Tempus AI, Inc. Q4 FY2025 Earnings Call

Good day, and thank you for standing by. My name is Tina, and I will be your conference operator today. At this time, I would like to welcome everyone to the Tempus AI Fourth Quarter 2025 Fiscal Results Conference Call. It is now my pleasure to turn today's call over to Liz Krutoholow, Vice President of Investor Relations. Please go ahead.

Thank you, Tina. Good afternoon, and welcome to Tempus' Fourth Quarter 2025 and Full Year 2025 Conference Call. This afternoon, Tempus released results for the quarter and year ended December 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 subsequent 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 earnings release, which is available on our IR page. I would now like to turn the call over to Eric.

Thanks, Liz. 2025 was an exceptional year for Tempus, with both of our businesses growing rapidly and performing well above expectation. Total revenue of our core business was up over 33% when you factor in the acquisition of Ambry, obviously much higher. If you look at our two main businesses, I'll start with Diagnostics. In Oncology, we had unit growth of 29%, which has been accelerating throughout the year. We called out that our MRD growth rate was actually 56% quarter-over-quarter, which is extraordinary. Hereditary held up well with 23% unit growth. So all in, our Diagnostic business is accelerating and performing above expectation. In terms of Data, that business is growing even faster. It's made up of really two product lines, our licensing business and our applications. Our licensing business, or Insights, was up 69% in the quarter when you factor in the one-time impact of the AstraZeneca warrant, and we're projecting roughly 40% growth this quarter. Total contract value was greater than $1.1 billion and most importantly, has been rising faster than revenue over the past several quarters. And net revenue retention was 126%, which is super strong, all things considered. We guided to $1.59 billion, which is in line with our 25% long-term growth expectations and approximately $65 million of positive adjusted EBITDA. Our balance sheet is in great shape. Our products are resonating. Our AI advantages are continuing to take hold. So all in, we're poised for a phenomenal 2026. With that, I'm happy to take questions.

Eric, I wanted to zoom out from the financials. I'm sure that will be covered and go a bit broader. The markets are a bit anxious around AI and how value is getting distributed within that ecosystem. And we're now obviously seeing kind of traditional AI players push into the healthcare sphere. So I'm curious how you feel your position is protected on the data side? And I guess, to that point, why you expect the large pharma companies you work with, particularly within Insights to keep coming back for more? The Q1 guide is obviously strong, but I'm kind of looking on that. And then finally, just on that point, I'm interested whether the feedback you're receiving from these customers suggests that they're getting better at what they do because of what you're providing them. Just any sense of success stories and what you're hearing on the ground would be appreciated.

Yes. I believe the most compelling business models related to AI, especially large language or multimodal models, focus on having exclusive data for training models and unique distribution of insights once the models are developed. Tempus stands out in this regard because we possess both aspects at scale. We have over 450 petabytes of connected multimodal data from our Diagnostic business, which offers real-time insights and tracks patients over time, along with rich molecular and imaging data. This gives us a distinctive proprietary dataset for training AI models. Furthermore, once we generate insights that can be shared with doctors, our connection to over 5,500 hospitals and more than 8,500 oncologists, along with many other physicians, allows us to deliver these insights in real time as part of their clinical practices. This is what sets us apart; replicating our data business would require duplicating that extensive real-time data, which is a challenging task. It involves negotiating contracts with providers, obtaining data, addressing IT issues, structuring the data, and developing technology to make it functional. This has been an immense effort over the past decade. Because of our commitment to building efficient data pipelines, harmonizing and structuring the data, and creating supportive technology, we offer something unique that others have struggled to replicate. In terms of our data business, just a few years back, some doubted we could reach $100 million in data revenue, yet we are now four times that and expect to grow by 40% at this scale. Our net revenue retention stands at 126%, indicating that, on average, our clients are ordering much more each year. Our largest clients continue to renew their contracts, and we've announced several extensions recently. This is due to the tangible value we provide; clients are able to leverage our data and technology to make smarter decisions about which assets to pursue, improve their early-stage discovery projects, design more effective Phase IIs and III trials, recruit the right populations at the right time, and expedite the approval and market launch of their drugs. That is why our data business is thriving and experiencing accelerating growth.

Earlier in Q1, you launched Paige Predict. Given you previously discussed that you don't expect this to contribute meaningfully to revenue this year, can you discuss the strategic value of the added capability when samples are QNS? How often is this case with xT and xR? And a quick one for Jim. Jim, ASPs are expected to reach over 2,200 in the coming years, but what are you expecting in '26 guidance?

I'll begin with Paige Predict. I encourage you to read the letter we published, as it outlines some important points. We have consistently mentioned the significant advantages of contextualizing diagnostics and the technology that supports them. It's not merely about having the latest assay or conducting studies that lead to improvements; that only tells part of the story. Our strong growth at this scale is largely a result of our technological edge, which makes physicians more inclined to order our products because they gain more valuable insights from them. For instance, we have Paige Predict and our Immune Profile Score, among many others. With Paige Predict, we've developed technology that allows us at scale to digitize pathology slides and derive insights that help predict mutations observable in next-generation sequencing. The advantage of sequencing a large number of patients while also digitizing and tracking pathology slides is that we can correlate the two. Although all next-generation sequencing comes with a certain level of error—an inherent aspect of using companies like Illumina, where you don’t achieve 100% output—digitizing the pathology slide helps enhance our tests when sequencing results are limited. Additionally, our ability to provide those results in hours allows us to deliver insights faster. Similarly, with our Immune Profile Score, we're able to analyze various types of multimodal data, which may include digitized pathology slides, RNA transcriptomic data, and DNA data, to improve traditional biomarkers like tumor mutational burden. Each insight we create may not individually sway a physician’s decision, but they accumulate over time. Further, the foundational model efforts we are undertaking will likely accelerate this process. Therefore, if we are currently outperforming competitors by generating a certain percentage of additional insights, expect that difference to expand significantly over the coming years.

In response to the ASP question, ASPs in Q4 were approximately $1,640, which was an increase of around $40 from the previous quarter. As mentioned in our investor presentation and during discussions at JPM, we believe there is more than $500 of potential upside to ASP based on the current mix, driven by several factors. Firstly, there is the ongoing transition from the LDT version of xT CDx to the FDA-approved version. We have previously indicated our aim to have the majority of our volume on the FDA-approved version by the end of 2026, making this a significant factor for ASP in 2026. Additionally, we announced that we have submitted our xF, our liquid biopsy, to the FDA. While this is not expected to impact ASP in 2026, it should start contributing in 2027. There is also some potential upside from commercial payers as we gradually address those. Overall, these three initiatives contribute to the anticipated $500 upside, with xT CDx being the primary driver, and this will unfold throughout 2026.

So my question for Eric is about the foundation model and its potential to significantly enhance the stacking initiative and the development of additional algorithms and modules over time. Can you provide an update on the progress of the foundation model? Last quarter, you mentioned that you aimed to have the first version ready in the first quarter of 2026. I'm curious if that has been launched yet and how it is performing compared to expectations. Additionally, for Jim, you mentioned ASPs, but I'm interested in your thoughts on the underlying assumptions for volume growth in Oncology and Hereditary in the 2026 guidance.

Yes, I'll begin. We had a goal in Q1 for the foundation model to meet specific benchmarks set by AstraZeneca, and we believe we have successfully achieved those benchmarks. We have submitted everything to them, and they are currently testing the model. We are very positive about its performance. Our work on this continues. We initially established a dedicated cluster of just over 1,000 H200s for the Oncology foundation model, and we have since acquired a second cluster of more than 500 GB200, providing even more computing power than the first. Additionally, we are running extra models not only in oncology but across all our data, as we possess vast amounts of radiology, pathology, cardiology, and neuropsych data. We are highly optimistic about the value these models will bring, and we are committed to these initiatives. We believe they will significantly enhance our diagnostic business by integrating insights into our tests, making them smarter and more competitive, which should drive growth. Furthermore, we plan to share these insights through our data business to deliver greater value to our clients.

Yes. Regarding the underlying volume assumptions in the Diagnostics business, Oncology experienced 29% growth in Q4. The first quarter is starting well, and we're not seeing a slowdown in that momentum. Hereditary testing saw a volume growth of 23% in Q4. As previously mentioned, we expect this growth to moderate as we compare against some of the share gains. There may also be fluctuations in the Hereditary growth rates in 2026. In our communication, we indicated a projected high teens longer-term growth rate, which might be slightly lower in Q1 but is expected to improve throughout the year. While there may be some variability, we believe that the high teens growth rate is still attainable. We are very optimistic about the Oncology business. On the Hereditary side, we had anticipated some slowdown due to the comparison against previous share gains.

Congrats on a strong year. So I think in your letter, you talked about MRD volumes came in around 4,700 tests in the quarter. And I think you made a reference that, that level could have been 20x higher if the entire sales force had been selling it. So am I understanding it correctly that approximately 5% of your sales force was selling MRD in Q4? And is the right way to interpret this that if everyone sold it, it could be 20x higher in the Q4? Or is the 20x higher more of an aspirational longer-term outlook?

Yes, it’s difficult to determine what could have happened in Q1 since this is hypothetical. We were emphasizing the remarkable strength of our MRD offering. We conducted 4,700 tests, representing 56% growth compared to the previous quarter, and we are currently limiting this effort. Only a small fraction of our sales force is engaged in selling MRD. If we allowed everyone to sell it without any restrictions, the potential could be 20 times greater. However, that can only be confirmed once we remove the restrictions. The key takeaway is the growth remains exceptionally strong despite a limited sales effort. Eventually, we will expand that effort; this depends on reimbursement and the appropriate timing. We plan to gradually remove barriers and fully enter the market. Other companies with strong reimbursement, like Natera, took a considerable amount of time to secure broad coverage for financial viability. We are fortunate to be able to proceed thoughtfully without causing financial disruption. When considering our diagnostic platform, which connects broadly for hereditary profiling and therapy selection in solid tumor profiling and liquid biopsy, along with the extensive number of EHR connections and integrations we have, and our significant presence in the workflows of a large portion of the U.S. oncology market, I believe that once we remove these restrictions, we will emerge as a significant MRD supplier.

It's Darren Kirstein on for Kyle Mikson. So just taking a step back, in Oncology, I was wondering if you could provide a bit of more clarity on your tests. You briefly touched on ASP and volume dynamics, but could you just kind of walk through what each of the main growth drivers will be for xT, xR, xF, xH and xE in 2026 and beyond?

I find that question challenging to answer because it is quite broad. However, I would highlight in our letter the key platform advantage that has been driving our growth. This advantage lies in our ability to contextualize diagnostics by integrating clinical data with molecular data, providing a comprehensive profile. The insights generated from this integration are present across all our offerings: xT for DNA profiling, xR for RNA profiling, xF for liquid biopsies, and xH for our heme offerings, which includes a whole genome heme offering launching this year. Additionally, we have xE for whole exome analysis. Thus, our core technology advantage fuels the growth of all five assays rather than having separate drivers for DNA and RNA.

The other thing I would add is obviously, the market itself is also growing. And so amongst our peers, everyone is experiencing kind of healthy growth rate. So clearly, sequencing is becoming more prevalent amongst our ordering physicians and ultimately patients. And as Eric noted, kind of that data advantage is what allows us to kind of capture additional market share.

Yes. With us, obviously, in solid tumor growing, it looks like at this point faster than others.

Can you walk us through what the guide embeds for data and services revenue in 2026? I know that you pointed to $350 million of current TCV being tied to '26. So can you just talk about the visibility into in-year bookings to get to that guide and the timing around that? And then maybe as just a follow-up, can you split out the guide of the 40% growth you're assuming in data in Q1? Maybe just talk through how that will shake out across insights and trials and any contribution embedded from the current foundation model with Pathos and AstraZeneca?

So, as we mentioned during the JPM conference, the bookings have been exceptionally strong, giving us better visibility into the 2026 revenue build than we've ever experienced before. We noted at JPM that it is typical for us to generate around $100 million in revenue from bookings within the same year, meaning we fulfill requests for data within that time frame. The significant percentage of our revenue already secured for 2026 indicates our focus on effectively managing the growth of our data business, which is driven by strong systemic growth factors leading into 2026. There's a direct link between bookings and revenue, and we currently have over $1.1 billion secured, a considerable portion of which applies to 2026. We're starting the year on a very strong note with a tremendous demand for our data products, and it feels like we're distancing ourselves from the competition. Our data business is flourishing. When it comes to how everything else fits together, most of our data and apps revenue comes from data licensing, which constitutes the largest portion. The remaining segments, including our clinical trial matching business (TIME), our care gap product (Next), and some additional connected products, contribute to the overall picture, but they are relatively minor by comparison.

This is Colleen on for Doug. So you delivered high 20s clinical oncology volume growth exiting 2025. How should we think about the durability of that volume growth into this year and next? And how should we be thinking about liquid versus tissue CGP growth throughout this year? Any color you can share on repeat testing with xF? And if we should continue to think about xF being about 25% of total clinical oncology volume going forward?

Yes. I'll begin and then Eric can add his thoughts. Regarding the 29% growth and its context, we're off to a solid start this first quarter. We don't anticipate a significant slowdown in the Oncology growth rates. As we scale up, it becomes challenging to maintain that same growth rate, but we are beginning positively. In terms of details, MRD is experiencing much faster growth than the rest of the portfolio, and we see strong growth in both xT and xF. XF might be growing a bit quicker than solid, but the difference isn't substantial. This trend has been consistent for quite some time, and we don't expect significant changes in the product mix as we look ahead to 2026.

Yes, I'll add that our guidance indicates a 25% growth year-over-year. Our Hereditary business is expected to grow at a slower rate, around the high teens. We also have a few other segments that are not experiencing growth, including a small CRO business that we are not focusing on. Some revenue from Ambry is growing at a rate significantly lower than 25%, meaning our data business and core Oncology Diagnostic business are growing over 30%. As we indicated last year, our core business was projected to grow around 30%, and it actually grew by 33%. We expect something similar this year, so those businesses are healthy. Liquid diagnostics are growing faster than solid diagnostics, which has been a long-term trend. While we don’t disclose specific breakdowns, both segments are performing well. There isn't a single factor driving the growth; we are seeing strong overall growth. Increasing numbers of people want the advantages of tumor-normal profiling, DNA, and RNA services, and our intelligent connected platform is also in high demand. Therefore, our unit growth in Oncology remains robust with no signs of slowing down.

Eric, it's sort of been just over a year since you closed on Ambry. And if we sort of go back to when that acquisition was announced, sort of a big part of that thesis was really around the data that you would be able to sort of generate across the entire drug oncology testing spectrum. So can you maybe just sort of talk about the acquisition in that lens now that it's sort of been a part of the company for some time, just sort of what you're seeing in that regard and how that's led to growth in the data business as well?

Yes. So I think we called out multiple reasons to acquire Ambry. The first by far was to broaden the comprehensive nature of our testing compendium. So when if someone said to me, why did you buy Ambry? I wouldn't say data. I would say the #1 reason we bought Ambry was they had a very strong hereditary profile. More and more of our clients want a comprehensive solution. I think we've said this historically, I very much believe that over the long term, when it comes to treating cancer patients, it's going to be like e-commerce shoppers going online. I don't go to one e-commerce site for books and another for clothes and another for consumer electronics and so on and so forth. I go to Amazon because they have kind of everything I need in one place. And I believe that's going to be the case as it relates to sequencing. So more and more providers want somebody who can help them manage risk, help them treat patients once they've been diagnosed and help them monitor those patients post treatment. And so we want to have a broad menu. That said, we're also seeing a trend of more and more of our provider partners and certainly, obviously, to the benefit of patients, wanting to contribute de-identified data to platforms like ours to be used to accelerate research and to accelerate drug discovery and development. This is a very big trend that I don't see stopping. We still have 600,000 people a year that die of cancer. We're not making nearly enough progress as it relates to eradicating that. And so I think you're seeing a movement among institutions that are saying, we need to help stop the waste to make sure these patients get better drugs and get them in market faster. And so we see that as a constant movement and a benefit of the kind of data we collect and then on a de-identified basis, take to market.

Sorry about that, Eric. Just maybe one on MRD and on Insights. Just on MRD, any update on the first-gen CRC assay? I know it's sitting at MolDX. Just any color there? And on the next-gen tumor-naive assay, have you guys discussed kind of what type of performance advantage you would expect to get out of that? Obviously, you're filing for two tumor types this year and then you have another two that you mentioned in the letter. Just wondering what kind of performance enhancements do you think that could offer versus the existing tumor-naive landscape?

We are currently in discussions with MolDx regarding reimbursement. I don't have control over their timeline, so it could be resolved in a month or take longer. We did not highlight this issue because it is not particularly relevant to our existing MRD offering, which is primarily tumor-informed. Since we are largely focused on a tumor-informed product in collaboration with Personalis, that constitutes the majority of our market presence. The product we anticipate will receive MolDx reimbursement is unlikely to have a significant impact, as the criteria are constantly changing. Consequently, tumor-naive products must continually improve to remain competitive. In colorectal cancer, it will take time before we see a substantial shift from tumor-informed to tumor-naive approaches. The tumor-naive market will likely be limited to occasional cases or patients unable to obtain tissue samples, but tissue is commonly available in colorectal cancer, so tumor-informed options will dominate in that area for the foreseeable future. Other cancer types, such as lung cancer, where tissue is less available, may allow tumor-naive products to perform better. We found that the performance of our initial assay version was not meeting expectations, so we decided to shift our focus to the second generation instead of risking valuable study samples on the outdated version. We were fortunate to have a successful tumor-naive product in the market, generating more volume than we anticipated, which meant we didn’t have to aggressively promote tumor-naive options. The second version of the assay is progressing well, and I believe we will eventually have a robust assay market.

Eric, the start of the year is typically when companies adjust their sales organizations and their go-to-market strategies. First, if you could just give us an idea whether you've implemented any meaningful changes to the sales org this year? And then as a follow-up, maybe you could just sketch out what you believe are the firm's kind of key investment priorities for the coming year.

We haven't made any significant changes to the sales force. We reorganized it earlier in 2025 and shared the effects of that. The positive news is that we are past that period now. Our priorities remain the same: to widely integrate the benefits of technology and AI into diagnostics, ensuring that every decision, whether in clinical settings or research, is based on data. Given our rapid growth and business acceleration, I believe we will continue on this path.

Just wanted to get into some of the ASP outlook and maybe the gross margin implication. Obviously, a lot of upside here with the 2,200 test outlook. I just wanted to kind of hear about what the expectation should be for gross margin. It assumes a big lift. I mean, if you assume that the costs hold and do some math, it kind of gets towards gross margin in the genomics side of 70% plus. What could you say about that progression? And is that 1:1 with ASP? Or are there some offsets we should be considering?

Yes. When average selling prices increase, it generally leads to a rise in gross margin. We have always approached this by reassessing how we can expand panel sizes to generate more data as ASPs rise or as sequencing costs decline. This is beneficial for both patients and doctors. We review this aspect annually. Given our two business lines, Diagnostics and Data, we are not as focused on maximizing gross profit in the Diagnostics product line compared to some competitors. However, as ASPs increase, we expect to see some growth in gross profit, while also ensuring that we provide the widest range of panels possible, which has significant downstream implications.

Thank you all for joining us today. We look forward to updating you again next quarter. Have a great day.

Thank you again for joining us today. This does conclude today's conference call. You may now disconnect.