Tempus AI, Inc. (TEM) Q1 2025 Earnings Report, Transcript and Summary

Good day everyone. And thank you for standing by. My name is Argie, and I will be your conference operator today. At this time, I would like to welcome everyone to the First Quarter 2025 Financial Results Conference Call. [Operator Instructions] Thank you. I would now like to turn the call over to Liz Krutoholow, VP for Investor Relations. Please go ahead.

Thank you. Good afternoon. And welcome to Tempus’ first quarter 2025 conference call. This afternoon, Tempus released results for the quarter ended March 31, 2025. The press release and overview of the quarter and our latest presentation are available on our IR website. Joining me today from Tempus are Eric Lefkofsky, Founder and CEO of Tempus; and Jim Rogers, CFO. Before we begin, I would like to remind you that during this call, management may make forward-looking statements that are subject to risks and uncertainties that could cause actual results to differ materially. For a discussion of these risks, please refer to our 10-K and other filings with the SEC. During the call, we will discuss non-GAAP financial measures, which are not prepared in accordance with generally accepted accounting principles. Definitions of these non-GAAP financial measures, along with reconciliations to the most directly comparable GAAP financial measures are included in our first quarter earnings release, which has been furnished to the SEC and is available on our website at investor.tempus.com. I would now like to turn the call over to Eric.

Thank you. And thanks for joining us today. Q1 was a record quarter for Tempus and we are off to a great start. I’ll provide just a super quick overview and then we can take questions. Quarterly revenue increased 75.4% year-over-year to $255.7 million. Genomics revenue was $193.8 million which is about 89% year-over-year growth. Oncology testing, which is how we’re going to refer to our legacy Tempus clinical testing grew 31% year-over-year with approximately 20% volume growth. Hereditary testing, which is how we’re going to refer to the legacy Ambry Genetics business, contributed $63.5 million in revenue and grew its units by 23%. Revenue from Data and Services totaled $61.9 million, which was about 43% year-over-year growth, led by our Insights or data licensing business, which grew 58% year-over-year. We generated $155.2 million in quarterly gross profit, which was 99.8% growth year-over-year. Adjusted EBITDA was negative $16.2 million in the first quarter of 2025 compared to negative $43.9 million in the first quarter of 2024, which was an improvement of $27.8 million year-over-year. As a result, we’re increasing our full year 2025 revenue guidance to $1.25 billion, representing about 80% year-over-year growth. So, I would say all in the company is performing super well, which was in my quote revenues are up, gross profit is up, both are growing nicely, we’re managing our costs, which is producing nice year-over-year operating leverage. In addition, I’ll highlight just one other big piece of news which we put out about a week ago, which is we announced a three-year, $200 million data and modeling license agreement with AstraZeneca and Pathos in April to build the world’s largest foundation model in oncology. This is big for a few reasons. One is it brings our total remaining contract value to greater than $1 billion as of April 30. It also allows us to take over 300 petabytes of data, which includes this really rich multimodal data set connected to outcomes, and use that to build a foundation model which is in addition to the data licensing, which is quite positive for us, also, the cost of compute is not small and AZ and Pathos are covering a significant portion of that. When the model is complete, which we expect, the first version of the model will be complete in about nine to twelve months, each party will get a copy. AZ and Pathos to advance their drug discovery efforts, and Tempus to advance its diagnostic and data products. Given that AstraZeneca is our longest standing client, actually was our first strategic collaboration, we couldn’t be more excited to be expanding our relationship in such a significant manner, I think, kind of further validating the value we’re providing to lots of biopharma clients. It’s also worth noting this is a non-exclusive agreement. We can essentially license data and build models with others and we hope to do so in the future. And as such this represents an entirely new category for us. It’s also important in that it’s a giant step in making precision medicine a reality. We’re closer than ever to understanding at a molecular level why patients do and don’t respond to cancer treatments. And we believe models like this will bring all kinds of insights into clear focus and we can see a day when our diagnostics are so smart that they’re actually playing a critical role in ensuring every patient is on the optimal therapeutic path and that drug companies are far more efficient, ideally in a perfect world, having clinical trials that fail far less often. On that note, we’re happy to take questions.

Thank you. [Operator Instructions] Your first question comes from the line of Tejas Savant from Morgan Stanley. Please go ahead.

Hey guys, good evening and appreciate the time here. Eric, congrats on a clean start to the year. I want to ask a two-parter on the AZ-Pathos deal that you just highlighted as well. So first up, can you just share some color on follow-up conversations with other pharma companies about the possibility of similar deals for foundational model development and oncology? And what is it about the Pathos approach that really is the hook here for drug developers? And on the deal again, one for Jim, I want to dig a little bit into the deal structure here. It’s a little bit sort of complicated. So, as we think about the rev rec on this $200 million amount, there is that $50 million upfront fee from Pathos to you guys, but then you guys are also paying them, I think, $35 million. And then in terms of upfronts, you also get an AZ, I think, paying $35 million. So, just pass that out for us in terms of how you expect it to play out on the P&L. And then the $150 million residual from Pathos, I think, there’s a stock component in it as well. So, just unpack that a little bit in terms of the next three years and how it flows through the P&L. Thank you.

Yes, so I can start and then Jim can jump in. So after we announced this deal, obviously there was quite a bit of excitement among other companies. We work with, I think, 19 of the 20 largest pharmaceutical companies in oncology and have good relationships with a bunch. And so people were quite interested. AstraZeneca is one of the leaders in oncology and has had a really strong track record over the past four, five years and so I think people were particularly focused on what was this going to mean for them going forward and how should they try to bring it into their own practice. Some of those conversations have already kicked off. I would say the excitement has been greater than I thought it was going to be. And I had pretty lofty expectations, so that’s awesome. But these are big deals, as Jim will cover in a second, this is $200 million of data licensing and real data revenue. And so somebody is got to be willing to sign up for something that significant. And it’s expensive. And so even though there is a ton of excitement, we have to turn that into tangible agreements, and tangible projects and kick those off. In the case of AZ and Pathos, AstraZeneca was a client of ours. They also had spent some time with Pathos and got to know that team. I think there was – independent of the Tempus relationship, they were exploring some different ideas together. And so I think when we began discussing this idea of building a foundation model, it made sense for them to want to have this be a three-way agreement whereby they could make a sizable investment and commit some of the attributes they have, but they also could leverage a bunch of work that Pathos had done, and then obviously leverage our data and the work we had done. And so it came together as a three-way partnership, but just as easily could have come together as a two-way partnership between us, and the pharmaceutical company and not involve Pathos. And I’ll let Jim cover the rev rec.

Yes, in terms of the revenue recognition, I think, the easiest way to think about it is we have a $200 million kind of data license to Pathos that is specifically related to building the foundational models. The data can only be used for that purpose as AZ-Pathos and Tempus work together to build that foundational model. There are some cash flows between AZ to Tempus over to Pathos, but there is no revenue recognition impact of that. And so the $200 million will be recognized, it ramps a little bit over the three-year period, but roughly ratably over the three-year term during which the model will be built. So, this license is no different than the other large subscriptions that we have. The upfront payment from Pathos to Tempus doesn’t trigger any revenue recognition upon that payment. It’s a three-year subscription, similar to the other large multi-year deals that we currently have.

Yes. And one last piece, the first payment was made in cash. We at this point, even though Pathos could make some payments in the future in part cash, part stock, we have every reason to think they will make it in cash. And so it’s very possible that we just collect cash the entire time.

Your next question comes from the line of Ryan MacDonald of Needham & Company. Please go ahead.

Hi. Thanks for taking my questions and congrats on a great quarter. Maybe you could talk about the hereditary business. I think the original expectation as that was integrated – as Ambry was integrated, was sort of a maybe a mid to high teens growth rate for this year, but obviously kicking off the year at much faster rate and around 23%. Can you just maybe talk about what surprised you to the upside in terms of the performance of that business and maybe how durable the sort of mid-20s growth rate was? Thank you.

Yes, I mean, obviously it’s early, so we don’t – there is no point getting too ahead of our skis. But we talked about this, I think, when we announced the Ambry acquisition, which was there was this kind of narrative that hereditary screening was either kind of in the twilight or sunset of its horizon or had become commoditized. And we just obviously couldn’t feel more strongly that that’s not accurate. I mean, I could foresee a day when tens of millions of people get this kind of sequencing on a regular basis, not just to understand their inherited cancer risk, but their inherited cardio risk, their inherited Alzheimer’s risk, their inherited risk of developing immunological disorder later in life. And so the target audience of people that might be at risk of disease is obviously much greater than the audience of people that have disease, generally speaking. So I think long term, we suspect Ambry will grow at high rates. In the near term we told people we thought the growth rate would be mid to high teens in large part because they experienced a lot of rapid growth previously. So you’re lapping that period. So, yes, the business is performing really strong. That may continue. But we’re not here to kind of highlight that for folks. We’re watching it month to month, but so far, they’re firing on all cylinders, and we see no sign of that slowing down.

Question comes from the line of Mark Schappel of Loop Capital. Please go ahead.

Good evening. Thank you for taking my question and nice job on the quarter. Eric, I have a question around the Deep 6 acquisition that was made during the quarter. I was wondering if you could just provide some additional details around what capabilities Deep 6 brings that you already did not have.

Yes. So we had – we also talked about this, I think, last quarter, which is that we felt like we had a really comprehensive molecular offering that with the acquisition of Ambry, we felt really good that between our MRD offerings and our therapy selection offerings and now our hereditary offerings, we felt really good and that if we were to make any acquisitions, they would likely be smaller and likely be on the data and services side. And so that – by the way, it’s exactly what Deep 6 is. Core to our business model is obviously combining large amounts of clinical and molecular data and being able to build these really rich data sets where you can understand at a molecular level what’s happening to a patient and then connect all those rich molecular insights to outcome and response data over time and say like, okay, if this is the molecular composition of this patient, this is the DNA profile, the RNA profile, the germline profile, the methylomic profile, this is the molecular profile of the patient, what drugs do they take? How do they respond? What adverse events do they have? How long are they on that drug? What was their progression-free survival? Did they – what was the overall survival, and all that clinical data. And so you have to have rich connections to pull that clinical data. We also mentioned this quarter, we’re now over 4,000 connections, which is up significantly from the past. And Deep 6 is a part of adding more connectivity, especially to some really high-quality institutions. They built a product that allows providers to interrogate their own data sets to advance analytics to get people on their own studies and clinical trials. And that product has good product market fit. People like it. And it allows us to kind of have another connection point to providers, another reason for them to share their data with us, another reason to be on our platform and that bidirectional feed of data. People sending us their clinical data, we’re generating some molecular insight, we’re putting the insight back into the hands of providers. That is at the core of how we build these very large data sets, which are now, in totality, like 40 million patients or something. So it’s become a huge data set.

Your next question comes from the line of Subbu Nambi of Guggenheim. Please go ahead.

Hey guys. Thank you for taking my question. I had one model-related question and one long-term. For the model-related, could you remind us the assumption baked in on gross margin and ASP improvement if you do receive MolDx xM reimbursement before the end of the year? And if it’s not baked in, could you quantify the upside? And the long-term question was when you do flip EBITDA positive, you said you are looking to invest back into the business. Curious, what are you looking at right now with the most ROI in 2025 and if that wish list changes at all as a result of flipping to profitability? Thank you.

Yes. So I’ll take the first question and Eric can take the second one. In terms of the ASP progression, obviously, we saw about a $60 increase in our oncology ASPs in Q1, largely the results of us migrating our xT volume over the ADLT version of the FDA-approved version of that assay, which has a higher ASP with Medicare. There was also a small impact from some ASP improvements for our liquid biopsy code as well. In terms of the progression over the balance of the year, we said that by the end of Q1, about 20% of our xT volume would be migrated to the FDA-approved version. That’s where the number kind of came in at. We will continue to migrate more over the balance of the year, targeting about 40% by the end of the year. And so the ASP improvements for the balance of the year largely come from just migrating xT volume from ADLT to the FDA-approved version. We haven’t baked in any xM reimbursement. We submitted our first CRC xM to MolDx in January. That process is playing out, and we wouldn’t anticipate anything until later on in the year. And so we haven’t baked any of that into the ASPs.

And in terms of kind of how we think about EBITDA and investments and ROI, it was very important to us, as we have said historically, that we are EBITDA positive by the time we turn 10, which is this year. We’ll turn 10 this year. And so I think we’re on track as providing our guidance to the adjusted EBITDA positive this year, which is a big milestone for us, especially given that there’s other companies that are older than us, similarly situated that are still losing $100 million a year, $200 million a year. So we feel really good that we have these rapid growth rates and kind of often best-in-class growth and the business producing lots of gross profit and lots of leverage. And so we’re able to run it in an EBITDA-positive manner – adjusted EBITDA positive manner. The issue for us, I think, is when you look at the opportunity set to bring AI to health care at scale, it’s not small. So you – and you don’t want to under-invest and just try to maximize every dollar of profit at the expense of long-term sustained growth and then miss out on what could be one of the biggest technology opportunities of all time. And so we’re mindful of that. And in particular, there’s lots of places to invest in both of our main businesses. For example, in genomics, the MRD space is super exciting. We’ve got a tumor-naive platform that we believe in. We’re running all kinds of studies in different disease areas, and we continue to make those – we’re making those investments now. We made them last year. We’ll make more in the future. And we feel good about those investments, but it’s certainly an area to put money to. And then the other is in terms of building out our core AI applications and product set, including the foundation model that power a lot of this. You’re constantly investing in data and compute, and those are not inexpensive. And we make lots of investments. Again, we made them last year. We make them this year. We’ll make them next year, make lots of investments in data and compute to be able to bring AI to diagnostics at scale. And so we have – we’re fortunate that the landscape in front of us is open. We can make – we have lots of things we would love to invest in. We’re also disciplined in that we’re not going to trying to get ahead of our skis and make sure that we’re investing appropriately. And I think lucky that these things are all coming together in a really nice way, where the growth is producing lots of additional dollars that we can invest to drive future growth.

Your next question comes from the line of Daniel Brennan of TD Cowen. Please go ahead.

Great. I was hoping maybe you could just speak to – the first question would just be on the Genomics volumes. Just give us a sense, I know some competitors have talked about some weather-induced issues in the first quarter. 20% was kind of towards the lower end of the range. So just wondering kind of how the quarter played out versus expectations. How do you think volumes will kind of play out the rest of – throughout the rest of the year? And then, b, just on the Insights business, super helpful upfront with the Astra deal. But I’m wondering, what else could you say just in terms of how that business is going, what the funnel looks like? Any qualitative color you can provide about the demand trends on your Insights business?

Yes. I’ll take the first. I don’t know about weather. I mean weather could have played some impact. We certainly had many days when FedEx was delayed or parts of the country were shut down. We don’t spend a ton of time focused on that because we’re not – we don’t think of ourselves as just a lab, and that’s not our – that’s not like our only business where we’re kind of micro fixated on it. We consider 20% growth, given our scale, given our volume, given the volume of tests that we’re running, to be pretty extraordinary. When you look at the unit growth of Tempus in the aggregate, I think we delivered something like 158,000 tests this quarter. So any time you’ve got something delivering 158,000 billable orders that is growing units not only double digits but in the 20-plus percent range, it’s pretty good. So we’re way more focused on long-term sustained growth than we are on short-term growth. And if we ever have a choice to grow at 20% for six or eight quarters or 10 quarters versus 22% or 20% for one quarter, we’ll always choose the former.

Yes. And then on your second question, Dan, regarding the Insights business and the Data Services business, it’s also off to a good start. As we mentioned, the whole Data and Services business growing more than 40%, the Insight business growing 58%. So we were very fortunate coming into the year. We had $940 million of real contract value that was yet to be delivered. And so delivering on those subscriptions and then adding additional deals in Q1, obviously, the highlight coming in, in April with AZ, which pushes the total remaining contract value over $1 billion for the first time ever. And so that forward-looking visibility that those contracts provide allow us to feel really confident about the data number for the balance of the year and into the next several years.

Your next question comes from the line of Mark Massaro of BTIG. Please go ahead.

Hey, guys. Thank you for taking the questions. The first one is for you, Eric. I was just curious if you could speak to how you think you can leverage your advantage with data and collaborations with pharma to some of your early traction in the MRD space. So if you could speak to how you’re – how you see the tumor-naive opportunity but also how you see the tumor-informed opportunity coming together with your partner. And then I’ll ask the second question, which is, can you speak to any puts and takes on any tests in your broader portfolio that might be picking up share in the marketplace? Thank you.

Yes. So let's start with the first. So I think, look, long-term – we have said this for years. We believe that AI and technology are the primary differentiator of diagnostics, which is kind of complicated because we spent so much time on the diagnostic side, talking about like sensitivity and specificity and limited detection and this study and that test and these 500 genes and 1,000 genes. We fundamentally believe that is not the differentiator. We're in a migration that we've been in for some long period of time where sequencing is getting less expensive, and I suspect, over time, we're all doing whole genome, whole transcriptome and the like quite regularly, and that's the bioinformatics landscape of the future. So what differentiates these tests is what insights can you drive for a clinician or a patient off of this massive amount of data. And that is where I think Tempus is so differentiated. And yet, we don't – kind of don't spend a lot of time ever talking about it, which is fine because it's – you want to see when it shows up. But if you look at the foundation model we're building, what it's essentially doing is pouring in an enormous amount of data, right, like hundreds of petabytes of data, looking for associations between vast amounts of molecular data that we have been unable to ever interrogate connected to vast amounts of outcomes. And likely, what should show up is all kinds of associations that none of us ever knew existed. So for example, if I'm a non-small cell lung cancer patient, one of the first things the NCC guideline would tell me is I should be profiled and to see if I'm eGFR positive. If I'm eGFR mutated, I should get an eGFR inhibitor. The challenge is about half the patients that get that drug respond and the other half don't. And even the half that do respond have different variations of response. Some might be on that drug for a year. Some might be on that drug for a decade. We have no way of stratifying those patients. And in cancer, the name of the game is to generate more insights earlier in the process. And I suspect AI and technology will produce that. We – I hope it's Tempus, but some company like Tempus will one day understand whether or not a patient is going to respond to an eGFR inhibitor before they ever go on the drug. And once you have that kind of information, you can design truly intelligent personalized diagnostics. That's going to hold true for therapy selection, that's going to hold true for hereditary, and that's going to hold true for MRD. We are going to know with far greater granularity not just whether or not a patient is likely to recur, which these tests are amazing in their ability to see recurrence long before a scan, but also how to intervene what it means when we're – when we see these signatures, these – whether they're methylomic signatures or whether they're just fragments of mutations in the blood. What does it mean? How do we analyze them? Is this patient going to – is this patient likely going to have a very bad recurrence? Is it going to be mild? Is it going – do we have a short amount of time, long amount of time? How aggressive should we be? All these insights will be data driven, I think. And so I would suspect that for both our tumor-naive and tumor-informed products, we enhance them and eventually, I think, make them totally differentiated from anything else out there by virtue of our investments in AI technology and the data that we've been able to amass, which others just don't have.

Then in terms of the performance of the different assays that we have in market today, I'd say we saw growth across the entire portfolio. Obviously, with MRD, we're still kind of metering the volume given the lack of reimbursement. And so we wouldn't anticipate volumes growing tremendously there only because since we don't have reimbursement, we're bearing the cost of running those tests. But the core assay is kind of all performing well in the quarter.

[Operator Instructions] And your next question comes from the line of Michael Ryskin of Bank of America. Please go ahead.

Great. Thanks for taking the question guys. I'll ask one. Maybe it has multiple parts, but I promise it's only one question. I just want to make sure sort of tying your earlier comments on Pathos on sort of how Ambry performed in the quarter. Just sort of what's included in the or what are your assumptions that add up to the new revenue guide for the year, the $1.25 billion? I mean you're raising it by $10 million versus prior. Does that include Pathos contribution? It sounds like it does. But if you just take that $200 million and prorate it over 12 quarters, just sort of how much of that is in there? And then previously, you talked about Ambry high-teens growth, the legacy Tempus business around 30%. Is that still unchanged? I just want to parse out the moving pieces of the guide change? Thanks.

Yes. I mean Jim and I can both answer. So certainly, in the guide there is some amount of the new Pathos, AZ revenue. We start every year. We have a very high degree of visibility to our revenue, especially our data revenues, but not 100% visibility. So we always expect to sign a certain amount of revenue that we both sign and deliver in the year, and that's been the case for some period of time. So we're fortunate that here we are at, whatever, May and we've – we're so far ahead, which is awesome. But this thing is going to ramp over time. We don't have – obviously, we only have a partial year anyway. We had some – we always had planned to go get some additional revenue. So we feel like this is appropriate. We also are being, I think, appropriately conservative in terms of Ambry and its growth rates and how they perform the balance of the year. Could there be some upside there? Sure. But at this point, our job is to basically kind of say, "Hey, this feels like an appropriate place to be," and we're glad that – by the way, we raised guidance last quarter by $10 million. We just raised it again by $10 million. So we're kind of $20 million above where we were, I don't know, 90 days ago. So we feel like we're in the right place, but we don't want to get ahead of our skis.

Next question comes from the line of Rachel Vatnsdal of JPMorgan. Please go ahead.

Hey. Good afternoon. Thank you guys for taking the question. So I wanted to dig into the data side of the portfolio and specifically what you're seeing on the TCV. So just given what we've seen from a macro sense, there's a lot of noise out there in regards to biotech funding but also pharma, these potential tariffs and everything as well. So can you talk about the risks that you see given the choppier macro environment that you could see some of the TCV either canceled or pulled out of that backlog? We've also heard some of the CROs kind of talk about these elevated cancellations in both preclinical and clinical studies. So curious, how are you assessing that risk? And have you seen any impact so far?

Yes. So just a reminder on the data business, we kind of have two customer groups. We work with 19 of the top 20 large pharma companies and then a couple of hundred biotechs. Obviously, the majority of the TCV and the revenue comes from the large pharma companies that have larger R&D budgets, but we do work with a number of biotechs. Certainly, on the biotech side, there has been some impact over the last, call it, 24 months of the lack of funding and kind of that coming through. But again, that represents a smaller percentage of our overall business. The relationships with big pharma tend to be multiyear subscriptions that are committed. And so we're delivering the data that they have agreed to license over these terms, and so we haven't seen a significant impact on the large pharma side. I'd also add that when budgets tended to be slashed or cut, we actually see a little bit of a benefit to the data business because leveraging the types of data that we license folks allows them to more effectively design their trials or identify targets in a more effective way. And so we actually see a little bit of a benefit of our business on the data side when it comes to budgets being shrunk because they can leverage our data and be more efficient and kind of stretch those dollars, and that's what we've seen in the market.

Your next question comes from the line of Dan Arias of Stifel. Please go ahead.

Hi guys. These are the question. Eric, on MRD, obviously, a lot going on in that space. You guys are coming up on one full year, I believe, of commercial availability for the xM assay. What's gone the way that you expected within your program? What's been a bit of a surprise? And then what would you say are the key things over the course of the next year in order to feel like you're on track and successful?

Yes. So I mean we – our portfolio, just to remind people, includes our tumor-naive assay in colorectal, and it includes Personalis' tumor-informed assay in non-small cell lung, breast and IO response. So we kind of run the market in four areas, four very big areas. We are managing volumes and metering volumes because none of these assays are currently reimbursed by MolDx. So not our assay, not their assay. And so we run these tests but we don't get paid. So we're metering volume. Otherwise, you could burn a lot of money. I would say the demand has been quite strong. We've been pleasantly surprised that there's a ton of interest in both products. I think there's a space for naive and a space for informed. Informed is clearly winning the day today because it has improved sensitivity and specificity. It's got lower limits of detection. Natera has done a great job setting the market. So informed is kind of the more conventional way people think about MRD today. We're fortunate that Personalis' platform is really best-in-class in terms of a bunch of those metrics because it's whole genome based instead of whole exome based. So I think people want it, and I would suspect the volumes will be really strong once we ungate it. In terms of tumor-naive, we're all doing a bunch of work, us and other people that have those products, to keep improving those assays, make sure that they perform well. I think they performed quite well today. Ours certainly does, performs quite well, especially in those instances where you don't have extra tissue, where you can't rerun a whole tumor-informed assay because you just literally don't have the tissue. And in certain areas like non-small cell lung cancer, where you have scant tissue to begin with, I think we – those products will perform really well. So I would say it's early days, but everything is moving along quite nicely, and I have not been negatively surprised in any big way. It's always kind of when you start a new product, you learn all kinds of stuff. And so far, I think we feel really good about our long-term position in MRD.

That ends our Q&A session, and we appreciate your participation. I will now turn the call back over to Liz Krutoholow, VP for Investor Relations. Please go ahead.

Thank you all for joining us today. As always, we're available for any follow-up questions. We look forward to updating you again next quarter.

Ladies and gentlemen, that concludes today's call. Thank you all for joining. You may now disconnect.