Seer, Inc. (SEER) Earnings Call Transcript & Summary

Hi. My name is Yuko Oku and I'm part of the life science tools and diagnostics team at Morgan Stanley. Before we begin, I'd like to remind our listeners that important disclosure information can be found at morganstanley.com/researchdisclosures. If you have any questions, please reach out to your Morgan Stanley sales rep. With that, it's my pleasure to host Seer, and speaking on behalf of the company, CEO, Omid Farokhzad. Thank you for joining us today.

Thanks, Yuko.

You made significant progress over the last couple of years and increasing awareness of the Proteograph suite and highlighting its use cases via over 52 preprints and publication from the scientific community. To set the stage, would you reflect on how customer discussion has evolved over the last year?

Of course, yes. So look, when we started Seer, the overarching hypothesis was that if we were right about it, that we would be able to give access to proteomic information in an unbiased way at scale, speed, cost robustness that wasn't previously possible. And we shipped our first instrument at the end of 2020, beginning of '21. And at the time, biggest study ever done that looked at plasma in a deep way was 48 samples that was published. The deepest study ever published was from Broad that was about 5,000 proteins. And we've now gone through 3 cycles of our product. We just released the latest the third generation Proteograph ONE in -- sorry, June of 2025. And with that, it's now possible to do really what we had predicted would happen, which is large-scale proteomics. And so first half of this year, we've already announced 1 corporate customer that did -- started a 10,000 sample study. We announced in Q2, the second customer, Korea University that announced a 20,000 sample study. We're in discussion with biobanks now and also soon be announcing a third one, which has already happened, which is a pilot study of 10,000 samples to pave the road to 100,000 sample study. So I think the -- we've now are able to impedance match proteomic and genomic at the same scale speed. And so the shift in conversations have been that we would have customers asking to do 10 or 20 proof of principle studies with us before they kind of adopt, I'm not seeing those anymore. And now customers are starting with our Seer Technology Access Center, which, by the way, has been a great asset. They're starting literally out of the gate with hundreds of samples at the time. And now we're in discussions with customers doing thousands or even tens of thousands of sample studies. So I think the validation and the proof points are driving customers to kind of bypass the very initial skeptical view, which scientists should be skeptical, but kind of now just adopting like the way they would the proteomic solution to for their application.

Great. Well, that's a great overview into what we'll be digging more into. There are many emerging competitors in the proteomic space, including those with affinity-based proteomic platform, such as Olink, and SomaLogic as well as those with touting higher sensitivity like Alamar and Quanterix. Moreover, [indiscernible] and PreOmics are using nanoparticle-based platform to improve proteomics workflow for mass spec as well. In the midst of so many options for proteomics, where does Seer's offering fit?

Yes.. Okay I think the -- I'll break up the offerings into really 2 buckets: targeted offerings and untargeted offerings. And before the world of untargeted at scale became possible, the only option you had if you wanted to do any study at scale with a targeted offering. And so that was Olink and Soma for a long part. And when we started Seer and we started becoming a commercial company, making it possible for the first time to do untargeted at scale, the conversations were always that the customer would say, well it's Olink or Soma and then should would consider Seer. I would say over the last 4 or 5 years, I've seen a shift where the conversations are more -- is it Olink or is it Seer? I'm hearing a lot less Soma out there in terms of customers options. But -- and so the wallet where that the customer needs to kind of open up and share is really between the targeted approach in Olink, let's say, versus Seer and actually see both platforms as really very complementary to each other. They answer different questions. And the questions are both needed for the scientists to be answered. In fact, if I just look at a company that we spun out PrognomIQ from Seer before we went public, it was to look at leveraging proteomic information for liquid biopsy and early detection of lung cancer. They were our largest customer, and we would report them on our earnings related for a transaction. They went from mid-30% of our revenue to mid-20% -- mid-30% in 2022 to mid-20% in '23 to mid-teen percent in '24. And if I look at '25, first half of the year, they were 6% and we'll probably finish the year with them being low single-digit percent. But what that means is they started off using Seer for discovery and then they shifted to targeted for their clinical and LDT. So both Olink and Seer have value to the customer. I think for discovery purposes, you need an untargeted approach for content discovery. But I think once you know what you're looking for, targeted is perfectly fine. Now you mentioned kind of other need to follow-ons that came after Seer did. Frankly, it's flattery for me that our approaches are being replicated. I mean -- but those are -- those me-too ones, they don't have the performance that Seer does. And they do create confusion and distraction in the customers' eyes because they go to them and say, here's something that looks similar to Seer and they offer it at, I don't know, 1/3, 1/4 to price. But then if you look at, a lot of our customers have begun to actually publish comparative studies, the performance of those things are terrible in terms of depth of coverage, reproducibility, batch to batch variability and robustness that one needs because if I look at a company like PrognomIQ or most biopharma companies for that matter, the most valuable commodity they have is their biological samples. I mean, PrognomIQ probably spend $2,500 for every patient sample that they collect it. So if you offer PrognomIQ, run Seer for a price of X, or run this me-too follow-on for 30% of X, they will never choose an inferior product because their major commitment financially was under sample. Seer is a very small part of that in terms of cost. But more importantly, the lifeline is developing a test. And if they use an inferior product and never end up with a test, boy they wasted hundreds of millions of dollars of invested capital. So I'm finding those solutions to be more of a distraction than relevant from a customer perspective. And I always say, if you give people a rope that's long enough, they hang themselves anyway. And so what's happening is instead of us needing to do anything about this, me-too follow-ons because we have a very robust IP portfolio. I'm seeing customers beginning to publish comparative studies and those comparative studies kind of answer the questions for most people. And those studies are now in the public domain by many customers that are unrelated to us as well.

And you mentioned complementary use cases using the unbiased approach for the discovery purposes and then maybe going into affinity. Do you ever see it going the other way around where you see -- maybe you see the affinity population that scale discovery studies using affinity-based approaches and then using mass spec-based Seer kind of approach to probe more deeper into the results, translational modification or other protein-protein specific interactions?

Yuko, I guess there's always an outside case where somebody may do X, and that X may be very different than what everybody else does. So I'll never say never, but I think that's kind of backward, meaning, when you're looking at something like the proteome, where the complexity is massive, and we know a tiny bit about it in terms of its content. You need untargeted approaches for discovery, not the other way around. Targeted approaches have no discovery power because you're interrogating the same thing. If you look at the prognomic test, that test will never exist without Seer. If you look at the biomarkers that were discovered for early detection of lung, majority of them were not in the public domain to go pick from, but once you found them, a targeted approach is a perfectly fine approach to utilize them. So now if you happen to like a targeted approach, sorry, a particular protein that you're interrogating a targeted approach, and you now want to go look at PTMs of that or I don't know, maybe protein interaction with that, sure, you can use an untargeted approach. But that, I would say, is 1% of the value add of that approach, meaning the lion's share of value proposition when you look at a hypothesis-free approach, untargeted approach is seeing things that you were not seeing before. And so I don't see the customers going from a targeted approach backward toward an untargeted approach. I would say for every -- probably 1,000 people, I would see 1 in that direction, I might see 1 thinking in the other direction because there's also other tools that one could use for protein-protein interaction, et cetera, other than this approach, that are easier for customers to adopt. So I don't see that Yuko actually.

Got it. Okay. Makes sense. And I wanted to dig into your product offering since you introduced a number of them this year. So starting with the Proteograph XT and cell lysate application, you launched this application early in the year. Could you provide some feedback on the application that you heard so far? And then how much did the ability to look at cell lysate theoretically expand the application samples that could be analyzed on the Proteograph?

So what we released at the ASMS conference, American Society of Mass Spectrometry Conference in June was 2 things: one, second generation of our instrument is the SP200. And then the second thing was Proteograph ONE, which is our third-generation assay and then we introduced a new assay, which is Proteograph assay direct, which is for cell and tissue. So let's break those 3 and say what each of them does. So in the first iteration of our instrument SP100, the key innovation was that Seer-enabled nanoparticle capture of intact whole proteins by compressing the dynamic range. That protein capture happened at the beginning step. And then as the assay moved on, it would get washed and then those protein get digested into peptides. And then the peptide will be purified through a method that was not proprietary to Seer, that method required kits from thermal for purification of the peptides, which would then go into the mass spec. With the second iteration of the instrument, the SP200, the workflow is now end-to-end Seer. So at the beginning, the nanoparticles do whole protein capture. So -- and this is relevant because if you're interested in PTMs, because you're capturing the whole protein, you would capture PTMs of that protein as well or variants of that protein. It then becomes peptides on the back end instead of using that other kit for peptide purification, we now have proprietary engineered nanoparticles that now do the peptide capture. So we have proprietary nanoparticle for protein capture and for peptide capture. So the total workflow is now end-to-end Seer. It's obviously for the proprietary and data generation and IP perspective, much stronger. So that's the instrument. The workflow also got compressed from about 8 hours to 4.5 hours in automated way by eliminating the -- optimizing some of the assay and eliminating the need for that additional peptide purification in the back end, we were able to compress the workflow. The other innovation was that the assay went from one run of the instrument being 40 to one run now being 80 assay because each sample is being analyzed through 1 well and a well of a 96-well plate has a multiplex nanoparticle that a single well does their job of multi-wells. And so that's the Proteograph ONE then enabled this assay to go through so that in 4.5 hours, you can sample 80. Now the direct product, the purpose of that was that customers were saying for the footprint that is occupying in my lab to run your assay, I would also like to be able to run other proteomic assays that may not require your particles, but are helpful to me, let's say, in cell and tissue. Now where Seer's value proposition is very distinct is when you have a high complexity sample, by compressing the dynamic range before it goes into the mass that you can see a lot of content. If you look at less complex sample like cell lysate or tissue where the dynamic range is much, much narrower, the relative value add of using proprietary engineered nanoparticles for content is smaller. And so the real value proposition is the automation. So the direct assay does not require our particles, but it runs the assay on our instrument so that for a very small dollars, the customer can take away a manual workflow that with it comes inherent reproducibility that happens with a manual assay, complexity of a workflow and it let you do cell and tissue on our instrument, broadening the utility of that instrument that occupies the footprint, and that was really kind of reacting to the market and reacting to what the customers' needs were. Now with every new product offering , Yuko, as you know, it takes time for adoption. So the adoption of the direct assay is foremost going to require an expansion of the installed base of the SP200 because the direct assay only runs on the SP200. So as the SP200 installed base grows, my expectation is that Proteograph ONE will feed the folks that are looking at complex sample and the direct assay will feed the folks that are looking at relatively less complex sample on the same instrument.

Okay. Got it. And before we get in -- dig a little bit more into Proteograph ONE, as we think about your workflow and menu broadening over time on direct, what are some other assays that you could offer that would enable customers to further their proteome?

Yes. So I mean, we've now done 10 -- about 10 white papers in various different kind of expanding low-volume animal model organisms, cell and tissue and various other ways of leveraging our platform, by the way, including also PTMs in collaboration with Professor Kelleher's group. So my expectation is that over time, Yuko you're going to see us addressing what the ask of the customers are. So what are they? So number one, the customer wants to see content. The customer wants to understand the value of that content that's different than seeing the content. And then the customer wants to do that efficiently. So every innovation and every advancement we make is along the line of one of these three accesses in terms of delivering the value that the customer wants to see it in the Seer platform. There's obviously interest in PTMs. And Seer's technology uniquely allows that because protein capture happens at the whole protein level where PTMs can be captured. And we're certainly thinking about that. But I think Proteograph ONE being a new assay has a lot of legs to run, you can see some almost like a label expansion, if you would, in that leveraging the Proteograph ONE in different ways for customers. It's probably the direction we're going to go in the next 12 to 18 months. And then trying to increase the throughput and adding flexibility like seeing PTMs that may be of interest to them, et cetera, should be forthcoming.

Got it. Okay. So going back to Proteograph ONE. So this workflow basically doubled the throughput to 1,000 samples per week, reused 1 time by 30% to 4.5 hours compared to XT. How does the workflow, if any, differ from Proteograph XT?

Yes. Very similar workflow, Yuko, with the following 2 exceptions. One, XT required 2 wells per sample. And each well would be separately injected into mass spec, so XT required 2 wells per sample, plus 2 injections of mass spec for samples. Proteograph ONE is 1 well per sample, 1 injection or mass spec for sample. So it halved the mass spec time, that was doubling the mass spec throughput, while concurrently doubling the throughput of the Proteograph because instead of now running 80 samples, you run 40 samples. But it also almost doubled the throughput in that the previous assay was, 8.5 hours is not 4.5 hours. So it became half the length, double the numbers. Mass spec time went down by half. The -- in terms of the chemistry differences, there's difference in the engineered particles. And we now also added our proprietary engineering nanoparticles for peptide purification in the back end as well in the Proteograph ONE. Now that all said, it's improvement in terms of throughput, speed, reduction in cost but in terms of content, the customer is not compromising because they're seeing the same number of proteins. They're seeing it with a very similar reproducibility and precision. And so if you happen to be a customer that is not a high-volume customer, then XT is actually still a perfect assay for you. But if you happen to be a high-volume customer in terms of your need then likely you would shift to Proteograph ONE. So if you're a population skills customer, you want to be on Proteograph ONE just because over time, in tens of thousands of samples, cost and time become very different between the an XT and ONE. But if you happen to be running hundreds of samples at a time, and maybe you run a couple of thousand samples in the course of a year, 2,000, 3,000, then it doesn't really make a difference to you overall. And you may continue to be on XT versus going on ONE. We'll continue to support both customer types and there are customers that have now shifted to ONE, meaning they upgraded to the next instrument. The customer that says, we want to bring in ONE in addition. But my expectation is Yuko that over the course of the coming couple of years, a lot of customers would probably shift to the ONE because you're getting the same information just a lot faster and why would you not do that?

What proportion of customers do you think will stay on X? Like what do you think the steady state mix to be?

I think it depends. I think if you happen to be a customer in the midst of a study, then you may not transition until your studies are done to the next assay, right, just for steadiness of your data. We certainly are seeing customers saying, boy, that sounds great, but I'm really, really happy with the XT, and we're in the midst of the study, and so we'll support those customers for the foreseeable future. Any new customer is going to get the ONE. And a subset of the customers have already opted to upgrade, but the sample size, Yuko is too small for me to have like a very precise number. But the general theme is, if you're a mid study, you tend to continue it. If you're starting from scratch, odd are you would pick the faster, more efficient system available to you.

Okay. That makes sense. Are there any price differences between price per sample or margins with Proteograph ONE version of the assay versus the XT?

Definitely, there is. So our COGS are significantly better would be with the ONE versus XT. It's 1 well versus 2 for us. And we want -- and a kit is 80 samples versus 40. And so a lot of those reagents like enzymes and buffers is the same volume, but now it does twice as much. And -- so from a COGS perspective, we're better off and that obviously helps our margin. That margin expansion can give us flexibility in terms of pricing if we want to. We have not found the need to discount because if anything, we're giving a better product to the customer. But the cost saving comes, for example, and the fact that they'll use half as much mass spec time. They're saving money because from an FTE perspective, running a Proteograph, it's half as much time of an FTE to run twice as much of the samples. So definitely, there's a cost saving in terms of operating the system and also running the mass spec. But from a kit perspective, pricing is very similar and the margins are better.

Okay. Great. And then I wanted to move on to some of the population scale studies that you announced. So you announced that Proteograph was selected to run a 20,000 sample with Korea University. From the press release, it sounded like they evaluated several proteomics platforms before settling on Seer. Tell us key features of a Proteograph that resonated with Korea University and ultimately led them to choose your platform over other proteomics solutions.

Yes. Yes, thank you, Yuko for that. And you're right, they did kind of boil the ocean and I'll give you one other boil the ocean in just a second. But in their case, they definitely boiled the ocean and looking at various different platforms. But at the end of the day, the reproducibility of the assay. And by that, I don't mean if I run your sample over and over again, of course, we're reproducible there. But by that, I mean, I need to be reproducible in the long run, meaning if you're running a multiyear study and you're going to cross manufacture lots of mine, that reproducibility needs to be high for longitudinal long-term study. The study needs to be robust. Let's say you are multisite study, and you may be running in hospital A and they're running in hospital B, different operators, different instruments, you need to generate same data. I mean the robustness of the Seer platform is fantastic. And nothing else comes close to that for unbiased proteomics. And it does that by the way, when you think of the CVs of Seer's, when you hear CV numbers, let's say, on an ELISA assay, it's usually 1 ELISA 1 antibody detects 1 analyte and so you look at the CV of that, you repeatedly mix that antibody with the antigen and you see how reproducible it is. But when we talk about the reproducibility of Seer, the CV is being measured not across 1 analyte, but across almost 10,000 analytes. So you're seeing CV reproducibility across 10,000 proteins that may be seen in plasma. So the robustness of the CV, you have to also put it in the context of how much am I seeing. So the reproducibility robustness was very important for the folks at Korea University. The automated system made a scalable for them. And by the way, I'm very proud to also share with you, and I'll get into it in more detail over the course of the coming maybe weeks. We just had another customer, a government entity that wanted to fund a multi-institutional team, multi-tens of millions of dollars study, and they boiled the ocean for proteomic solution. This is the U.S. government. And many of these multi institutions had comfort with the other proteomic platform, which I will not name because they've been around longer than we are. and yet the final solution that got picked for that gigantic study with Seer, which is fantastic for me to see because that is a highly informed group of scientists with extreme level of experience with the other platform, generating a lot of body on the other -- body of bid on the other platform and yet conclusively reaching a scientific answer, which is for our purpose, which is discovery of large amount of content when we're spending tens of millions of dollars, this is the right solution. And so anyway, that's great. So...

Great. Looking forward to hearing more on that. Are there other large -- as you mentioned in the opening comments, it sounds like you're getting a lot of traction with these large-scale projects. Are there any others that we should keep in mind and kind of monitor as we go through the remainder of the year?

Yes, we're definitely in discussion with biobanks. We've already signed one which we'll discuss more. I mean it's -- and we'll let the biobanks themselves kind of disclose it. But it's a 10,000 sample pilot to become 100,000-plus sample study. And then we're also in discussion with others with multi-tens of thousand samples. So I had by the way -- by the way the first time I predicted that we would be at 10,000 sample plus was at your conference. In fact, it was exactly a year ago at your conference, where I said, I think 2025 will be a year where we will see for the first time, population skill studies get done at tens of thousands of samples. Well, I'm thrilled to say that the first half of the year that we finished, 3 such customers were signed in 2025. So let me be also here to say that I predicted just a couple of weeks ago at the Canaccord conference that I think 2026 is the year that we'll see the first 100,000 sample study in terms of unbiased proteomics, Yuko. Now of course, a study like that, it's going to take time to finish. But I think studies like that will be initiated because the innovations that happen at Seer but also improvements that our colleagues in the mass spec space are making is really making it possible that a biobank can actually realistically choose to do this in an unbiased way because the speed, the depth and cost all make it possible for them to do a 100,000, 200,000 sample study in an unbiased way using a mass spec.

So macro environment continues to be pretty challenging for instrument vendors broadly within life science tools. Could you provide some color around how the environment has trended since about a year ago in terms of sales cycle and funding availability? And how does the dynamic differ between biopharma and academic customers?

Yes. I mean the macro picture has been challenging for multiple quarters for us and our peers. And I think that continues to be the case. There's volatility on top of that with tariffs and questions about NIH budget. You cut the budget, then it comes back and then tariffs are there, and then they come back and then they go away again. And so all of that makes any reasonable customer kind of pause or at the very least slowdown in terms of the decision-making, and we are observing that as everybody else is. I think what has helped us and I want to be cautious because the path isn't going to be linear for us. and we'll go through ups and downs, no question about that. But what has helped us is in the setting of that headwind that's been reasonably strong, our tailwind has continued to get stronger and stronger, which has helped us kind of maneuver this and push us through. We'll continue to have challenges, I think, over the course of the coming -- certainly the balance of the year, but maybe even into next year, as we maneuver this, the macro picture, Yuko.

Great. And then you ended 2Q with a healthy balance sheet with $263 million in cash. Walk us through your capital allocation priorities between share repo, M&A or internal investments?

Yes. Look, we've been very disciplined about the way we spend money. Despite of our balance sheet, our cash burn, and I say that in the context of free cash flow, it's come down year after year. This year, we'll probably do low 40s. And my expectation is that next year will be even lower than that. It will be somewhere in the 30s. And this is in the context of heavily investing in innovation and product development. We launched our third generation of products just a couple of months ago. And what's in our R&D road map is incredibly exciting, and I'll make announcements about that over the course of the coming months. And we -- and given our balance sheet -- and frankly, no debt, we are constantly being asked by others to look at opportunities for us to do inorganic growth. And I have no ego in this game nor do I think that we have a monopoly innovation. And so we're always looking for other verticals to kind of add that is synergistic and valuable to our customers. And I've seen a lot of things come on my way. We have not acted on any and the reason for that is that I have not found any that we would want to deploy our own cash on, but we're always looking. What I think has been the most incredible value that I do want to invest in is I do our own share. We've now -- we authorized a $25 million buyback, as of the end of June, we had done $20 million of that. My expectation is that we'll try to finish the rest of that $5 million balance of this year. And if the stock continues to be where it is, I hope the Board would authorize another $25 million buyback after that. We bought back about 10.8 million shares of the -- as of end of June, which is roughly about 14% of our stock. And we have more than enough cash on the balance sheet that I think we're going to get to a cash flow positive company with a significant cushion. So we'll continue to invest internally. We'll continue to look outside and we'll continue to buy back our stock.

Right. And then I wanted to wrap up with a quick like bigger picture question. How do you anticipate proteomics to evolve in the future with new emerging technologies, improving scalability of proteomics as well as increasing the number of targets that can be identified on the platform? And what will become the key differentiating factor in your view for those that take majority of the market versus those that are limited to niche applications?

Yes. Look, I break up this space in discovery, translational clinical. And for discovery, you need to find new content. When you're starting in a space like proteomic, we're the tip of the iceberg in terms of content, then the question is what solution offers content discovery. I think the only solution that offers content discovery is untargeted approaches and to that Seer is a leader, frankly, the only person or the only organization that can deliver on that. When you shift on translational clinical, the need shifts a bit in that for those applications, you need to reproducibly and robustly interrogate a set of known proteins to do it efficiently -- and by the way, it isn't always that you need to be the most sensitive because sensitivity gives you an edge if you happen to be looking at something that is very low abundant, but a lot of important proteins may not be low abundant. But you need to do it in a reproducible, scalable, robust way and you need to do it very cost effectively. So those targeted approaches will dominate there. And I think the real money at the end of the day is in the clinical part of it. And so the platforms that can grab clinical are going to take a significant part of it. Now if you happen to be a platform that can go all the way from discovery to clinical then I think you will likely dominate that. And let me give you one prediction here, which is when content discovery velocity is slow, frankly, relatively speaking, at snail's pace, then targeted approaches have the time to discover an analyte-specific reagent for a particular protein of interest. When content discovery velocity picks up as it now has because of technologies like Seer, you no longer have the time, the capability or the dollars to begin to interrogate analyte-specific reagents for all that content that is forthcoming. And I think, and that's a prediction that we're going to see innovation comes on the mass spec side, that it's going to leverage a lot of those targeted -- specific targeted approaches. that they will actually effectively compete with the targeted panels today. And so I think the jury is out with who's going to own from discovery all the way to clinical. Certainly, discovering these untargeted approaches. So I don't think any of the targeted approaches are going to get there. But the question is, would an untargeted approach get themselves all the way to clinical or not, and I think time will tell.

Great. Thank you so much.

Thank you. Really appreciate, Yuko. Thank you.