Hi, welcome to the Canaccord Genuity growth conference. I'm Kyle Mikson. I cover Life Science Tools and Diagnostics for Canaccord. Please welcome you to this fireside chat with Quantum-Si. Quantum-Si is developing and has commercialized the first of its kind next-generation protein sequencer. From the company, we have Jeff Hawkins, CEO. Thanks, Jeff, for joining us today.
Thanks for having us.
For those of us familiar with Quantum-Si, why don't you walk through the company's technology vision and the roadmap, basically, as you see this industry playing out?
Sure. As Kyle mentioned, we're the first company to commercialize next-generation protein sequencing. What does that mean? We have a single-molecule detection technology that detects individual amino acids. Most measurements in the field of proteomics are sort of saying there's a protein there, its presence or absence. They're not giving you that level of fidelity. Think of it like DNA sequencing and your ability to do a whole genome. That's the type of approach we're taking. Today, our technology is leveraged for more targeted applications. The roadmap, which I'm sure we'll get into today, helps expand the sequencing output, the coverage of amino acids towards our end-state goal, which is really de novo sequencing of proteins.
Okay. You just had earnings, the second quarter earnings recently. You noted, really throughout the first half of the year, you noted obviously headwinds on the U.S. academic side. You've consistently had some positive comments and I guess traction on the biopharma side too. Could you just talk about what you saw in each end market year to date, especially recently?
Sure. Yeah, I think the U.S. academic market is probably experiencing what most companies in our space are experiencing, you know, general slowdowns in capital purchasing. On the consumable side, though, with our existing installed base, continuing to see those customers sort of purchase at their normal pace. We actually said on our call it was a little ahead of our expectations. On the pharma biotech side, you know, people are applying our technology in that space largely around something called protein barcoding. We had launched a kit in that area late in 2024. We made that a big priority given what we saw in Q1 with the U.S. academic market and really grew that funnel of opportunities during the quarter.
We had about 30 or so going into the quarter, exited the quarter with 60 opportunities in that space across both large Pharma and smaller biotechs, both here, in the U.S., and some in Western Europe as well.
Okay. With the update on the call, you talked about some new capital acquisition models, I guess for customers, especially in this academic research market. Maybe just provide a little bit of flavor for what those are and then how, you know, are those going to stay around for the long term?
Sure. What Kyle's alluding to is we began just a couple of weeks ago offering other ways for a customer to acquire our platform. If we take a step back, we think a large user base, you know, running this product, the technology routinely, publishing their data, talking with their peers about it, is a, you know, sort of a critical part of the long-term strategy. We also have seen customers who have had our instrument for, let's say, three or four quarters, so about 9- 12 months. When they've had that level of experience and utilization, they're the ones who are starting to come to us and say, "Hey, when am I going to get budgetary information for Proteus, which is our next platform?" You can tell that they're thinking about where we're going and thinking about how to incorporate it.
Given the macro backdrop, you know, having a strong balance sheet and an instrument that, you know, has a pretty modest cost to make, we had sort of two options. Customers were approaching us and saying, "Hey, we have consumable budgets. Could we send you samples?" Many people in our industry have used sort of service models. Our experience is when people use it in their lab, that's when it's stickier. That's when you get that sort of link over time. We said, "Let's give different options so people can reagent rent, they can lease the platform, and, you know, in select accounts that, you know, maybe are key opinion leaders or maybe are really great application for the tech, we may place the device and capture that consumable revenue that we would not otherwise, you know, capture if we were holding out for the capital dollars.
There's a decision window for these models too, I think six months?
Yeah, so the way we've set it up right now is if an instrument, you know, a reagent rental is different if someone signs up for that. Those are normally long-term commitments. If we're placing the device and they're just purchasing consumables, what we've told customers and what the agreement says is that we'll leave the device there for six months and we'll evaluate their usage over that time. If the usage is really strong at the end of it, some of them we hope will purchase the machine, others might move to a rental contract and continue. If somebody really hasn't purchased to our expectations, we have the ability to pull that instrument and take it to another place. One thing we've learned in the early commercial time with the platform is we've been able to move it around for the purpose of evaluations, and it's very robust.
We can ship it somewhere, unpack it at the next place, and it works. It's a fairly low risk and low cost endeavor for us if that instrument doesn't stick, pull it out, get it to another account, let them use it and generate data. You still have the options at the end of that six-month period.
Okay. In terms of, just going back to the end markets, academic and biopharma, talk about sales cycles in each of those, you know, almost like the pre-NIH kind of headwind, let's say, and then also how biopharma sort of is doing, just given the evolution over there.
Yeah, sure. I think, prior to the NIH funding, I would say the typical academic sales cycle was around three to four months. Some people would want to evaluate the technology, some wouldn't, sort of a little bit of a mix depending on just specific decision makers. biopharma was more like six, maybe nine months on the long side. What we're seeing right now is Academia is obviously stretched out and in some cases just paused. biopharma tends to be with, especially with that protein barcoding application I mentioned, it takes about 9-12 months. The reason for that is they're implementing the barcode into an active program, running it all the way through to the readout, which often means through a mouse model or some other animal model to really prove out that it does exactly what they want.
They can multiplex, they get the fidelity of the result they need, and they can then calculate the savings they're going to experience by doing it. Running through that process, on one hand, you can say that it sort of delays the revenue opportunity, but the flip side is they're committing that amount of time and energy. When we get that business, it's going to be a high barrier to entry for the next person who might want to come in and try to take that business from us in the future.
Gotcha. Okay. When you think about being almost entrenched right now, you're penetrating pretty well in this biopharma market, maybe better than your expectations. How do you expand within the biopharma customer beyond this barcoding application?
Yeah, I think it's a good question. Barcoding was a really good entry point. It offered them a type of result and capability they didn't have access to. It gives them that multiplexing. With some of the changes in the rules around the number of animals being used, a lot of the large Pharmas have initiatives to reduce the number of animals being used. Multiplexing is obviously a cost savings. If you try to look at eight payloads at once versus one at a time and like one per mouse, there's a cost saving. It checks a lot of those boxes of the large Pharmas certainly. For the biotechs, it's those benefits, but also the platform is just much easier to run. It's small and it's a lot lower cost than, say, sending out to Mass Spec or somewhere else for some type of a result.
Slightly different models, but similar sort of both economic and workflow reasons to do it. The other part you asked was just how do you, I think extending beyond that. I think the next place for us, as we expand the coverage of the technology and launch some of the kits we're working on that will create a deeper sort of post-translational modification discoveries and move more into discovery. A lot of what we're in today is what's called like candidate selection or candidate screening. As we bring out more and more capability of the tech, we can move back upstream and get even into the discovery side, the biomarker discovery side over time. I think that's the next place for us to go.
In the biologics world, you could conceive maybe not in that Pharma, but with their contract manufacturing partner being used in more of a QC type of environment for those, for the actual biologic on the downstream side. I think a couple of different ways to go with the tech as it continues to sort of expand in its capability.
All right. Thus far, how is utilization? The numbers are a little bit small, but how is utilization different between the academic customers and the biopharma customers?
Yeah, so we see a couple of patterns. In Academia, what we tend to see is more episodic use. They buy a certain number of kits, they run their experiment, their study, then they sort of stop buying, they write up their data, they publish it, or they present it at a meeting, and then they start up their next study. You see this lumpiness, buying kits, then off, then buying again. What we see in both pharma and biotech, and also in government accounts, we have a nice footprint in the Department of Defense, is a much steadier, consistent purchasing pattern. They purchase at some rate, whether that's monthly or a certain amount every quarter, and they're doing that very consistently, and the number of kits is going up over time.
A much more consistent pattern, a little longer to get them going, but once going, a more consistent pattern with some level of growth on it.
Got you. Good. You have a handful of boxes out there that have been deployed, they've been installed. How are they, you know, being, just broadly, how is utilization going? I know, you know, pull through is something that you don't want to probably quantify right now, but is this something that's in line with your expectations, or is this being, again, impacted by the macro environment?
Yeah, I think we can say one thing on the utilization. We think at scale, when the platform has been implemented, people are running it across multiple sort of research projects. The platform's capable of pulling in about the list price of the machine per year in consumables. Our machine is $125,000. That $100,000 - $125,000 range is what we would expect at scale a customer would be pulling through with this platform. We haven't given out exactly where we are today. I think to your comment, Kyle, people are tracking the way we want. We're happy with it. I think the adoption has been a little longer sales cycle than we like in pharma biotech, but the ones who are live with the platform, we've been really happy with the purchase levels. The defense side has been very strong, probably the most ahead of our expectations.
I think academic, I would say it's in line with our expectations. It's lumpy, so it's a little bit harder to say we had an exact number there, but people are, no one's going sort of dark with the equipment, which is what we measure. We measure what's the buying pattern, and then we define as a user idled versus active. We're not having people go active and the instrument just sit there, which has been challenges in this industry for many years. Something gets bought, gets used once and doesn't get used again. We don't have any sitting in that category, so we're pleased with that, but we're working to continue to raise the level across the different segments.
Okay. You launched the Platinum Pro , which is like the next-gen sequencer that you have. You launched that earlier this year, a little bit higher price point, and it allows this pro mode, basically. How has that rollout progressed? How has that storyline kind of progressed so far with customers?
Yeah, so it offers two things. It offers Pro mode, which you mentioned, Kyle, which is the ability to use our technology sort of in an open mode. If you break down our tech at the most fundamental level, you know, we're essentially an exquisitely good technology for single-molecule binding kinetics. Now we're applying that to sequencing, but there's a lot of other applications where people want to measure binding affinities. Our tech is sort of an alternative to some of the other options that are out there to do that. We opened up a channel, gave people access to the dyes, opened up the software so that they would be capable of doing that type of work.
The other thing the Platinum Pro offers is some customers, especially when you get into pharma biotech or you get into defense, they will not, they're not able to connect your machine to the cloud, and all of our analysis tools sit in the cloud. The Platinum Pro gives people the ability to do onboard analysis and not need to be connected to the cloud. That helps when you're trying to implement in one of those environments where someone doesn't have cloud connectivity or has a firewall that doesn't allow for it. So far it's gone well. We continue to sell through the last of our Platinum units, and we have a mix right now. Some people buying the Legacy Platinum machine and some buying Platinum Pro, but we'll soon here be sort of out of that inventory, and it'll be exclusively pros moving forward.
Got it. Also, on new products or pipeline, I think you have like a V3 Library Prep Kit coming out soon. You also have the V4 Sequencing Kit coming out soon as well. You keep rolling these versions out of these kits. What's the kind of the goal, the end game when it comes to having all these kits roll out? I mean, given you don't have like a ton of placed boxes, it would just seem like maybe you don't have to do a new version every several months, but why is that necessary for this type of technology?
Yeah, I think our approach has been continually improving the technology. The new sequencing kit, the Version 4 Kit, which we expect to happen this quarter, really does two things. It continues to expand the amino acid coverage, and it also allows us to cut through the amino acid called proline. For the deep proteomics people, you talk to core labs, proline-rich proteins, antibodies, membrane proteins, these are very difficult to analyze with mass spec. We've historically not been very good at those either because we couldn't cut consistently through that amino acid. This kit will do that. It's going to open up some new applications of our technology in an area that is widely desired to be studied, but very, very difficult to do with existing technology. The Version 3 Library Prep is really about lowering the sample input.
A lot of our customers to date have worked with recombinant proteins, maybe not super complex biological samples, higher concentration sort of proteins. The new library prep, we expect to lower that input by more than a hundredfold, which is really going to let us get into more complex biological samples, more complex mixtures, things that we think opens up new opportunities. Our general strategy has been this sort of steady cadence of improvement. The other thing happening in the background that often doesn't get much attention or credit to the people doing the work is over the last two years, we've brought up and insourced all of the manufacturing capabilities to make these kits.
What that is going to allow us to do going forward is really start to look at sequencing kits, perhaps as more of just a continuous improvement to the kit rather than having to do these in chunks. Each time we identify a new recognizer for a new amino acid, rather than having to sequence it into a kit because of lead times with vendors, et cetera, we can flow it in in more of a continuous fashion. We think that could be very beneficial for our customers. We're in research markets, so there's not some of the complexities there are for clinical customers. We can turn kits over and people tend to adopt very quickly. I think we'll look to move to a more continuous process on that front and then with the library prep combined, get into some of these lower concentration complex biological samples.
Okay. You also are going to roll out PTM detection kits in the next few quarters, let's say. You haven't provided the exact timing. That's been PTMs that have been viewed as this killer app on most protein sequencing or one of the killer apps. What is that? How does that, how would those kits really expand your customer base or product opportunity, let's say, and also expand your revenue per customer potentially?
Sure. Yeah. I mean, PTMs and single amino acid variants are sort of what people really want to deal with, with protein sequencing, especially if you're in, you know, a developed market like the U.S. or Western Europe. I think when you get outside of those areas, protein identification quantitation becomes more interesting to customers because they don't have access to other existing technologies. On the PTM front, we have some customers who have already published either in pre-print or peer review on the use of our tech for PTMs. Sequencing today doesn't go 100% of the way all the way through the protein or the peptides, and we don't have 100% coverage today.
One of the things we showed at our analyst day is the ability to combine modalities, meaning we can do a pre-sequencing detection run and probe for PTMs anywhere in the peptides, then move into a sequencing run, combine that data and get an even deeper sort of profile of those peptides and of those proteins in terms of their PTMs. What we announced on our call and what we'll share more data on, sort of the exact roadmap and timing at our investor day coming up in November, is really how do we see those kits rolling out? How broad could this be applicable to various PTMs?
Based on the feedback we've had from customers, this is really an area that outside of a small number of people who have sort of the highest end equipment, the sort of bespoke pipelines on the bioinformatics side, this is not really something that's reproducible and routinely able to be done in the market. We think it represents a really good opportunity both on the translational side in Academia. We think it could start to open up some of those discovery opportunities on the Pharma side. We're sort of excited to share some more coming up and then roll out those kits soon after and see how the adoption goes.
Yeah, in the PTM detection, basically like reagents, those are going to be compatible with Platinum and Platinum Pro or with the Proteus.
With Platinum Pro, and then they will also port over onto the Proteus device.
All right, so the Proteus device that's going to be launched in the second half of 2026, different architecture. It's an optics-based architecture compared to the kinetic space that the Platinum device s are based on, basically, w ith the fluorophores and so forth. Maybe talk about the development of Proteus so far, how that's gone, because you announced it late 2024 and you've given almost a two-year kind of your roadmap. How is that progressing?
Yeah, so I think maybe just on the technology side. Our existing system, the consumables, you know, sort of basically like think of it like a digital camera, right? A CMOS chip with nanowells over the top. Then it's placed into a device that, because the optics of the device are in the chip, the instrument itself is a pretty, you know, simple device, has a laser, has the electronics, the collection, the processing of data. When we go to Proteus, the consumable becomes a very simple sort of, you know, fused silica substrate with nanowells patterned over the top, and then the optics go into the machine. A couple of reasons to do this. One is we can get to a significantly higher density of features of these nanowells.
That's important as you think about getting into more and more complex samples and wanting to see more and more proteins per sample. You can also do that at a significantly lower cost to make that chip than it costs to make our current chip, which gives us a lot of pricing sort of leverage, whether that's to get business in certain accounts, whether that's for future, you know, competitive sort of defensive reasons. We sort of have optionality with that type of price point on that side. You're right. We announced the program at our investor day in November of last year. It's really probably one of the faster programs I've seen in the industry. This is something we started sort of in the middle of 2024, and we're saying we're going to launch it in the second half of next year. We feel comfortable with that expectation.
About a 2.5 year long program for a complete architecture change. The reason we believe that's feasible is a lot of the parts we think are even harder to do in our field, the surface chemistries, the recognizers development, the enzymes, the manufacturing infrastructure, all those things are in place with Platinum and Platinum Pro, and they poured over into the Proteus program. Really what we're changing is the consumable engineering architecture, but a lot of the other aspects of the technology are directly portable into that. We've been giving milestones throughout the year. The big one before the end of the year is to demonstrate sequencing on a prototype platform. We've said we'll do that by the end of the year. We, you know, feel very good about that milestone.
Assuming we hit that milestone, then being, you know, in good shape to be able to get the platform out, you know, before the end of 2026.
Okay. Just as we think about projecting that the product, let's say like, why would Proteus be, you know, shipped more, placed more per quarter than Platinum would?
Yeah, I think, you know, Proteus is maybe backing up one step. I think Platinum, right, was brand new tech into a brand new market. I think, you know, you're doing a combination of market development and learning about your tech. You're evolving it and improving it over time. I think we get all that learning as we go into Proteus. We know more complex samples and more proteins per sample is things customers want to do. That's something that will only be unlocked when we get to Proteus. I think the other piece of it is we know sort of that menu of applications we want to have, you know, the capabilities around PTMs, the capabilities for those more complex biological samples, you know, even higher plexing barcoding. We feel like we can come to market with a really robust set of applications.
We think we can get the, you know, the pricing right for each of those markets in terms of the consumable side. You know, we feel good about it. Obviously, we've got to do the work to build into that. We're going to want to pull customers into the development process throughout next year so that, you know, people have used it and run it and have data before we even launch. I think it's a project and we just have to keep plugging away at it and, you know, sort of launch it on time with that customer engagement all the way.
Okay. You've also discussed the use of AI models to develop like amino acid recognizers that'll help you expand your coverage of the proteome. I guess what is that? How do you kind of commercialize that? What's the kind of roadmap to do that and how important is that going to be in an industry where there is no reference proteome? How important is like AI going to be?
Yeah, I think we use AI in two ways. One is in the development of the recognizers that bind to amino acids, and that's what we talked about. I'll expand on that in just a second. The other is we use it in the development of the kinetic database, and that's sort of what you expect to see when you see the sequencing reaction happening. That allows us to sort of evolve the detection algorithms and improve that over time. On the internal front, I think people talk a lot about using AI, and AI tools are obviously out there in the public domain. Even more important than the tool is the underlying training data. What we talked about on our call is, over the last eight years, the company's screened over 1 m illion different binder candidates.
We have all this data about when you insert mutations into a protein, it behaves in the following way. It binds to this amino acid, it doesn't bind to this one, it has this stability profile, it has this manufacturing sort of success. What we've done most recently is taken that data and trained AI on that to generate an AI-designed binder. We did that and saw a significant improvement in our first candidate of a binder that sort of isn't even in the ballpark of what you see when you try to develop them using more classic techniques. What we think that's going to do is allow us to really go from where we are to complete proteome coverage in a much faster period of time. We've got some case studies in the works that we're going to put out to try to elucidate this a little bit more.
At our investor day in November, really try to help quantitate that for investors of what's that mean, when exactly do you think you can get to that whole coverage. We think it radically reduces the number of cycles we need to go through to develop coverage for, to get from where we are to complete coverage of the proteome.
Can you use something like AlphaFold to help, I guess, or is it?
You can use these tools. We've been using AI tools, like we have a computational biology team, we've been using these tools, but I think the big breakthrough is training these tools on this very rich set of data that we have, that, you know, these tools coming off the shelf don't have access to this data. This is all proprietary. This is our internal data generated through years of empirical work in the laboratory across many different orthogonal methods. The tools are there, but the real lift we saw was when we trained it on our internal data and generated candidates with that and then tested those candidates in sequencing.
The rich data comes, it's just because you're, you know, you're just sequencing proteins.
We've been doing binder development for almost eight years. We have sequencing data, we have other data from our directed evolution program, we have data from other orthogonal methods to look at binding kinetics on off rates. We have data on any of these that have ever gone into manufacturing, what is stable, what's not, what's easy to express or not. You have an amazing amount of data when you've been doing this for this long. If you give all of that to AI, it's able to take that into consideration when it's designing the next binder.
Is it like 0.5 million samples have been?
Over a million , over a million sort of candidate binders have been screened in the history of the company.
Okay, interesting.
It's a big body of data that's available for us to tap into.
All right, awesome. Finally, just what are investors and maybe even academics or customers underappreciating about Quantum-Si in this whole protein sequencing industry that definitely is early stages, but it could be super promising.
Yeah, I think people are probably, you know, underestimating, you know, really what the true potential could be. I think some of that's tied to, I don't think we've yet really helped people understand how different Proteus is going to be in terms of its output and its capabilities when combined with the level of sequencing improvement we're making in parallel to that. I think that's something we have to help bring to light in the market over the coming year. The other side is I think it's a much more difficult problem to solve than people think. I think because of that, the barrier to entry for others is very high. This is an extraordinarily difficult problem. You have to have talent density across so many different areas on the scientific side to pull this off.
I think some people might be undercalling, you know, the risk of sort of like how quick people will get here. I think it's very difficult to do, you know, sort of competitive fears might be overstated by some in that regard. I think it's also a market development and we're doing it. I think the data's coming out, the pipeline of data from customers looks good, and I think that'll come out over the next few quarters and help as well.
Okay, that's great. Let's leave it there. Thanks, Jeff. Appreciate the time.
Yep, thank you.