Twist Bioscience Corporation (TWST) Earnings Call Transcript & Summary

All right, guys. Thanks so much for joining. We are starting now our innovation across drug discovery, molecule synthesis and precision medicine. So we've thrown everything that we possibly can on the title. Hopefully, we've encapsulated enough. But Liza Garcia, UBS life science tools and diagnostics analyst. I'm being joined by my co-coverage John Sourbeer and then we have Sean McClain, the CEO of Absci. We have Patrick Finn the Chief Operating Officer and President of Twist, we have Anna Marie Wagner, SVP, Corporate Development at Ginkgo, and we have Ross Muken here as well, the CFO of SOPHiA GENETICS. And so we have quite a number of companies to take it off. And I think we'll start on some broader themes and then we'll drill down to the company level questions. But just to start it off, obviously, maybe we'll go down the row on this, from Ross -- from left to right. But -- all right.

So a broad one to start. So just thinking about utilization of data and new technologies to better deliver on molecule synthesis, drug discovery, patient care, why don't you kind of start -- let's start with the quick intro actually, just so that everybody can kind of get the scene and how -- and a high-level overview of how your company sort of -- you feel like you're tackling these areas.

Perfect. Thank you, and thank you for having us. So here at SOPHiA, we're really focused on what we would consider data-driven medicine, right? So this is the incorporation of community insights around the globe into all elements of the care continuum, including obviously, clinical care at the bedside, if you're an oncologist, but also in the drug discovery and drug development and commercialization process. And so for us, it's really around the animation of data at scale. We started in genomics now that we've moved into what we would consider multimodal data. we can support radiology, so images, EMR data, pathology data, any type you want being cleaned and organized, right, in a common data lake and then used for the purposes of drawing insights, whether you're trying to figure out the correct patient population for your new IO drug or how -- what therapy someone should be on, right, in a category where there's many options or if you're trying to figure out why I had super responders or nonresponders in a given area. So there's a full spectrum of use cases. But for us, we're trying to embed this collective intelligence and the power of being in 750 institutions in 70 countries in the world, producing that data at scale and what you can learn from it. And the algorithms we have that power that, enable that in a more predictive way to aid in all elements of what you sort of described. So that's at SOPHiA in a nutshell and how we're positioned in this space.

Great. Great. Thanks. So at Ginkgo, we're really focused on the early side of drug discovery, although it can also play across the value chain. And our mission is to make biology easier to engineer. So what does that mean? It means making it predictable, making it more like an engineering or science discipline than just an art. And the way that we approach that is really on the one side, trying to decouple the R&D process from manual human labor. So that you're driving scalability, accuracy and quality data generation at a lower and lower cost over time as you scale. And then on the other side, capturing that data to better inform future experiments and to create reusable architectures and reasonable biological parts, if you will, so that we're not recreating the wheel every single time. And our view is that one of the things that's really held biotechs back over the last 40 or so years since we created our first major biological drug is the fact that the tools, techniques and IP that is enabling the industry is not being shared. And it's being siloed in companies that are focused on individual medicines in their individual space, but the reality is that a lot of that learning can be cross applicable across many different areas. And so Ginkgo is working on bringing that together and bringing together the tool set that is required to derisk the process of doing R&D in this space. As it relates to data, in particular, we do view data as one of our core assets, and the foundry, this automated lab is the core tool that we use to then generate that data. And we're able to apply that data in many different ways, creating better predictions in our traditional kind of AI and ML tools to design the DNA sequences that we're testing for our customers and to create very specific applications that we're then able to use across the platform. And then we're able to use the foundry to then test the quality of those predictions and iterate those tools over time. So I think it's a very, very interesting time to be sort of in our sector and at this sort of confluence of big data, the ability to generate lots of data at a lower and lower cost and then use some of these advanced tools to be able to really drive the space forward. We also buy a lot of DNA from this guy right here.

So yes, Twist Bioscience, we have a truly disruptive DNA synthesis platform that we think has really opened up in the application space. I think to describe the scale, we're capable of synthesizing tens of millions of DNA building blocks every day, which is opening up all sorts of interesting applications and business areas. So for that -- that's in synthetic biology, whether it's a clonal gene that's used by Ginkgo with prodigious volume, in pharma for antibody discovery, it's oligonucleotide pools feeding the gene editing space, libraries, antibody discovery, any product at all in the industrial biotech segment to break out off our dependence on oil. So we're enabling the community there. For next-gen sequencing, we have a target enrichment product and some workflow solutions that really truly maximize how people use sequencers. And what we've learned over the last few years of that product is that the clue's in the name, customers want to customize their assays. And so that incredible scale allows customers to get onto our platform and build an assay that they need for applications like liquid biopsy and other research applications. We have an antibody discovery program underpinned by our ability to make incredible amount of molecular diversity, whether that's synthetically, a recent acquisition of a hyperresponding mouse allows us to create even more diversity. And then we'll have a layer of AI or machine learning that helps with really improving the quality and the execution of those designs and ultimately, the final products we produce. And then last, by no means least, and pre-revenue is a really interesting application where we can take what we know in DNA synthesis really truly scale it up, orders of magnitude where we are or from where we are today to then incorporate that into the data storage space which although it sounds like science fiction, we're all living proof that DNA is a stable data storage medium, and we're starting to see the sort of cold data layer and actually be -- we're seeing the proof-of-concept experiments where people are now starting to store modest bits of data in DNA. Over to you.

Thank you, Patrick. And by the way, I like your socks, I need to get some protein socks. I'm Sean McClain, Founder and CEO at Absci, we're a generative AI drug creation company, really going from this paradigm of drug discovery where you're searching for a needle in a haystack to drug creation where you're actually creating the needle in our case, a biologic. Now there's been a lot of buzz around generative AI and like what does that actually mean when it applies to discovery of biologics. Maybe we can just take a step back and look at how have antibodies traditionally been discovered. I mean, Regeneron was really the first company that developed humanized mice. And you inject -- you do immunization, you inject a mouse with a target, the mouse generates the antibodies. Well, the issue with that is you have no control over what the mouse gives you. The mouse will bind to a particular epitope of interest, it will have a particular affinity, it will have developability parameters, but it may not be what you want. And what we've been doing here at Absci is using generative AI to actually specifically design antibodies with the attributes that you want. Hitting the particular area of the target you want, having the affinity, the developability and manufacturability. And ultimately, that's what's going to dramatically decrease the amount of time it takes to get new therapies into the clinic and increase the overall success rate because we have these very challenging targets that are out there, GPCRs, membrane protein -- GPCRs, ion channels that are really hard to hit with traditional means. And if you can hone in with generative AI to be able to hit the specific area that gives you the biology, gives you -- has the affinity that you want that gives you that particular biology. That's what's going to start unlocking that new biology and increasing overall success in the overall clinic. And one of the ways that we've been successful with generating this generative AI platform is exactly with what everyone else has been talking about, synthetic biology. We have this platform that allows us to generate billions of protein-protein interaction data points, that we used to train the model. And then we can also then validate the model in the wet lab, looking at over 3 million unique AI-generated designs in a given 6-week time period. So it allows us to train our models, validate them and make very rapid advancements for being able to use generative AI for de novo design or creation of antibodies.

All right. Great. So a very diverse group. But I guess let's start with numbers. As you guys think about kind of your companies and obviously, the challenges you're addressing, how should we think about kind of if you have any frame of reference or numbers, the potential cost savings that you're trying to deliver here and how to think about kind of output and obviously with data generation becoming more and more complex, how to think about kind of addressing those -- how you're addressing those issues? And kind of what you think about in terms of solvability for your customer set? I think we'll start with Ross. And then we'll go down the line again, and then John will go the other way.

So cost savings for us can be a couple of different ways you think about it, right? So on the core business for us where we're helping with sort of the production of genomic information at the sequencer side, right? If you think about an average LDT or other diagnostic. You're talking about probably reimbursement, let's say, in the U.S. -- and we're global, but in the U.S., let's say it's $3,000. Of that, the sort of prep and bioinformatic costs that Pat and I sort of help address is probably $1,000 of that right? So it's a pretty big number. However, that's based on folks with their own bioinformatic teams doing it internally and maybe they're using one of your competitors that's not as if scaled as Pat. And so if we come together or we can bring to the market a combination solution that allows for AI and ML and other pieces to automate some parts of the labor and then scale economies to lower the chemistry cost, you can take that down pretty materially, right? And I think most of us know that a lot of the labs don't make money yet even at scale. And so aiding on that side, just purely from an industrial standpoint, can help in the efficiency and I would say, ability to scale to many more patients, no matter the institution and can also flex down as we look in emerging markets to price points, and we were talking about before, India and other places, Africa, where you have to be at a very different cost level to be able to do that, right? So there's that part of savings. But I would say that the sexier part for us, when you think about today, and again, a few folks are talking about the clinical trial process, as most of you know, if you look at a Kaplan-Meier curve, right, there's responders and nonresponders. But what if you could know who would fall into what group before you recruited them into a trial, right? That's what -- part of what we're trying to solve with multimodal data sets and multimodal signatures where you're taking genetic information, information out of the image, information from the pathology scan information from the EMR, putting that together, creating a specific signature and algorithm and trying to figure out, "all right, can I predict how someone will respond before they go into trials, right?" And there's a lot of use for that different information. If you think about the cost of discovering a drug today or you think about trying to get more and more targeted with therapies, that's the kind of information you're going to need to be able to take those decisions, right? And so I would say it's hard to put numbers around that, but we know the [indiscernible] outs of a failed drug is incredibly high and the cost to a patient, right? Imagine it was a friend or loved one who was in that pool that we knew was going to be a nonresponder. That's even more devastating in terms of the cost, right? So I think there's the aspects, obviously, on the data side of being able to identify right patient, right therapy, right time. But then there's also the piece of it where for the system, that's also a huge savings as folks aren't getting unnecessary therapies or other treatments that we know ultimately won't be successful in the end therapeutically.

That's super helpful just before we move on. The $3,000 number was really helpful and kind of the $1,000. Do you have any sense of -- or any numbers around the setup of the clinical trial and choosing patients and what that number, that spend for a company might be?

It's huge, right? I mean I think if you had the CEO of most pharma companies in today and you asked them, what is the greatest challenge or one of the greatest challenge, it's opening up new sites, going into new markets, moving into new postal codes, bringing it to new parts of the world where there's naive patient populations, it's incredibly costly, right? And one of the challenges there has also been while they're great and some of us are our partners, the central lab model really was born out of the U.S. And so serving many other parts of the world in a centralized faction can be quite challenging, right? In the U.S. market, it works reasonably well. We think a hybrid approach of central and decentralized right can help on that. And for us, part of the excitement is we can look -- I can pull up my phone now and tell you what patients have walked into what institutions in what part of the world and have been tested within the last 24 hours or even less, right? And so having real-world, real-time information at your fingertips, I think on that and the ability to then recruit patients based off of that when someone comes in and not relying on the clinician can be something quite powerful. But we really need to get the scale of adoption of the technology all over the world and be able to get the cost down to where it's viable, but the savings could be tremendous because it's one of the biggest challenges to most pharmas in terms of the overall drug development process.

Great. So I am to piggyback on your term like the sexy place to save the money because I think we sort of have these two opportunities. And for us, the real magic happens when we can turn it into our business model. So the sort of unsexy side, if you will, is like we're going to drive down cost to do R&D that is our core, kind of, technology offering in the foundry is working with folks like Twist and benefiting from their scale economics as they advance using automation, et cetera, to drive down our costs. That's like the unsexy stuff, it is table stakes, we have to do it. Even as that has been happening across the industry over the past decade or two, the cost to develop new drugs as an example, has been skyrocketing despite the advances that have been made in the underlying technology, and it makes, in many ways, absolutely no sense. What Ginkgo has now been able to do in certain areas of our business and what I would love to be able to continue to do, is offer a completely different value proposition to the market. We just launched for -- our enzyme engineering business so think biocatalysis in pharma as an example. Certainly, there's a large industrial enzyme segment. Enzymes are used across many, many different applications, whether it's a component of a small molecule manufacturing process or a drug itself. We can now offer the market success-based pricing for that. You don't have to pay us a dollar unless we deliver you a successful enzyme that meets your spec. We have to underwrite that, but we have now gotten our platform to a point where we can offer a value proposition that nobody else in our industry has because we have made it predictable enough and affordable enough to do that work on our platform. Now imagine if we could do that in gene therapy or in cell therapy or in traditional biologic drug production, like that would be absolutely game-changing for the industry. And that's really how we think about developing our platform is being able to completely change the paradigm from I'm just throwing a bunch of money into an R&D problem that will probably not be solved to I am paying for a product that works. And now I just need to figure out how to get it to patients and get it to market, which is a much different question and value proposition. And to me, that's what's really exciting about how we're able to really influence the field.

Tough to build on that and those of you that haven't worked...

If you want to give a...

You can feel the price every minute of the day. The comment is absolutely right. Just I'm trying to [ swap up ] for being on stage, I think it's something like approximately $60 billion of R&D budget wasted in failed drugs in the oncology space, which is meaningful. And so when you look at that, I see Twist's role here. We're an enabler we're picks and shovels. So people will come and go, and they'll always need nucleic acid-based tools. And I think what's crucial here is actually, there's 3 things that matter: quality. If you're making a nucleic acid-based anything, there's nowhere hide you either made the right molecule or you didn't. And that is hard to get right. That's taken us 10 years to hone that game. You have the quality component. We have a scale component. Just as a reminder tens of millions of oligonucleotides every day. And the third part that we need there is speed to truly enable. And so that's something we're working very hard on building.

Speed's huge.

Absolutely. And so what we're seeing now, if you think about the true precision medicine solution, there's a couple of things you have to do. You have to read the disease, you have to understand what that means, and you're probably going to bolt on some other modalities to understand what's going on with the disease. Precision-based therapeutic is going to come from one of us here and a few others, obviously, but it's going to be underpinned by genetic content. So that speed to right effectively and then get to the point where whatever biomolecules going into the patient can be made as quickly as possible, truly matters. And so we've had it described to as sort of needle-to-needle in days rather than years for the true precision approaches to solving complex disease, we really do think that scale and speed and quality is going to matter.

Yes. So we're really focused on solving two problems which are big cost savers for our partners. The first is time to clinic. It takes on average 5.5 years to get a drug into the clinic. And we are going to be putting in the first generative AI designed antibody in the clinic within 24 months, so going from 5.5 years down to 24 months. That's a huge time savings. And that time savings, you get it to patients sooner. But also if you look at the patent cliff, you actually get additional 3.5 years of patent life and that's huge for biopharma, our biopharma partners. And then I would say the second big savings is being able to increase that overall success rate. It's about 4% success rate from start to finish. If we can actually start designing drugs that have the attributes that we want. That's what's going to start increasing success rate. Additionally, at Absci, we have another novel target discovery platform that's based on reverse immunology, that's actually delivered new novel targets to go after. So not only do you have to design better molecules to increase that overall success rate, but we also have to be looking at new novel targets to be going after as well. And so I would say, with our platform, what gets pharma most excited is being able to use this to discover brand-new novel biology. And that ultimately is the value prop that brings in the large upfront payments, the milestones, the royalties. And at the end of the day, pharma CEOs are looking to how do I continue to develop my pipeline to get first-in-class, best-in-class assets and that's really what we're delivering on at the end of the day. So it's increasing the success rate and decreasing the amount of time it takes to get these therapies to patients.

And maybe, Sean, starting with you and going back the other way. When you think of that next stage of genomic analysis or molecule synthesis, in your view, where do you think companies are going to differentiate to gain a stronger foothold within the market?

Yes, absolutely. I think it goes back to being able to do things that traditional technologies cannot do. Again, looking at phage display or immunization like that can only get you so far and being able to use technologies like we've been talking about, like generative AI, to be able to design, again, molecules that create new biologies that current technologies cannot deliver on. And that's really where we've been focused. And again, it's all that -- the focus on new biology, being able to create, again, first-in-class, best-in-class assets for our pharma partners. And that's where I feel like the traction is going. Ultimately, everyone has a better mousetrap out there, whether it's better phage display, better immunization. But if again, you can actually design something that no large pharma or other company can design and create that's a huge value prop for pharma and I feel like that's ultimately what gets these new emerging technologies into pharma, especially in this type of environment.

A few comments, just coming back to speed and scale for Twist. There's lots of great ideas. We're incorporating AI, ML ourselves into some of our designs for smarter design. At the end of the day, we still have to produce a high-quality molecule for testing as quickly as reasonably possible. Now that's absolutely fundamental to the industry. And so that's where we'll focus our efforts over the coming few quarters and really support folks chasing the big discoveries like Sean and Ginkgo and obviously on the reading side we will draw and we'll push it forward.

Yes. You'll never hear Ginkgo bet against speed and scale, but I would add a third -- I would add a third component to it, which is really around IP and this ability to try to democratize access to tools, I think, again, in the last 40 or so years, as IP has developed, let's say, traditionally in an academic lab, company gets formed around that new technology and looks for a nail for that hammer that's been developed. It turns out you need a lot of different tools to build a successful drug, not just a hammer. And like just to put an example out there, you could have one company that's done some amazing work developing a therapy for a particular, let's just say, cardiac disease and another company that's developed an awesome vector that targets the heart, but has no idea how to develop therapies, you will never see those companies cross-license to each other. Instead, you'll see them try and largely fail to fill in the gaps around their core technology, and you end up with a bunch of suboptimal drugs trying to make it to clinic and through clinic. Ginkgo has approached the market with the mind that if we're going to advance this field, we need to bring together the tools that customers need, especially as therapies get more and more complex and more advanced. No one vertically integrated therapeutic developer is going to have the breadth of tools they need, especially as the market continues to evolve. And so one of the things Ginkgo is really focused on, in addition to being able to just experiment faster and cheaper to develop new technologies is bringing together that much wider toolkit so that when you come to us, no, it's not that we're trying to sell you the best little mousetrap for the one little thing and give that to you exclusively, it's that we can really integrate a wider solution so that you're really filling all the gaps that you might otherwise have. And we can incorporate whatever you think your special sauce is but we can help round out the program to optimize the chance that you make it through successfully. So again, it's really about bringing together tools and not just trying to optimize around one particular technique.

I think I'll build on sort of the comment around scale because I do think scale, particularly when you're talking about the problems we're trying to solve is incredibly important. I think for us, as we step back, there's a lot of amazing technology in the market, and there's a lot of impactful potential modalities. But in most institutions, right, whether it's pharma or it's academics, or central laboratories, that data and those modalities even within the same institution is siloed and sort of not unlocked, right, at its greatest power. And so for us, I think in terms of the ability to differentiate and where the future is going, you need to be able to touch these institutions across many different data modalities, but in a way that you get true network effects, right? This hasn't really existed in health care before, right? People kind of kept their data siloed. They used it for themselves in their own sort of R&D purposes. But the reality is if we really want to do much of what we're talking about on the stage, it needs to be unlocked and unlocked at massive scale because the only way AI and machine learning and all these different techniques work is a massive data scale, right? And so what we're looking to enable is sort of that collective intelligence network around the world where data streams -- no one owns it. We don't believe we need to own the data, but we can control it and enable it to be safely utilized and transferred. And by doing that, it will enable more production of it because people will know it's safe and then we'll have good use. And so our view is -- the more data that is produced and the more usability of that data at scale around the world, right, in a diverse global population, the more that each of us will be able to ultimately do more of what we want to do, which is find new discoveries and extend life. And help people live, right, which is a lot of what the end goals of our customers are because at the end of the day, there is a patient, right, in all of what we do in at least on the drug side, there's an individual that is afflicted with something, and we need to find them a solution to be able to live.

Great. I think we're going to transition to the company-specific lightning round. So, Ross, since you're next to me, we'll start with you. And I think John will have a couple as well for some other -- but all right, why don't -- I mean you've talked a lot, but, I guess, kind of let's talk about the portfolio that SOPHiA brings, you have your roots kind of in advanced cloud-based genomics, data analytics. You talked about kind of the balance between the decentralized and centralized models. But can you bring us to kind of where you think about the portfolio today and kind of the pieces that it's tackling in terms of the data analytics side?

Sure. So we've started being able to support any precision medicine application in oncology or rare disease. That was kind of our core backbone. You buy whatever sequencer you want. Hopefully, you buy Paddy's chemistry and then you use and create your own tests, and we can support that at scale. That was sort of where we were, now as we think about sort of some of our newer solutions as we think about CarePath, right? So that was sort of the core DDM for genomics. As we think about CarePath, which is more of our multimodal tool, now we're taking that great genomic data, we're generating and enabling it to be comboed with other modalities to again lead to, let's say, more conclusive outcomes. So an example would be what we're doing with our deep lung study, where in the case of several thousand non-small cell lung cancer patients, we developed a very complex signature using our proprietary algorithms that can be, let's say, 80% predictive in response to a PARP inhibitor. So before scan 1 that happens, we can tell you what the outcome is going to be for that patient, where they fall in the Kaplan-Meier curve. Imagine that across many different indications and diseases. That's kind of where we're driving and where our portfolio is going. So we've talked about that is the balance of, let's say, our clinical customers, which are mainly labs and AMCs and others. And then as you think about CarePath, primarily the customer there is pharma, who is using a lot of what we're talking about there. So for us, that's where the portfolio over time is going, to where those two pieces balance a bit, but they also feed each other, right, because to the degree that pharma wants to deploy a decentralized solution or tool in the market, it needs to be deployed in the clinical market, right? And so they want to sort of improve upon that and have that in a decentralized world, so they can have data at scale, and then that helps those customers, which feeds back to pharma. And so that's the flywheel we're trying to enable.

Great. All right. We'll move it around it. I think we'll go to Paddy now, the Factory of the Future is up.

Goodness, [ factory up ], right now. Yes, we're very proud of our achievements to bring the factory to scale. So from a commercial standpoint, obviously, we've got the technology working over the last few years of products, certainly around the sequencing side and synthetic biology side of the business, we've done well in showing that we can take a segment of the market, we can actually reach into high-volume users and compete very, very effectively. Second part to ensuring that the entire globe has access to our platform is scaled, sorry for the theme. And so adding the factory of the future has really allowed us to one up the number of different nucleic acid sequences we can synthesize. But in addition, it's going to enable us to drive much quicker to a product to the customer. So the factory is up and running. We have a big investment in making a superfast gene happen, which is going to open up the pharma drug discovery space. It's going to keep one of my very, very important customers to my right here, very, very happy and will also allow us to drive into the academic segment and you just open up interesting new areas of research as well. So that's good. I'm really very privileged to serve with the team to make that happen. And then the other thing that's going to enable for us is what do we do downstream of DNA synthesis and there's a whole value map of what a customer does want to have a piece of DNA. So again, our job is to enable the community, is to enable companies like Ginkgo, SOPHiA and -- we'll talk later, to do a very, very good job, make sure we can stand behind them. So that downstream product mix will make us a very valuable partner as we go forward. It just needs square footage once you've come off the microfluidic chips, steady progress.

One for Anna Marie. When you think about Ginkgo's approach in terms of unlocking productivity gains, where I guess is the sweet spot when you talk about cell engineering and the flywheel you're creating there in any way just to come talk about or quantify the number of programs or targets that you're working on?

Oh, man, I think our sweet spot is insatiable. So again, our goal would be that -- let me step back for a second, we have a little over 100 active programs on the platform right now. Some of those programs are multi-tens of millions of dollar bespoke programs, where we are working hand-in-hand with the customer to figure out a brand-new scientific challenge. Others of those programs could be 6 months enzyme engineering campaigns that are largely solved by AI, and we run a few experiments in the lab to validate that we've met the customer's spec and we hand it back relatively quickly. My goal would be that over time, the projects today that feel like big $40 million bespoke messy projects become what the enzyme engineering projects are today because as we continue to work in these new areas, we get better, more scalable, faster, more predictive with our algorithmic tools and we can take stuff out of the lab and bring it into a more predictive space. And that should allow us to work on not just 100, not just a few hundred, not just a few thousand, but tens -- I mean all the programs, but I would like to make it as easy to make a new drug as it is to write a neat little piece of code for a website today. That is really our vision. And to make it such that if you've got an idea, you can create it and you're not limited by the science, you're really only limited by whether it's going to actually add any value to the world. So I don't think there is a sweet spot. I think if the question is more, at what point do we start to reach kind of operational profitability or something like that, that's a little bit more investor driven. I think we're solidly on that path. We've now made the investments in the platform that allow us to really now take advantage of efficiency gains as we add more programs. Just on the upfront economics that we received from our customers. But again, I think our vision is that as soon as we get products more to that predictive state where we can really rely on the downstream economics we get from the customers, we would like to give away the platform as cheaply as possible to get as many programs out in the world as we can because there's so much more value for us to capture as well when those products are successful and they make it to market. And so our ambition is insatiable. And while we're solidly marching towards kind of profitable unit economics today and a profitable platform today, I'm much more excited about the opportunities when we cross that threshold, for us to go beyond that.

And Sean, how should we think about the benefit of leveraging AI to improve the clinical discovery and any financial aspects or commitments you can share with us there?

Yes. No, absolutely. I mean, one of the marquee partners that we have is Merck last year, we closed a $610 million deal with Merck to discover three new AI discovered antibodies to three targets. And really the reason why they had reached out to us was again, that ability to specifically hit targets that they previously couldn't have hit themselves as well as actually using, we have a nonstandard amino acid platform that allows us to do quick chemistry of different payloads for ADC development as well. And so being able to make next-gen ADCs. That's really been a huge kind of value driver for us in addition to the AI drug discovery, being able to kind of create those new modalities. And again, the AI is really being used to specifically target the attributes that you want. Because a lot of times what you see is drugs that have actually very promising functionality, actually never make it into the clinic because you can't manufacture it. It doesn't have the best, let's say, half-life or the best PK/PD properties. And again, if you can use AI to hone in on all the attributes that you want the functionality, the manufacturability, the developability. That's what's going to increase that overall success in the clinic.

Great. All right. Back again. Another round. Ross start it off. So you talked about traction in some of the initiatives like HRD, liquid biopsy. Can you help frame kind of the opportunities that you're seeing there? And then kind of a more thought -- how do you think about like comprehensive genomic profiling and the uptake curve, right? I think right now, people have said it's about 30% of advanced cancer patients in the U.S. are actually kind of even getting CGP and kind of how do you think about the evolution of that?

Sounds pretty high, by the way. I think it's probably lower, but maybe that's a U.S.-specific comment versus the rest of the world. I mean, I think most parts of the world, and again, we address the globe, the access to many of these I would say advanced products are much less, right? I think from our perspective, when we take a step back, the market grew up over the last number of years on small panels, right? And it's still primarily what's being used across the board. But I think -- if you think about it from pharma's perspective, right, and the type of information they want to have for their drug discovery purposes or drug development purposes, bigger is better, right, in terms of the amount of data you can get on the patient. And so I think a lot of this is now around how you can push into some of these newer areas with larger panels and more content and be able to drive that information in a way that can then inform on what you're doing next. And so HRD was a great example in terms of that relative to the PARP market, right, which has been exploding on the pharmaceutical side, and you can look up some of our partners, of where we've helped them there. But I think they're still sort of just scratching the surface given the number of indications those PARPs are likely to expand into of where we are. So we're very excited about HRD on a global basis, including in the U.S. and think there's still quite a lot of legs to go there. But I think if you think about that story versus the others, it's not too different, right? I think in general, if every pharmaceutical company can have a CGP or a whole transcriptome or a liquid biopsy available for every patient in some areas, they would want it, right? Particularly liquid is really useful as you go to many parts of the world, just given the challenges on tissue and storage and sampling and other elements. Blood draw's a more I would say, easier way to do it. And so in our mind, it's where the market is going. If you think about for us, from a business model perspective, it's why our ASP keeps moving higher because it's higher and higher value, some tests that are being computed where the amount of information we're processing on our platform keeps going up. But we're still not even scratching the surface, particularly on a global basis of any of those tools in terms of their penetration. And I think what you're going to start hearing more and more is more post codes or more ZIP codes in more parts of the world for all patients, right? And equal access and how you get into different geographies. And that even is in the U.S., right? If you look -- so I was wondering on the adoption because I think that's maybe the addressable market within major metros in the U.S. But if you go out to the community or you go to the more rural parts of the U.S., you're definitely not seeing CGP testing. So I think as we just think about that availability, et cetera, that's what we're trying to allow for that all populations get access and I still think we're probably 5, 10 years away from that reality, but the market is moving there. And I think for everyone, including us, that's a positive.

Hopefully. So CGP for everybody 10 years from now.

Yes or even -- and again, we didn't even talk about whole genome, but like that's another conversation, but it isn't appropriate for all things -- or exomes, I think Paddy would tell you, are still quite...

Very relevant.

And popular, our growth there was also tremendous and this is all, I think, a positive development for the market because it's more information, more data on all patients.

Great. All right. Paddy. I guess, why don't you talk about -- you alluded to the DNA data storage and kind of how that's going to be a different or a distinguishing feature for the company. Can you kind of dive into that a little bit and help people kind of understand kind of the work and how you're scaling these solutions.

Sure thing. So again, pre-revenue, I think we're talking about potential product in calendar year '25. It's kind of core to who Twist is as a business. We're an interesting combination collection of skills and microfluidic, DNA chemistry, silicon, fascinating and eclectic and gathering of scientists and engineers. The key, quite frankly, is density. It's very expensive to go into DNA as a storage medium today. So our investment and our focus is on really driving the density at which we can write high-quality nucleic [ assets ]. So we've got some proof of concept work coming out in the next 12 months or so that's going to show the end-to-end prototype and workflow with a view to having a much higher -- or excuse me, a quite higher density coming through calendar year '25. But it's all underpinned by what got us to the start line, going from the 96 well plate which is constrained biology for the last however many years you want to look back, breaking out of that on to our silicon chip platform, which is now with tens of millions of all the oligonucleotides per day, now coming up to this where it's going to be many millions of oligonucleotides per run. Which, again, that's what's going to open up that space. And there's a meaningful part of the market segment that really is about cold data or write once, we've never resell them, which we do think will -- DNA as a medium is going to make a meaningful contribution to enabling capacity for other more rapid read-type applications. So we've been here before. We have the right collection of people to make it happen. It's real. We have done proof of concept work in existing platform with many customers. So I think it's just a matter of just proving out the technology and bringing it to scale.

And Anna Marie, I was wondering if you could talk a little bit about some of your recently launched Enzyme Services. Remind us of the offerings there? And just how is that adoption curve trending?

Sure. So the background for this is that we discovered that throughout our history when we had been going to customers and saying, "what is it that you want us to do for you?" They would ask, "what is it that you do?" And we would say, "we can do anything. What do you need?" And you go in circles for a little while before anyone actually figures out what we're going to do together. And so we realized it would be helpful if we started going out to the market and saying, "okay, well, in addition to being happy to discuss with you whatever your needs are, these are our few products that we actually offer." And some of the earliest ones we sort of formally launched were in the protein space because if you think about it, effectively, any program that Ginkgo works on, whether the product is a small molecule or a biologic drug involves some element of enzyme engineering inside of a cell. And so it was an area where we had a lot of experience. And so that was the first sort of service that we productized and we started going out to the market to talk to folks about this offering and this tool as part of our broader portfolio. We've since launched a number of other formal kind of product lines. Again, there's no real difference to Ginkgo. It's just a way to start the conversation with the customer about what they might need to do. But the real innovation, as I mentioned a little bit earlier, has been that in some of these more mature areas now, including enzyme engineering, we're able to offer a very different value proposition around success-based pricing as it becomes predictable and very high probability of success for us to be able to do, and that's something I'm really excited about that's recently launched.

And Sean, how do you think about pharma companies will eventually think of R&D? If this generative AI and the Absci model becomes a part of the standing -- operating within pharma, how do you see that?

Yes, absolutely. I think one of the things that you're going to start to see is pharma companies start partnering at a later stage. So instead of partnering at the target they're going to start partnering at IND because what we're going to start seeing as the costs start to shrink in terms of getting to in vivo validation, let's say, instead of costing $15 million, it can cost you $3 million and you can get there in 18 months. And that's what we're seeing is actually a big value inflection point when you partner not at the target stage, but when you have in vivo validation or IND, you can start looking at much higher upfront payments. And ultimately, again, putting on my large pharma CEO hat, they're looking for assets. They don't care how you got the assets. They just want assets that are going to be first-in-class, that are going to be highly differentiated and have the ability to be best-in-class. And if you can -- with an in vivo model or an animal model that is -- that pharma believes in and you can show some really great data there. You can start cranking out IND packages in a manner that you haven't been able to before and capture much more value, but you're also not taking the clinical risk as well. And I see -- that's how can I see business models shifting, I think even Absci is starting to see that as well as we've started to build out our own pipeline and starting to take things later and later because we're able to do so in a much shorter amount of time and in a more cost-effective way, delivering differentiated assets.

Great. Well, we're already a minute over. So thank you so much, everybody, for joining. This was a great panel and a great discussion.