it. Turning to the numbers here in the last minute or so, Anna Marie, just, you know, your cell engineering guide of $145 million-$160 million, can you talk about how you're thinking about the cadence there and just some sort of drivers of upside/downside?
Sure. So, again, I'm trying to like... I would likely just refer folks back to our, our last earnings call. The leading indicator for us is always new program additions.
Yeah.
So that's what I would look to. Obviously, the mix of those programs can have some impact.
Mm-hmm.
But the real unlock for us, honestly, in going back to one of your earlier questions, in terms of revenue, has been some of these operational investments we've made. You know, for a while there, for a few quarters, you know, we were having trouble kind of breaking out of the kind of $20 million-$25 million service revenue range.
Mm-hmm.
And some of these steps that we've been taking have really allowed us to drive that. And so it's really around the pace of our revenue recognition, in other words, the pace of our progress on these programs, is driving some of our revenue as well. And so those two are the biggest drivers.
Got it. Last five seconds, quick preview of the Investor Day. You know, you're working hard on it, I'm sure.
Yeah.
Um-
Investor Day, October 3rd. If folks are gonna be in Boston, please, please shoot me a note. We'll do some Foundry tours for folks that are in the area. You know, we'll do an overview of our core businesses and meet the leadership of our cell engineering and biosecurity businesses. And then we will do a deeper dive in AI, because that's a newer part of the conversation for our investors.
Mm-hmm.
So we wanna make sure folks really understand what those differentiated assets are that we believe allow us to win.
Awesome. Thank you so much.
Thanks for having me.
Appreciate it.