Tempus AI, Inc. (TEM) Earnings Call Transcript & Summary

Okay. Welcome back, everyone, to the 2025 Stifel Healthcare Conference. We are back on the Life Sciences and Diagnostics track. I'm Dan Arias. I'm the Life Sciences and Diagnostics analyst here at the firm. We're happy to have Tempus AI with us. Speaking for Tempus AI is CFO, Jim Rogers. Jim, thanks a bunch for agreeing to join us here today.

Yes. Thanks for having us.

Yes, my pleasure.

What I have been doing is sort of starting with the quarter for a lot of these sessions. But for this one, what I want to do is actually just start with sort of a high-level question about AI because we get a lot of questions about AI. As you can imagine, Tempus is one of the companies that like fits the bill as an AI company. So can you just maybe talk to us about how you use AI on the diagnostics side and then also on the biopharma side? What those capabilities really do bring to you?

Yes. So we get this question quite often, as you can imagine. And AI is obviously a very broad term. But we've been focused from day 1 on how do you leverage data and technology to really impact patient care. And so AI is really embedded in everything that we do. A few examples. So on the diagnostics side, we have a platform called Hub that our physicians can log into. There's a feature in there called Tempus One that allows them to interact with the diagnostic test. So they can ask it questions, what does this mutation mean, show me the guidelines when this mutation exists, what are the side effects of this therapy. It really allows them to interact directly with the diagnostic, and that's a patent we filed, I believe, in 2019. So it's something that we've been working on for a number of years. Obviously, we're the beneficiaries of some of these advancements in technology. On the data side or data licensing side, the same thing is true. So we have a software tool called Lens that our researchers can use to kind of build cohorts. Again, they can talk to the database, ask it, how many patients have this mutation, build this cohort for me, again, embedded in kind of the core business. The next step in that is since we have this multi-model database, you can kind of create these algorithmic diagnostics or AI-enabled diagnostics that are purely just algorithms that look at different data modalities and can be predictive in nature. We have a number of those that are in market today. And then lastly, we announced earlier this year, building a foundation model with AstraZeneca and Pathos. There, we're leveraging the entire database and really building a model to identify insights. So you really can't see if you're not looking at massive amounts of structured data, multimodal structured data. We believe over time, that will be catalytic to both of those businesses because you're going to find insights that we can embed back into our core products.

Okay. And so on the diagnostics side, it's fair to say that the clinician gets information that is AI informed. Because one of the questions I get is, okay, so late-stage cancer, lung cancer patient comes in the door. You want to know whether his or her cancer has an EGFR mutation. You could get that information from several different panels that are available in the market, but it sounds like the oncologist is more well informed in certain cases, maybe not in the specific EGFR case, but maybe cell. I don't know. I'm just sort of like thinking about how it is that your information that's provided to a clinician would be AI informed and differentiated.

Yes. So from a sequencing standpoint, there's a number of companies that are obviously great sequencers. We all use Illumina equipment. We all can make those types of calls. For us, it was always that is one piece of the puzzle, and you need to connect into the institutions, into the EHRs to pull out other data for that patient because then you can contextualize the result. So that means if we recommend a clinical trial to a doctor for specific patient, we've looked at the inclusion/exclusion criteria that others would have a difficult time doing so due to the lack of those data connections.

Okay. Very helpful. So now let's talk about the quarter. Third quarter was a 28% organic growth quarter for you guys. It was high 20s on genomics. 33% on Ambry, I believe. And then mid-20s on the data side, margins expanded. EBITDA was positive, a couple of million below Street, but positive, which is an inflection that we'll talk about. How would you characterize where you are with the business and what drove the 3Q results? I'll ask you a bunch of specific questions that you can take as you decide to, but at a high level, what would you say about where we are now with the business?

Yes. At a high level, it was a great quarter for us within -- so our genomics business is broken down into oncology and hereditary. Oncology volumes grew accelerated in terms of growth again to 27%. We're still seeing obviously some ASP tailwind, so revenue growth was slightly higher than that. And then the hereditary -- Ambry has been the beneficiary of some disruption, obviously, in the hereditary screening space. So they had a really strong quarter. We've highlighted that, that will moderate as we get into Q4 as those share gains are compressed. And then on the data side, the data licensing business is performing really, really strong, about 37% growth in the quarter. We do see some headwinds in our CRO business, which we highlighted, and I'm sure we'll get into. So the total data and services growth rate was kind of mid-20s. But the one that drives the majority of that business is data licensing, which is growing higher than that. So we were thrilled. And then lastly, on the adjusted EBITDA side, we long said that our goal of ours was by the time we turn 10 to be adjusted EBITDA positive. We have turned that corner even with adding in some additional kind of Paige expenses with the Paige acquisition. So we're really happy with where we are.

Okay. Helpful overview. Let's dive into therapy selection a little bit, which is kind of like the bread and butter for a lot of these companies that are doing either tissue or liquid. It's a very established market, reimbursement is good. There are several elements of the business that you have. And I'd love to just see whether there are some things that we can kind of point to say, "Okay, here is why volumes are accelerating." There is a tissue assay. There is a blood assay. First question is, are you seeing some of those start to be run concurrently in a way that they haven't? There's tumor normal sequencing. You offer an assay that does that. Are you starting to see that be something that's more frequently run in your case? There is a large panel, there's a small panel. Could you talk about whether or not there's a shift there or something going on that is helpful for you. So essentially, what I'm asking for is like are there things in the therapy selection portfolio that maybe don't rise to the level of obviousness, but that are like sort of underpinning this across the space, really obvious acceleration in therapy selection usage.

Yes. So I'd start by saying you've seen all the -- our public comps have had really strong quarters as well from a volume perspective. So clearly, the entire market is expanding, and we're a beneficiary of that. We haven't seen any significant shifts within ordering habits or kind of mix within our therapy selection assays. All of them grew at similar rates to what we've seen previously. Certainly, when we introduce a new product, for example, when we launched the larger liquid panel a few years ago, there was a shift from kind of smaller to larger just because the doctors tend to like more information, but nothing within the quarter. For us, a lot of it came down to kind of sales force execution. We hired a lot of reps in kind of the beginning of 2024. They're now kind of hitting their stride, which we noted in Q2 and that's continuing in Q3. But no significant shifts in terms of ordering habits or increased concurrent testing or anything like that, really strong growth kind of across the entire portfolio.

Okay. On the blood side, you have the 100 gene panel, 100-ish and then 500-ish. I think it's like 523. The larger panels tend to be the pharma-focused ones. Is that the case here? Or is it a mix of oncology use and biopharma use?

It's definitely a mix. I think that when we originally brought the larger panel to market, it represented a smaller percentage than it does today, but you do have oncologists that migrate to those larger panels just given the additional information that's available.

Okay. About hospital connectivity. I mean when we first started doing our work, it became very clear that, that was going to be an advantage for you just in the sense that you are touching quite a few institutions in a meaningful way. Is that a number that's stable, that's growing? And how important do you see that as a driver going forward? How much greenfield do you think there is when it comes to just sort of like reaching out and touching them, but also getting yourself really integrated into their EMR systems?

Yes. So the number of connected institutions has grown. It's over 5,000 today. Back several years ago, it was probably a few thousand. So we continue to enhance those connections each quarter. We've reached a scale, we're obviously touching a large percentage of oncology patients in hospitals broadly. But for really people to get the true value out of the Tempus platform, those connections are important. So it's obviously important to us because we get the data, but that also allows us to make these intelligence or make these diagnostics intelligent and return those results. So that's why these things kind of get built is. The hospitals and institutions really see the value of doing those connections, and that also just makes the platform more sticky.

To what extent when you sort of survey the landscape of what it is that you're competing against, to what extent do you see EMR integration and just the ease of ordering and sort of streamlined behavior for the oncologist? How much of a decision-maker is that or decision point is that for oncologists? And would you say that that's something that kind of sticks out for you guys?

Yes. I'd say for us, we think oncologists -- you have to cross a bar with the performance of these assays, right? And all the big sequencers obviously are good at doing the sequencing. Where we win is really ease-of-use, connectivity being a one-stop shop, so kind of having a very broad offering that meets all their needs. And then what additional insights are you able to provide as a result of that connectivity. So when we kind of bump up against competitors, that's where we end up kind of winning share. And I think for doctors, they're incredibly busy individuals. And so they're really looking for who can give me results very quickly, very easily and provide additional insights, and that's why we've kind of designed the platform as well.

Okay. Let's touch on ASPs. I know we're bouncing around a little bit, but I've got a whole list here, I got to try and 19 minutes work through as much as possible. ASPs have been on their way up. They're right around 1,600 now. I think we started the year closer to 1,500. You've got the ADLT reimbursement trend for you on the xT assay. I think you've said 40% of total xT volumes should be ported over to the ADL price by the end of the year. Is that still the target? And then the follow-on is that or the follow-on question is when you say the majority should be shifted over in 2026, am I thinking more like 55%? Or am I thinking more like 85%?

Yes. So the reimbursement landscape obviously is always changing, but the long-term reimbursement trends are certainly going to be in our favor. So for xT, we have about 30% today. Target was 40%. We'll see where that ends up, but we're on that track. And then by majority, we mean the vast majority. So closer to that 85% by the end of '26, not 55%. So within solid -- the tissue DNA test, we have ADLT status. We're going to submit the liquid biopsy, the larger liquid biopsy panel to the FDA shortly here. And so that will be another catalyst as we get into '27 and then followed by kind of the RNA panel after that. So if you look at our reimbursement relative to our peers, we certainly are still lagging, but we're doing all the things that they did. We're just kind of earlier on in that journey.

Yes. The broader trend is certainly up across the space. But to your point, I mean, as far as shorter-term mix goes the xM assay and the xE assay will be things that sort of I don't want to say headwinds, but they will tamp down any increase that you're getting elsewhere?

Yes. So for xM, we have that in front of MolDx right now in CRC, hoping for reimbursement shortly here. But as you scale, there certainly are some headwinds there. But again, longer term, as you kind of look across the entire portfolio, we're certainly behind our peers and looking to close that gap.

Have you ever thought about renaming these assays in a way that makes sense?

I stay out of the naming of the assays.

Humble opinion from analysts would be anything you could do to make this easier would be great.

Fair enough.

Okay. Maybe on MRD, the early traction for you guys has been good. This is a really highly competitive space. It's going to get more competitive as we go along. And I'm always asking you about data generation or at least you guys about data generation because from where we sit, it seems like it will become increasingly important to put something in front of a clinician that's now got 10 MRD options, and try and convince him or her that this is the assay that you should use. How do you see -- what is your philosophy or what is the firm philosophy on data generation, clinical evidence? I say this knowing that your MRD assay is doing well out of the gate. And you've got a couple of studies, but it doesn't -- it's not voluminous the way that some of the other players are in the market. Do you think that, that's something that you need to do more of going forward? Or is what you have subset analysis of CIRCULATE study? Is that going to sort of carry the weight going forward as this field evolves?

Yes. I mean our view on studies, and we've had this conversation in the past is that we certainly do studies to prove clinical -- or validate the assays, clinical utility and certainly the study is necessary to secure reimbursement. We take a slightly different approach than the market in terms of kind of running these very large studies. We choose to make our investments kind of elsewhere on the technology and data side than on these large studies. But we're the beneficiaries of kind of the work at large across all of these companies is demonstrating the validity and the utility of these types of tests. And we often point to therapy selection. The early movers and therapy selection had to run very large studies to convince payers to pay for these tests and then to be included in guidelines and all those things. And the companies that were later on didn't have to do the same amount of investment to secure reimbursement. And we think the same is going to be true in MRD. And when we started in therapy selection, there were very large established players in the space that we were able to displace by leveraging our data connections and the insights and ease of use. And we really think that the MRD market will play out similar. It would be odd for us to be -- as we're sitting here a few years down the road to believe that somebody would choose somebody for therapy selection and somebody different for MRD. We think that those are going to be very related decisions and one that we should be able to carve out a piece of the market.

Okay. So the idea is draft behind the work that has been done in order to sort of get the field where it needs to be. And then to your point earlier, rely on sort of the comprehensiveness of your portfolio.

Yes. And I would say that we don't run -- I mean, like we publish a lot, but we just take a slightly different approach with these very large studies.

Yes. There will be some studies coming out. I don't know if it's 2026, whether you've said when we should expect data. But you have an ultra-high sensitivity assay or a higher sensitivity assay that you're working on, tumor-naive for MRD. It did sound like there was a study that we would -- that we hear about at some point. Is there a timing that you would kind of orient us on in terms of when we could start to hear more about that?

Yes. So we're working on a second version of our xM assay. First indication will be non-small cell lung cancer. There will be data coming out in '26. We haven't specified exactly the timing there. And then that will follow with breast and IO, so -- which would be in '27. So a lot of work going on in MRD. We obviously are with CRC to start, but highlighting for folks that there are more things coming, and we're excited with where those assays are turning.

Okay. The next assay, the Personalis assay that you're kind of aligned at the hip with those guys on is focused on lung cancer. Is that a situation where there needs to be anything different that happens internally when you have 2 assays pointed at the same cancer type in the same application?

So we don't think that we have to kind of worry about that. There's going to be kind of a bifurcation within the market where certain physicians prefer tissue, and they want kind of that ultrasensitive test that -- of which Personalis is one. And then you're going to have other doctors that just prefer give me a result very quickly. And so again, similar to therapy selection, we think liquid and solid tumor naive are both going to have a large market going forward.

Okay. I'll ask some number of questions. I'll say the clinical details for another conversation. But I do want to touch on Ambry because the growth is really good. I mean I think if I go back to our model in 3Q, most of the outperformance came on the Ambry side. When you did the deal, the idea was that, that would be a high-teens grower. And I think in the quarter, you were more like low 30s, if I quoted the numbers right early on. Some of that is due to market share change, that's my belief based on what you guys have kind of talked about. But some of it isn't, can you just sort of take apart those 2 things and talk about, a, how you get comfortable with what is versus what isn't market share change? And then b, where the rest of the volume acceleration is coming from?

Yes. So obviously, parsing out market share versus organic growth is a little bit difficult, but we think it's about 50% organic, 50% coming from some of the competitors in the space. Obviously, there's been some disruption with some of those competitors and Ambry has benefited from that. So we've told people that is going to moderate those share gains can't continue forever. But Ambry has a best-in-class kind of hereditary offering. So I think naturally, as some of that disruption occurred, people kind of migrated to the Ambry platform. So the hereditary space performed, and we think will continue to perform well. One thing we've noted is a good opportunity for them is in rare and undiagnosed disorders. Still a relatively small percentage of their testing volume today, but one that we'll look to double down on in '26.

Yes. Okay. And so if you take out the market share piece, are you able to sort of give us a thought on the long-term growth rate on Ambry?

Yes. We said at the quarter kind of low to mid-20s. Obviously, the hereditary kind of market within cancer might be on the lower end. But as you kind of layer on some of these other disease areas that improves a little bit.

Okay. Have you put a number on the rare disease and peds portion of that?

We haven't disclosed that publicly yet.

Okay. Maybe moving on to the data side. You grew the total contract value that you had booked thus far by $150 million in the quarter. That gets a lot of focus because up until the point that you do that, it's kind of like a low visibility exercise to understand what's going on there. Can you talk about how that bucket of revenue or future revenue grows, where it's coming from? And then how we should think about sort of the frequency of step-ups from here just so that when investors do start to think about where things are going or are not going, they can gain comfort with that -- that bucket of revenue that's sizable, but low visibility?

Yes. So I mean on the data side, we disclosed two metrics. One is total remaining contract value and one is net revenue retention. The total remaining contract value we highlighted in Q2 was north of $1 billion. Net revenue retention in 2024 was 140%. Those are annual metrics because, as we've noted before, there's always some fluctuation in terms of when these deals get signed. But we now have two very strong quarters back-to-back. We signed the $200 million deal in Q2 with AstraZeneca and Pathos. Over $150 million of contracts signed after Q2. And the nice thing about Q3's number is it was a combination of different factors. We had a new biotech customer that signed a $66 million deal, a couple of existing customers that expanded their relationships, both biotechs. And then also a association effectively that is licensing data as well. So expanding beyond kind of just the traditional pharma and biotech side. We also noted at the quarter that the pipeline remains very strong. Q4 is always the largest quarter for us, typically, from a revenue perspective, certainly, but typically a strong bookings quarter as well. And so we are continuing to see an uptick in engagement both across big pharma and biotech, such that we're excited about where that business is at.

Okay. There is a re-up decision on the part of -- GSK and AstraZeneca are two big partners for you. They've been sort of cornerstones on the data side. There's a re-up decision that's on its way, not imminently, but in the next couple of years. Is there any reason that you could see why they wouldn't decide to re-up on a contract?

Yes. So AZ and GSK have been two of our longest-standing customers. We don't have any reason to believe that they would not re-up. These strategic collaborations are interesting. Because typically, what happens is someone starts by kind of licensing one cohort of data, and that expands over time. And these examples have kind of reached where they've embedded it across their entire portfolio. And the reason why they kind of sign up for these strategic collaborations is really that they get a larger discount. So the pricing is similar to some of the cloud providers. So it would be odd for us to believe that somebody who is spending a significant amount of money suddenly would not spend that on a go-forward basis. And both of these relationships are really strong. We'd point to AZ that even kind of double down in the interim with a separate deal as a proof point though.

Yes. That certainly makes sense. Has there ever been a large customer that has not done that? I mean I get this conversation on just, again, visibility on the data business, what do you think will happen with the existing customers? How will you grow new customers? Has there ever been a case where you haven't had someone resign a license after the initial, however, many years?

The only time -- we've quoted that we work with 19 of the top 20 big pharma companies, a couple under biotechs, and that was probably the same list that we talked about over the last couple of years. So the only kind of churn that we typically see is, if a biotech has one asset and they choose not to move forward, they may be a onetime purchaser and then move away. But most of the time, these relationships expand over time. And we're still relatively early days. We've been able to establish a business of some scale. But for the vast majority of those customers, they're still only licensing kind of 1 or 2 cohorts. So as we continue to kind of march them up over time, there's a tremendous opportunity for growth there.

Yes. It always feels like companies are working with 19 out of 20. I have to find out who this one ops in pharma company as it finds a way to not round out the group for several types of companies. Okay. So the data side feels good. You've talked about your strategy around growth that relates to data, but also just bigger picture, at least in our conversations. I'd love to just have you and maybe a public forum, talk about how you think about long-term growth versus short-term growth. Because one of the things that I'll ask you about is, for instance, a 3Q to 4Q transition and a step down that looks like for someone, it might be meaningful but for you is not because you think in longer-term time frame. So how is your revenue recognition philosophy relative to some of these shorter thought -- shorter time frame thoughts just in the sense that like the priority is, if in one quarter, you need to recognize a little bit less revenue, but it sets you up better for the long term, that's what you will end up doing. And I don't know whether I've perfectly captured that, but I know that like the way in which you think about safeguarding growth over time and keeping that consistent is not always the way that some investors think about 90-day time frames.

Yes. Yes. We want this business to grow at the 25% to 30% range for a long time. And that's always how we've thought about the business. And so we don't try to jam as much revenue in any given quarter because we don't think that, that sets you up for kind of healthy long-term growth. And so the data business is a good example of could you go out and sign more data in any given quarter? Yes, probably, but are you really establishing the value for biopharma such that they're going to come back and re-up those contracts and extend them. So we spend a lot of time making sure that customers are seeing value on the data side and really building these long-term relationships that will fuel that growth over many years versus trying to maximize any revenue in any given quarter.

Yes. Okay. So the point is there could be some quarterly fluctuation and that is not something that you feel is overly meaningful.

Yes. I think on quarterly, you're always lapping something, there could be a project that started or stopped and so there's some noise in there. And so the annual growth rate is what we're kind of focused on.

Okay. Okay. A couple of minutes left. I want to make sure I do touch on the CRO business, which is seeing some headwinds. What are the prospects for a rebound there? And then does this ARPA deal that you announced recently, does that change the trajectory at all?

Yes. So we acquired a CRO back in 2022, a small CRO that primarily worked with biotechs. At the time we did it because we were launching kind of this clinical trial matching, and we really wanted to understand or had folks that were familiar with the clinical trial process. And they certainly started hitting some headwinds last year, like a lot of the other CROs did as studies kind of dried up. It's not a huge area of focus for us, meaning like we're not investing a bunch there to try to get it to grow. But as kind of funding generally for CROs opens up, we might see some benefit there. The ARPA deal is a good example. That's a little bit interesting because some of that will go to the CRO kind of line item. There's also a sequencing direct bill sequencing component of that. So it will be a little bit split revenue in terms of where it comes through on the P&L. But it's not -- we're starting to see things stabilize, but we don't think that it's something going to return to be a high growth catalyst for us.

Yes. Okay. Maybe just moving down the P&L and thinking about investments. How do you think the investment needs will change over the next couple of years as this expansion mission that you're on evolves and presumably some of this trial work at whatever level it takes place gets going? I have you at 60% of revenues as an OpEx total for the year. That's down from 80% last year, so it's coming down, but I also have you flat as a percentage next year. Is there a justification for that? Or is the idea to slowly as you grow the business, start working that number down?

Yes. I mean I think our view is that it's still very early days across all of our kind of offerings. And so while we've been able to achieve adjusted EBITDA breakeven, we don't think it's prudent to try to maximize adjusted EBITDA in the short term. And so what we've told people is over the next, call it, 3 years, if the business grows 25% top line, we'll reinvest about 2/3 of the incremental gross profit dollars back in the business. And then in the fourth year, that probably drops to a 1/3. So we will continue to see improvement in adjusted EBITDA over that time period, obviously, as top line grows, but there's a tremendous opportunity for us here, again, across all of our offerings that we think that, that's kind of a healthy level of investment over that time frame.

Okay. How about on the gross margin line? How optimized are you on the infrastructure that you're working on -- working with when it comes to just sequencing platforms, efficiencies derived from the sequencers? Are you running the xs at this point? Do you see any gross margin benefit that can be had from just sort of production initiatives going forward?

Yes. So on the genomics side, I think we've long taken the approach that as costs come down, you reinvest some amount of that into kind of broader panels. And so that trend certainly will likely continue. We're not running anything clinically on the x today. Certainly, that will present some savings when we migrate those things. But long term, we think kind of the low 60s margins for the genomics business are kind of a healthy spot to be. So short term, as ASPs go up, you'll likely see an increase. But long term, we think kind of low 60s makes a lot of sense. On the data side, that obviously operates at a higher margin. Q3 was a little bit lower because we had some kind of start-up costs associated with the foundation model. We think that kind of rebounds into the mid-70s in Q4. So quarterly growth rates in data can be a little bit more volatile just given the starting and stopping of projects. But if you look at kind of for the year, that will operate at a mid-70s margin compared to genomics. And then the last piece I should actually add on margins is some of these AI applications operate at a very high margin. They represent a very small amount of revenue today. But if you go out long term, to the extent that those scale, which we believe that they will, that would skew margins more positively.

Yes. What is the time frame for which we should think about those scaling?

I mean, given that there's not really a reimbursement framework for a lot of those today, we haven't disclosed what we think at that time frame could be, but it's not going to be a '26 event.

Okay. All right. We're at time here. Jim, I do appreciate you come in. Good to see you.

Yes. Thanks, Dan. I appreciate.