Meeting The Moment with Tools of Tomorrow with Twist Bioscience

Speaker 1:
Welcome to TD Cowen Insights, a space that brings leading thinkers together to share insights and ideas shaping the world around us. Join us as we converse with the top minds who are influencing our global sectors.
Brendan Smith:
All right. Welcome back to another episode of Machine Medicine: AI and Healthcare, TD Cowen's podcast series where we bring you the latest and most important takeaways from the state of AI in the healthcare sector today. I'm your host and TD Cowen Healthcare Analyst, Brendan Smith. And today I'm joined by Twist Bioscience's President and Chief Operating Officer, Patrick Finn. Patty, it's great to have you and welcome.
Patrick Finn:
It's great to be here. Always good to spend time with you, Brendan.
Brendan Smith:
Yeah, so for anyone new to our podcast series, Machine Medicine aims to break down the use of artificial intelligence across healthcare into bite-sized digestible points one episode at a time, highlight the biggest misconceptions and then re-contextualize each piece back into the bigger picture. Today, Patty and I are exploring how companies like Twist are capitalizing on the evolving drug development landscape today and what impact biopharma's use of AI in their own R&D workflows is having on life science tools providers across the industry. So I guess with that hutty, let's just dive right in.
If we rewind the clock back even a year or so ago, Twist was likely not the first name that would come to mind when folks think about AI in healthcare. I mean, you all offer so many different products across all these different end markets, but I think it's just worth contextualizing for today's discussion. So I guess for those listening in who are maybe newer to Twist's story, help us understand where Twist sits in this ecosystem here in July 2026 and how you all have found yourself at this pretty unique intersection between traditional drug development tools and the wet lab of the future.
Patrick Finn:
Great question. Thanks for that. And now as an 11-year Twist veteran, it's true, we've had some remarkable growth in the platform. I think if you go back in time, Twist started by solving a fundamental problem in biology. DNA synthesis was slow, it's expensive, and it's low quality, or historically has been low quality. Now we've built out a semiconductor-based platform that writes DNA on silicon chips at massive scale. I mean, if you look towards the upper end of what we can deliver today, it's about 32 million oligonucleotides or DNA building blocks per day, which is just incredible, a super grade of molecular quality. And that's changed the economics of the field. It's a structural cost advantage with that throughput, meaning we can basically load more products onto each chip, which drives a business that ultimately is going to become quite profitable.
From the foundation of the chip, we've launched our first commercial offerings back in 2015, 2016 around clonal genes and gene fragments. And over time, we've just been a new product introduction machine, building a full platform of products that moves us along the value chain and all the way through to a service offering for antibody discovery. That's one of the later additions to the portfolio in 2020-ish, which has actually become a foundational product for our role in AI-enabled discovery today. But the scale of the platform's a little bit more than that, and actually it's meaningfully more than that. 2018, we added NGS tools to compliment our first, what was synthetic biology offerings, gene product line, giving us basically a new product portfolio that's really grown substantially over the last eight years. I think that AI intersection's a natural consequence of what we've built.
When AI companies start generating thousands of antibody sequences, someone has to build them. Someone has to make the molecule, express a protein, characterize the protein or the antibody, and deliver data back to the customer. And so when you think about that with the color of our throughput, 32 million oligos per day, potential capacity of three million clonal genes per year, Twist is the best manufacturer. And quite frankly, we have the scale and throughput to actually deliver on those AI driven requirements. And I think what's making this year different is that we don't stop at DNA, that scale in writing oligonucleotides and genes, we've added the same capacity and strength and protein expression, characterization and data delivery, which makes a full sequence to data partner. Which is exactly what AI-driven drug discovery workflow requires.
So Brendan, you're right in the opening comment. It's not the most obvious thought, but when you look at the infrastructure and our molecular prowess, it's actually a very, very nice match in the platform. So if AI designs the sequences, Twist builds them and gives data back. I think of it as a wet lab executional layer for this revolution that's happening right now. I think it's not by any stretch of imagination niche a player. It's top 20 pharma, it's AI native, dry lab biotechs, and the magnificent seven tech companies are all coming into our customer base. So we're just getting going, but you can see it's a good application of the platform.
Brendan Smith:
Yeah, I think that's great. Thanks, Patty. I think that helps level set where all of this is and frankly, how it stands in contrast to what we've seen in a few other sectors. I think most people are probably familiar with the large frontier models from Anthropics Cloud, OpenAI's ChatGPT, both of which we're realistically trained on just billions of parameters across the corpus of the internet, what have you. But I guess given how different the models being developed in healthcare are versus some of those chatbots and search engine-like offerings, and I know you just touched on this a bit, but I guess how should we think about some of those differences and really the importance of wet lab tools, like the ones that you all provide really as we see the drug development landscape increasingly become more and more digitized?
Patrick Finn:
Another great question. So I think a lot of the analogies break down quite quickly. I think if you look at something like ChatGPT, it's trained on text that already existed on the internet. That's a wee bit tougher to do with drug discovery because a lot of the data you're looking for, positive and negative, it hasn't existed up to this point. And it's such a simple statement, but it does need restating. You have to go generate the molecules and ultimately the characterization in the wet lab. And there's just no getting away from that basic chemistry and biology. And basically if you think about what the drug discovery AI model needs to know is does the antibody sequence, does the antibody bind to the target? With what affinity? What's the expression level? What's the stability?
What's the likelihood that this discovery, this molecule is going to be readily scaled and is going to be viable for driving up to scales that's going to serve a broad patient population? And there's an awful lot of missing data. In fact, the predominance of that data isn't sitting in a database somewhere. It has to be experimentally generated. And that's exactly what Twist does. The more powerful the model, more high quality experimental data it needs to train them. And also that another really important point is data quality, which matters enormously in a way it may not matter for some of the things like chatbot, for example. A model trained on bad binding data is going to generate bad candidates.
That's something we've obsessed about from a product standpoint too, just to go super deep. Every data set we're creating or has controls run within the data set, so that when our customers come onto the platform, the quality, integrity and consistency of that data remains tight. So when we think about the idea of the digitization of drug development, it doesn't shrink the wet lab. It's actually going to industrialize it. The question's shifting from should we run the experiment, to how many can we run? How fast and with what data fidelity? And that's the question that Twist was built to answer. And it's underpinned by our obsession around molecular quality. You can't be in the DNA synthesis business and ultimately the biological synthesis business if you don't make an incredibly high quality molecule.
Brendan Smith:
Yeah, that's great. I mean, I think you're getting at a lot of important underlying questions here that people are trying to wrap their head around as they think, not just about how we got to this point and how the goalposts are moving in some respects and how all of this is evolving frankly on a daily basis it feels, but also trying to think forward about this. I guess I have a suspicion on how you answer this, but I'm curious what kind of color you can provide around it. But do you expect realistically that wet lab validation will remain an essential part of drug development, at least for the foreseeable future? And I guess maybe to your last point there, I mean, how do you think about maybe the evolution of the durability in this data demand over the coming years? Never know what we don't know and it's hard to predict some of this, but just based on what you guys have seen so far, how should we think about that space evolving moving forward?
Patrick Finn:
Good one. And of course, yeah, you are right. I mean, it's an unequivocal yes to the need for ongoing wet lab validation. And quite frankly, I look forward to debating anyone with a different opinion. The biology still has to work in the real world. And that means make the molecule and test it for the long foreseeable future. What we're actually seeing is true durability. We're in the early stages of commercialization. We're starting to understand customer buying behaviors, how they're thinking about the projects. And for anyone who missed our investor day in May, the customer talks we have there on our fireside chats with our customers say it again and again, the market and the opportunity is only increasing, not going backwards. And I think you'll get a virtuous cycle as we start to see successes as molecules progress towards and through the clinic.
We also see customers who've deployed the most sophisticated models are generating the most wet lab demand. The better the model, the more experiments it wants to run. And the learning loop is self-reinforcing better data trains, better models, better models create better hypotheses, better hypotheses require more experimental validation. I think as every turn of the crank creates demand for the next turn of the crank. And so from a technology standpoint and from a business standpoint, we're quite bullish on that. Then when you think about it, and I talked about this a fair older mind, this is absolutely in its infancy. We do see the models moving into harder, more complex problems and molecules. I think we can start with something like even just the skew towards smaller biological structures today.
And once you start to see full VHH structures become more predominant, I think that's very obviously how companies will be thinking about the product roadmaps. And bispecifics, it's a relatively unchartered territory. We know it's a rapidly growing segment of the drug discovery continuum. Higher order multi-specifics, novel scaffolds, nucleic acid drive therapeutics, again, can leverage a Twist platform and is relatively unchartered. And so I think as AI leans into, or sorry, customers utilizing AI tools are leaning into more complex modalities, the wet lab complexity and the need for Twist operating scale is going to go up, not down. So it was an obvious answer from a yes or no standpoint to your question, Brendan, but that's our thinking behind what we see as we think about our outlook and the outlook for the community.
Brendan Smith:
Yeah, that's great. And I think again, it also gets at what is really just top of mind for a lot of people is just understanding how it's changed so much even in the last year and how we can possibly try to wrap our head around where this goes over the coming 2, 3, 5, 10. And I guess to your point there, it sounds like the breadth of offerings and optionality is increasingly a key ingredient to success here. I know I argue one of Twist's biggest strength in recent years has been what Emily loves to say is this, "We'll meet the customer wherever they are."
Whether they're looking for DNA, oligos or recombinant proteins, IgG, all the analytical data stuff that you're talking about, it really does feel like having a degree of portfolio optionality is part of the recipe here. But I guess looking at how this has evolved now, what has maybe surprised you the most about how the impact or just what the impact AI is having on your and your customer's needs and really how we should think about ultimately what it takes to win in this space moving forward?
Patrick Finn:
Another good question. I talked in the opening comments about us being a new product introduction machine. And so we're not surprised by customer demand for new product. We've built a platform and a team that can respond quickly to market needs. And the flex we have on the platform has allowed us to really address the product challenges. We have good form over the last 10 years of driving the business through new product introductions and organic growth. I think the real surprise is how quickly the center of gravity in this segment shifted towards data. Meeting the customer wherever they are, it's not just a philosophy. It's a competitive and commercial strategy we employ from the moment we started. We're in the sport of business, we're in the business of delighting customers. And the moment we're going to tell a customer we only do arrayed workflows or we don't do bispecifics yet, we've handed the account over to a competitor. And yeah, that's just not what we do.
So I think what it's going to take to win is that full stack, synthesis, expression of the protein and antibody, the characterization, data delivery. You need to be fast. You need to be the fastest at all of it because as usual, the customer base is incredibly sensitive to turnaround time. And quite frankly, being great at two out of four, it's not enough when the customer's AI model is waiting on the slowest step. So speed and scale are non-negotiable. I think there's some early customer data out there showing Twist at 17 days, versus 41 to 57 days for the nearest competitor. That's not a marginal advantage. If you're making a decision to go to 41 to 57 days for your experiment versus 17 days with Twist, you're going to lose in that market segment. And so if you look at it, our platform's built for this revolution. And I'll borrow another Emily-ism, which I think is absolutely super. GPS needed Uber as one of the outstanding apps. And AI-enabled Discovery is another outstanding app that fits perfectly onto our platform.
Brendan Smith:
That's great. I'm not sure that I've heard you all make that kind of correlation before. It's great. Very soon-
Patrick Finn:
That's a good one. I always try to beat it, but I never can.
Brendan Smith:
No, that's great. And I think that again lends itself to which something I would be remiss if I didn't ask about, but this is the recently announced partnership with Amazon, AWS, really helping position you all as pretty key wet lab provider in that Amazon Bio Discovery platform that they've launched. I still think there's been a lot of interest in understanding even more broadly just how big tech fits into this picture of healthcare AI moving forward. I know they've all been throwing ideas out there for years. It's not brand new that a lot of those guys would want to get a foothold somewhere that they can. So I guess maybe through that lens, how do you think about this growing integration of big tech platforms within the life sciences? And maybe what are some of the ways that you feel are easiest for you all to compliment each other's offerings?
Patrick Finn:
Yeah, another good one. When I look at the scale of the Twist platform, the global community deserves access to the platform. Speed, cost, quality, it's unmatched. And what we need to deliver on that comment is quite frankly, is reach and scale. And yeah, my understanding of Amazon, I think all of our understanding of Amazon is they are absolutely exceptional at building platforms that are somewhat effective at aggregating users, maybe the understatement of this podcast so far, and delivering on a really seamless user experience. And so the partnership's going to lower essentially the activation energy for a researcher to go from AI design sequence to wet lab order. If I go back to classic commercialization strategies and crossing the chasm and taking technology and capability out to the majority of users, and that's incredibly important. Now when you partner Amazon's strengths with our ability at building DNA, expressing proteins and delivering data fast, no matter, cloud infrastructure solves that.
And so the partnership becomes truly complimentary. They bring customer funnel, compute layer, and Twist essentially brings a foundational molecular layer, whether it's wet lab execution or that data. And so what it unlocks for us is access to customers who might not fund us directly. Someone who needs a seamless digital workflow before they commit to outsourcing wet lab work. And it just makes it easier to get started on the platform. And quite frankly, the Amazon validation matters too. We're one of the most analytically vigorous technology companies in the world. So let's use a wet lab execution partner. That's a signal the market notices. Customers who've come to us directly are increasingly referencing it. So for Twist, we view big tech as a channel partner. Certainly we don't see them as a competitor. Our molecular strength is so beautifully complimented by what they do.
Brendan Smith:
Yeah, that's great. I mean, I think in terms even of competitive landscape, I mean it seems to us that in many respects, AI is pouring fuel on the fire of what was already true in some ways. Pharma companies prioritize speed. And you mentioned this a couple of times now. Equality is of course important. That's entry stakes for a lot of this, but it feels like time is increasingly an equally valuable resource in this. Is that the right way to think about it? And maybe why might that be now more than in the past?
Patrick Finn:
Yeah. I mean, as usual, to me, you're spot on. It's exactly the right way to look at it. Speed and quality to us are no longer trade-offs. If I go back to an earlier comment, we've obsessed about molecular quality. The company's founded and led by a DNA chemist. Our CTO is a DNA chemist. I'm a severely lapsed DNA chemist. And we knew and we learned in the early days of commercializing Twist that molecular quality is absolutely fundamental. And we've obsessed about delivering the best quality molecule, a throughput nobody else can match. And so now basically speed and quality, they're no longer trade-offs. They're table stakes to be in this space. You can't win on speed if you're delivering a poor molecule or if your data's not trustworthy, and you can't win on quality alone if your competitor delivers in half the time.
And again, it's another one of the beautiful applications of the Twist platform. This AI scale has shown the cost of slowness is now more visible in a way that it wasn't before. If I go back in time, until we launch the express gene product, you can get a gene or one gene in five to seven business days. If you ordered a thousand genes, you can get it in five to seven business days. If you order 10,000 genes from us, you get your product in five to seven business days. Competitors can't manage that scale. And what you'll see is multiphasic deliveries. So by the time you go end to end across these higher volume orders, your turnaround time is, well, we showed some data or mentioned some data earlier. It's in the months, not in days.
And I think the scale of what these AI models are driving in terms of what's required from data generation, the molecules required to generate that data, that's what's really uncovered that all of these components which are normally opposing each other are important to this customer base and this segment. And so when you think about that, when you're doing traditional discovery, I think that a six-week turnaround time, and we know this from talking to customers, is frustrating, but it's manageable and acceptable. And at the time customers were delighted with that. But when you look at an AI model that's ready to crank on the next iteration of four to eight hours, a six-week turnaround time, it just doesn't fit. And so the benchmarking I mentioned earlier tells a story. Top five pharma, five data providers head-to-head delivered or we delivered in 17 days. Our fastest competitor was 41 days, our slowest is 57 days.
And by the way, these are companies that are advertising sub 20-day turnaround. So I think the iteration loop, design, build, test, learn, repeat is now the unit of competition. It's not just one experiment. It's not just one gene or one sequence. It's about how many learning cycles at scale you can complete in a quarter. So every day a turnaround time, it matters and it compounds across each of the programs. So that's why today we're in a good position. And we have track record for scaling up not just volume, but improving turnaround times ourselves. And so as you look forward, and that compounds across every program, every modality, and the multi-year of runway that we see here, we start from a good competitive position and we are going to be relentless in improving the quality of the product and the data that we deliver to our customers.
Brendan Smith:
Yeah. I so appreciate that context too, because I feel like it's one thing to look at some of the old battle lines through volume lens of the past. But to really extrapolate to how a lot of this is likely to evolve moving forward, you have to think if you turn up the volume really rapidly, instantaneously in some respects, who actually has the capacity and ability to meet that without compromising some of these other KPIs? I think that's increasingly a super critical part of the conversation too. So I guess from really where you sit and I guess what you've seen over the last couple of years, what do you feel the investment community likely underappreciates or maybe even just misunderstands altogether about AI within your ecosystem that you think is especially important for anyone listening in to know here in the summer of 2026?
Patrick Finn:
Oh, another good one. Probably the most underappreciated thing, this is already a real business for us. It's not future hand waving. It's not sitting on a bunch of PowerPoint slides being debated in some long-winded business strategy session. It's happening right now. I think we're about 111 million in therapeutic drug discovery revenue last year, greater than 25% growth, top 20 pharma driving into a customer base, dozens of AI native partners running repeat production workflows. So I think there's still a lot of external view of its AI hype that what we're in the business of is AI demand and delivering on that demand. I think the durability question is one that's come to us many, many times. We see durability. I've talked a little bit about how we're seeing or learning how our customers buy repeat orders from multiple customers. Customers who are trained on models and known targets are now back with novel targets, doing real drug discovery.
Every learning cycle creates demand for the next one. I talked a little bit about the different modalities that will feature in future design build test learn cycles, bispecifics, higher complexity, nucleic acid therapeutics. And I think probably the last point, we've spent a fair bit of time over the last few quarters talking about the AI story. But as a business, it's very clear that our NGS product offering is growing double digits independently. So coming back to and talking about the resilience of our business, we're a diversified company with thousands of customers serving multiple really high growth opportunity markets where our technology point of differentiation, driving products that our customers love and scale well into industrial applications sets us up for a pretty rosy future.
Brendan Smith:
That's fantastic. So I mean, I know we've covered some great ground today, Patty, and it's a conversation I'm sure you and I will continue to have for the foreseeable future, let alone within a few weeks, I'm sure. But before I let you go, one thing I like to ask all of our guests is if everything we've discussed today goes over or somebody said, whether by an inch or a mile, but they've made it with us this far and they're still listening, I guess, what is maybe the one point you would really want everyone listening in to remember and just take away from our conversation today?
Patrick Finn:
Good one. I think everything that Twist makes leverages the same silicon chip platform. And I think from a business standpoint, that's insight that unlocks everything else. The chip is largely fixed cost. And so as we load more product and more volume onto the chip, the incremental economics are very, very favorable. They're extraordinary. And so every new product we introduce, it's not a new factory, it's not a new cost structure, it's more volume on the same chip. And so therefore the MPI machine, essentially, it's a margin machine. So as we add new products, and again, remember the technical strength we have at the top of the company, it's a deliberate, very thoughtful, customer-centric effort and initiative to compound the leverage of that fixed cost platform. So if that was the simple take-home message, I think that would be what I'd like to try and articulate as we wrap up here.
Brendan Smith:
Yeah, that's great. And I think with that, I want to thank you, Patty, for hopping on and talking us through really this marriage of software and healthcare tech innovation. I'm sure we'll have plenty more to discuss over the weeks and months ahead as it pertains to all of this. So again, thank you for joining and thank you for everyone else listening in.
Patrick Finn:
In standing, thank you. And Brendan, as always, I thoroughly enjoy our time together. You're always so on the money, well-informed, thoughtful about our business and our markets. I really do appreciate it.
Speaker 1:
Thanks for joining us. Stay tuned for the next episode of TD Cowen Insights.