This webcast. I'm David Westenberg, the Life Science Tools and Diagnostics Analyst at Piper. With me today is Tempus AI. I'm happy to have the CFO, Jim Rogers. Liz is in the audience as well, if any investor here has any questions here. So let's just kick it off with the quarter here. Tempus AI achieved its first positive Adjusted EBITDA in Q3. How do you envision profitability trajectory evolving in 2026, and what operating levers will be prioritized to sustain a profitability, but while still, you've got to be a growth company too, you know?
Yeah. Well, first, appreciate you having us, David, and everybody for joining. Yeah, so achieving adjusted EBITDA positive was kind of a long-term goal for us. We said by the time we were 10, we wanted to kind of achieve that, which we did in Q3, and on a go-forward basis, we don't intend on obviously taking a left turn and going back to being negative from an adjusted EBITDA standpoint, so we said in our Q3 letter, the way that we're kind of viewing the world on a go-forward basis is assuming we're growing about 25% a year top line over the next three years, that we would reinvest about 2/3 of that, 2/3 of the incremental growth profit dollars back into the business, and let 1/3 drop down to adjusted EBITDA.
After kind of that three-year period, that probably flips to a third being reinvested in the business and 2/3 dropping down. But given kind of where we're at in our life cycle of many of our products, we feel that it wouldn't be prudent to kind of maximize adjusted EBITDA in the short term, but we do want to demonstrate continued discipline and improvement in the metric.
Sounds great. Let me maybe talk about your data differentiation relative to competitors, and if you can go at this angle, not just kind of the AI component, but the software business component.
Yeah. So from very early on, and when we were building out our wet lab capabilities, we felt it was very important to kind of build these connections into the EHR systems to pull out clinical data on the patients that we were sequencing. We felt like we could provide additional value to physicians by contextualizing the results for those patients. And now we're connected to over 5,000 institutions. We have over 45 million patient records. And so a very large database. And so we de-identify that information, and we license that broadly to biopharma. We work with 19 of the top 20 big pharma companies and a couple hundred biotechs to leverage their data for drug development purposes.
These businesses are very much interconnected, but we think that not only do we want to impact patient care by providing information to physicians, leveraging this data to impact kind of drug development, improving patient care in the future is equally as important.
Got it. Maybe you could talk about that, sorry, the biopharma/software data business in a little bit more details. I think the average investor doesn't have any difficulty in understanding what you do on the CGP front. They have Guardant. They have Caris. I mean, it's Foundation has been around for years and used to be public. I do think that they probably have a little bit less understanding on the data services business and how you're differentiated. Can you talk about what you do differently than other CGP companies for pharma in a little bit more detail?
Yeah. And so those relationships are typically kind of multi-year subscriptions where they're licensing cohorts of data. R aw molecular information tied to longitudinally updating clinical information for the cohort of data. So our data business actually doesn't really compete with the other big CGP players. You can obviously see their financials and see kind of how large their data businesses are. And so we kind of set ourselves apart and have a fundamentally different business model than many of them do. And the use cases for this data are, you know, how do you improve clinical trial design and set yourself up for success? By leveraging real-world data, again, raw molecular data tied to this longitudinally updating clinical information so you know who these patients were, how they were treated, and how they responded to those treatments is immensely valuable to pharma.
Got you. Well, and that makes a lot of sense, and particularly in somatic oncology. Now, germline hereditary cancer is a little bit less established. Can you talk about the acquisition of Ambry and how you integrate the Ambry data into the overall Tempus platform for finding additional data insights?
Yeah. So the Ambry transaction, which we announced last year and closed in February of this year, you know, we did that strategically for a couple of different reasons. One is they were a vendor of ours. We had actually outsourced our hereditary screening to them. And so we knew the team quite well and were quite impressed with the product offering that they had. From a data perspective, it allows us to, one, start interacting with patients earlier on in their cancer journey, either newly diagnosed patients or at-risk patients where most of our database originally comes with metastatic or later-stage cancer patients. So getting a more longitudinal perspective of these patients from a data perspective is interesting. In terms of the insights that ultimately will be derived, we're in that process now. We closed the transaction in February.
Embedding the hereditary data with ours will be a kind of multi-year effort, so it's not a near-term driver, but one that we are excited about. The other reasons for the deal were they have an offering in rare and undiagnosed disorder that allows us to kind of move into that space from a testing perspective, but also build out interesting data sets leveraging our data connections, kind of expanding beyond oncology, both in testing and data, and then lastly, the asset had been cleaned up, so they were already EBITDA positive, so back to our conversation around kind of path to profitability, for us to do an acquisition that large, we want to make sure it didn't result in us kind of taking that left turn that we were trying to avoid.
Yeah. Got you. Yeah. And can you talk about some of the different products that come out of the germline testing and the Ambry integration? I know Eric mentioned on the earnings call a little bit more interest in rare disease. Is there any, can you talk about all the different growth contributions that could come out of maybe some of the adjacencies to what Ambry does?
Yeah. We think the hereditary screening market is going to have a period of growth. And Eric has mentioned this previously as well, if people kind of viewed this market as being very mature and saturated. And while it's probably more mature in areas like oncology, it's less mature in other disease areas. And for the Tempus platform, it can really be applicable to any disease area where there's multiple kind of therapeutic options and a diagnostic test that can be useful. So where genetic sequencing is important, which we believe it will become more and more important in other disease areas. So they have other offerings in cardiology as well, meaning Ambry does. And so we'll look to continue to expand those as well.
Got you. Maybe we could talk about the AstraZeneca and then Pathos AI deal. This is a multi-year deal with expected revenue recognition scheduling over a multi-year period. Can you talk about both the coming together of that deal and how you expect to recognize revenue over the next few years?
Yeah. So AstraZeneca has been kind of one of our longest-standing customers on the data upfront. So they signed a $320 million deal back in 2021 and then also signed this $200 million deal with Pathos involved this year. And this is kind of the culmination of many of our data collection efforts of, obviously, AstraZeneca has seen tremendous value in leveraging the data in their portfolio. And so we started having conversations around, what if we looked at all of the data? They're still licensing a relatively small subset of our overall database. And could we train a model on the 400 petabytes of data that we've collected over the last 10 years? Pathos was a company that we obviously have a relationship with, but as does AstraZeneca as well. And so it's a $200 million kind of data license to build this model.
Ultimately, at the end of it, we all get a copy of the model for use in our respective areas. So AZ and Pathos can use it for drug discovery. We can embed the insights to make our genomic tests smarter and also improve our data offerings to biopharma. In terms of revenue recognition, the $200 million is largely ratable over the next three years.
All right. Okay. So can you talk about the commercial strategy for Tempus Pixel and the ECG algorithm? What are the projected revenue contributions and margin profile for 2026 from this offering?
Yeah, so Pixel and some of the ECG offerings, we have a number of these kind of algorithmic diagnostics that we've created over the years. As we've talked about, there's not really a reimbursement mechanism for many of them. There are some early signs, the cardiology algorithms being one of them, where there are reimbursements in select areas, but it's still very early days, and so when we think about kind of the guide and the contribution for next year, we don't have anything baked in associated with those businesses. Long term, though, we don't see a reason why these types of technologies would not be reimbursed. They have clinical utility. They're very useful for doctors, and in many times, they're cheaper than the alternatives in terms of either the alternative testing or the adverse events that come without identifying these patients earlier on.
So we don't envision a world where, as you zoom out, whatever, seven, 10 years that these things would not be reimbursed. We're fortunate that we have a very strong core business in genomics and data that have allowed us to achieve break-even from an adjusted EBITDA standpoint while still making very large investments in these spaces. So while we're doing all of the same things that others are in terms of convincing people that these things should be paid for, we're fortunate that our business model doesn't hinge on them being paid for in 2026. But as we zoom out kind of longer term, we think this will be a significant portion of the business.
Got you. Can you talk about the launch of xT CDx? You got ADLT status, sorry, $4,500 reimbursement. What's been the market uptake since this launch? What's the strategy to drive further adoption in 2026 and beyond?
Yeah. So xT is our solid tumor DNA test. We got FDA approval and ADLT status. And then we kind of nationally launched the test earlier this year. By the end of Q3, we had about 30% of our solid tumor DNA testing volume had been migrated over to that version. And again, for context, that's a $4,500 price point with Medicare versus the roughly $3,000 we get for the non-ADLT version. And we have plans in 2026 to get the vast, vast majority of that remaining 70% migrated over. So as we exit 2026, we would envision, again, most of it being on the ADLT version.
Got it. It's interesting with the reimbursement rates across the industry. I mean, we're seeing CGP tests in the high thousands and MRD tests being frequently reimbursed. So I think there's some a little bit concern on investors about when exactly CMS could come in and maybe cut prices. Can you just remind investors about the value provided by these tests in oncology and how they can necessarily save the system money and better outcomes for patients, etc.?
Yeah. I mean, within therapy selection, obviously, many of these cancer treatments can run in the hundreds of thousands of dollars, and so several thousand dollars for ensuring that we're getting the patient on the right treatment at the right time makes a ton of sense financially, and then with respect to monitoring and MRD, obviously, the sooner that you can identify recurrence and get somebody on treatment relative to the adverse events of not identifying that early on, we think that that makes a ton of sense. In terms of kind of overall reimbursement, our average reimbursement is about $1,600. Obviously, David, you know our competitors are significantly higher than that, and so we don't envision a world where the long-term reimbursement trends aren't positive to where we sit here today.
Yeah. Fair point, so one of the interesting things is you're not just oncology and you have broader ambitions here. So can you talk about some of the newer disease categories in which you could use your data advantage and what the strategy is for expanding into these areas?
Yeah, so again, it's kind of disease area by disease area, and we've long said we don't have to run every diagnostic test. So in cardiology, for example, we obviously are looking at the output of ECGs. In radiology, we're looking at the scans. In other areas such as rare, we may actually run the diagnostics. So we look at it disease by disease area. We also are very thoughtful in the investments that we make in new disease areas, so these aren't $100 million investments. We start these very small to prove out that, one, there's a need that we can provide insights to physicians, but then also that there's a data business that can further support kind of additional investment in the area, so we will continue to expand in other disease areas, but again, it's all with that kind of profitability framework that I discussed earlier.
Got you. And then what are the key milestones in 2026 for new product launches? What are some of the enhancements that you're working on, both genomics and in the data segment?
Yeah. So in genomics, obviously, we said in the Q3 earnings release while we're submitting xF, our liquid biopsy test to the FDA here shortly, we're awaiting reimbursement for xM, our tumor naive MRD test. And then there's always iterations of assays that occur during the year and ramping rare within Ambry is another kind of area of focus for us. On the data side, we've said that the first iteration of the foundation model that we're building would take about a year. So that obviously will come during the year. And then just continuing to build out those capabilities and advance those partnerships with biopharma because, again, it's still very early innings on that side of the house.
Got it. We do have to talk about the things that you ask a CFO. So we're going to talk about margin progression, GM, whether or not you have still a NovaSeq X upgrade that you could be doing, whether or not just volume going through the system is going to help you with some of the margin progression. I mean, anything you can give us in terms of reductions in cost of goods sold that we should expect over the next couple of years?
Yeah. So within genomics, excluding Ambry, margin expansion comes from increases in ASP kind of going down that ADLT path that we talked about. We still run the 6,000 from Illumina, so we're not on the X at this stage. Obviously, as we launch new versions of the assay, we'll look to migrate over to that. We do have a fleet of those in-house today. We've long taken the approach of, as costs come down, kind of reinvesting those back into larger panels, eventually getting the whole genome. And so unlike a traditional diagnostic company that is hyper-focused on what the genomics margin is, while we should see some expansion, we're okay with it being probably less so than what you would otherwise see if you were only focused on that if it's generating more data that leads to improvements on the data or continued growth on the data side.
Data margins are pretty stable. You'll have given quarters where if a new project starts, margins may dip or vice versa. They've been in the kind of mid-70s range, and we don't see any kind of shift there other than if the AI applications business takes off, as we think that it will at some point, that operates at a very high margin. So expansion there would be kind of longer term as that AI applications business grows.
Yeah. This brings up kind of an interesting point because we hear from investors talking about the justification of price, but you mentioned kind of here, well, the cost of goods sold savings just goes into expanding the assay. So I mean, can you just maybe think about that in terms of how ASPs could progress and actually be stable in the face of continuous development? I kind of answered that myself, but.
Yeah. I mean, I think that the ASPs will land where they are. And obviously, you adapt the assays depending on that to maintain a healthy margin. So our overall margin, we think, is in a healthy spot today. And we continue to evaluate all kinds of the moving parts to determine what that long-term margin profile would look like.
Got you. What do you talk about some of the expenses on the OpEx side, priorities for you? Is it where are you going to work at, s ales force expansion, R&D? I think you're probably fairly decent on the G&A front, but maybe you have some sort of international expansion or something that you would consider?
Yeah. Most of the investments are going to come on. So we have kind of two R&D lines in our P&L: traditional R&D, which would be kind of your wet lab R&D akin to most of our diagnostic peers. And then we have a technology line, which is all of the R&D that kind of goes into the tech platform. Those are the two biggest areas of investment for us. The sales force, it grows as we continue to kind of bring in more volume and you split territories. But there's not a need to kind of have a massive increase in the overall sales force. Almost all the investment is going into those two R&D lines.
Got it. Can you talk about where you see the business in a few years from now in terms of just because you brought up the long-term software margins are quite attractive in this kind of business? Can you talk about the revenue mix you could see in the future between software and, say, wet lab and dry lab?
Yeah. I mean, I think you have periods where one way grows quicker because you get ASP tailwinds or whatever it is, and that leads to outsized growth maybe in genomics o r you may sign a big data deal, and that results in greater growth within the data segment in every given period. We think both genomics and data can be very large businesses. AI applications is the one that is kind of the unknown, right? Because that may grow, it could go from zero to something very large overnight if reimbursement kicked in for some of these algorithms. But again, as we think about guides and all that and how we think about the world, we don't include these things until we have a clear line of sight of them landing. So we're doing all the work in the background to have them land.
But you could envision a world where if you zoom out 10, 15 years, that the AI business is the largest of the three of them. But again, until we have a clear line of sight of that, we don't base that in.
Got you. Can you talk about the relationship that you have with Personalis and where you see the mix, the market is going between tumor-informed and tumor naive?
Yeah. So we struck an agreement with Personalis last year to be the exclusive distributor of their tumor-informed assay in breast, lung, and IO. They just received coverage in breast a few weeks ago. And so we're starting to kind of scale the commercialization of that effort. The MRD market, we think, is going to play out pretty similarly to how therapy selection played out. You're going to have some doctors that really prefer the sensitivity and specificity of a test like Personalis, and you're going to have other doctors that really prefer the operational ease of a tumor naive assay. And at some point, the performance of these tests are going to kind of reach some level of parity, no different than they did in therapy selection. So we felt it was important to have both offerings.
Our view has long been to be kind of a one-stop shop and offer everything to physicians because we do think that physicians don't want to be ordering from three or four different providers. They would prefer to have one that's integrated, that is very easy to use, that provides additional insights, and again, we don't have to always run that diagnostic. We can embed that into our platform, so that's what we did with Personalis, and so the relationship has been growing great. We're excited that they finally received coverage in breast and excited about what that leads to in 2026.
Got you. Well, you do have the sales force and the strategic reach that a lot of companies don't have. Do you see yourself having more of these kind of relationships like you do with Personalis to kind of expand the overall offering? And do you think the market is migrating to a kind of a one-stop shop in different oncology tests?
Yeah. I mean, ultimately, we think the MRD market will be a natural extension of the therapy selection market. It would be very odd for us to believe that somebody is going to want to use, again, multiple kind of providers, and so we think over time that we'll have a sizable kind of MRD offering. In terms of other tests, we're constantly evaluating what we think the doctors need by listening to them and having them tell us what they need and see if we can find things that can augment the portfolio. We're also cautious, though, to not overload our sales force with too many different things to disrupt kind of the core offering, so it's that balance of making sure that we're meeting doctors' needs but maintaining that customer experience that they're used to.
Yeah. But this brings up to the point of the apps. Can you talk about some of the different applications that you're seeing in your therapy selection business and how that's been contributing to growth and essentially how you can balance talking about just therapy selection, but hey, you know what, knowing tumor of unknown primary. I don't know. I think you guys have that one, but I'm not sure.
Yeah. I mean, ultimately, the same data that our biopharma customers leverage, we leverage internally to see can we identify insights, again, that can make our tests smarter. Some of these get turned into what we call algos, which are basically add-ons to a genomic sequencing test that somebody can order, and oftentimes, they're built on some combination of molecular and clinical data. So we have a handful of these in market today. You mentioned tumor of unknown primary. We have another one called IPS, which is immunoprofiling score. Again, doctors are overwhelmed with the amount of information that they're being provided. They're obviously seeing a bunch of patients, and so if we have access to this information, to the extent that we can provide insight by having these available to doctors, then we make them available, and the uptake has been great.
It's not an insignificant amount of orders that are adding these things on because doctors obviously see value in what they're providing, and we'll continue to bring more and more of these to market.
Got it. Lastly, in terms of capital deployment strategy, how are you thinking about either acquisitions or deploying more of the capital back into the business? And how far are we away from an idea like share repurchases? I'm guessing we're still three to five years away from something like that, but anyway, longer term.
Yeah, so I think on the M&A front, we've obviously been fairly acquisitive over the last couple of years, Ambry being kind of outsized relative to some of the other deals that we've done. We did Paige in the digital pathology space, Deep 6, which has kind of a connectivity platform. There's a lot of very interesting companies out there, especially in the data and AI front. We feel like the genomic portfolio is pretty well-rounded out, but there's a lot of companies that created really interesting technologies but kind of couldn't get to commercialization, and to the extent that it makes sense, again, within kind of our profitability framework to bring those in, then we may do so, but that's more of a build versus buy discussion than anything else.
Okay. Thank you very much for your time.
Thank you. Thank you, everybody.
Sorry. We plowed through the questions, BIll so I had to make up half.