Tempus AI, Inc. (TEM) Earnings Call Transcript & Summary
July 30, 2026
Earnings Call Speaker Segments
Elizabeth Krutoholow
executiveThanks, everyone. Good afternoon, and thanks for joining us on Tempus' conference call for the second quarter of 2026. Earlier today, Tempus posted its results for the quarter that closed on June 30th, 2026. You can find the press release, a quarter overview, and our newest presentation on our IR website, investors.tempus.com. With me today from Tempus are Eric Lefkofsky, our Founder and CEO, and Jim Rogers, our CFO. Before we get started, I want to note that management will make forward-looking statements on this call that carry risks and uncertainties which might cause actual results to differ materially; to review those risks, please see our 10-K and other filings we've made subsequently with the SEC. We'll also talk about non-GAAP financial measures, which aren't prepared under generally accepted accounting principles. You can find definitions of those non-GAAP financial measures, plus reconciliations to the most directly comparable GAAP financial measures, in our earnings release, which is posted on our IR page. I'd now like to hand the call over to Eric.
Eric Lefkofsky
executiveThank you, and good afternoon to all of you. For Tempus, the second quarter proved to be a standout period. On the whole, revenue rose 22%, reaching $382.5 million, and this marked the first time we were comparing against a period that includes Ambry fully folded into our numbers. Revenue from our Diagnostics segment came to $289.3 million, which is 20% higher than the same period a year ago, since weaker expansion in hereditary cancer testing was balanced out by stronger growth in CGP testing thanks to momentum building in that business. That momentum is carrying forward, with June delivering among the best growth figures we've recorded so far across our offerings. Hereditary revenue during the quarter climbed 5%, hitting $107.4 million, because the second quarter of 2025 represented an unusually strong stretch of growth that we are now up against. Data and Apps brought in $93.2 million, up 28% versus a year earlier, while Insights, our data licensing and modeling operation, expanded 36% during the quarter. The quarter also brought a number of noteworthy developments. We secured FDA approval for tumor-only xT CDx; with this clearance, our whole solid tumor DNA portfolio can move under a single ADLT pricing structure. We anticipate an ASP increase of roughly $200, which works out to about $85 million yearly starting in 2027. It's also worth pointing out that our liquid biopsy product, xF, is currently under FDA review, and once it gains approval and reaches the market, expected during the back half of 2027, we foresee a further ASP gain of $550. Taken together, the xT CDx and xF approvals should generate around $400 million in additional revenue during 2028; we shared early findings from our foundation model and successfully handed AstraZeneca the initial version. That model was applied to forecast which patients showed a response across multiple public and blinded clinical trials. We're delighted to have reached this milestone and are currently developing the next iteration of the model; we entered into a sizable multi-year data licensing and modeling arrangement with BioNTech, which now stands alongside AstraZeneca, GlaxoSmithKline, Bristol Myers Squibb, and other partners. This deal, together with Merck last quarter, offers more proof that our data and modeling strengths are turning into essential tools for pharma. We additionally closed major agreements with Daiichi Sankyo, LevelSet Bio, and Incyte Pharmaceuticals, which helped push total bookings for the quarter to roughly $200 million; we wrapped up a $460 million offering of 0.0% convertible senior notes maturing in 2032. Part of the proceeds went toward paying off an outstanding loan from Ares Capital. Crucially, this deal lets us cut more than $30 million in annual interest expense, which puts us on track for positive free cash flow by the end of the year. GAAP net income came in at $5.6 million, while Adjusted EBITDA reached $8 million, an improvement of $13.6 million compared with a year ago. We closed the quarter holding $820.7 million in cash equivalents and marketable securities, up from $643.8 million in the prior quarter. As we had anticipated, cash consumed by operating activities dropped considerably, to negative $7.5 million for the quarter; beyond all of that, on July 20th, we disclosed an arrangement to purchase Personalis. When it comes to minimal residual disease, or MRD, testing, that's a market worth more than $20 billion, and it's among the quickest-expanding areas within oncology diagnostics. By acquiring Personalis, we speed up commercial uptake of our MRD test, complete the picture across our portfolio, and reinforce the multimodal data flywheel that sets our business apart. Because their financial trajectory keeps getting better, we concluded the moment had arrived to go after this strategic acquisition. Until now, our sales push has been rolled out in phases, since just about 10% of our sales force currently sells MRD, limited to those indications that have reimbursement. Despite that, the growth rates we're posting have come in above what we expected, with roughly 6,500 tests run in Q1 and roughly 9,000 in Q2, which is 38% growth quarter-over-quarter. Now that reimbursement exists for multiple indications and additional ones are on the way, we think volumes will rise substantially as we gradually arm more sales reps with NeXT. As for how the deal is put together, it's a 100% stock transaction, and Tempus holds the option to choose cash payment, though that's capped at 50% of the consideration paid. We've already started engaging with counterparties to arrange a debt facility, since our plan — clearly contingent on where our stock price sits — is to fund a substantial share of the proceeds through debt so that shareholder dilution stays minimal. Even after this acquisition, we still expect Adjusted EBITDA and free cash flow to keep improving in 2027. Now to guidance. We're raising it to $1.595 billion-$1.605 billion for 2026, which works out to roughly 25% growth. For 2026, we anticipate Adjusted EBITDA of about $65 million, which is an improvement of roughly $72 million compared with 2025. We're incredibly proud of what we delivered this quarter and can't wait to bring this momentum into the back half of the year. Operator, the line is yours for questions.
Kallum Titchmarsh
analystThanks very much for that. Perhaps this one's for Jim, regarding Personalis. A number of inquiries have come in about what you're assuming underneath on the average selling price side, and about how those economics might turn more in your favor as time goes by. Even a small bit of extra detail there would be great. Eric, I believe you mentioned this briefly during last week's call, but could you walk us through the way the additional MRD data you're going to have could flow back into your data business, and I suppose why that might make for a more persuasive data set for customers at this point. Many thanks.
Jim Rogers
executiveI'll take the ASP part first, and then Eric can handle the second one. Regarding ASP, clearly they've gained coverage across a number of indications during the past few quarters, so improvement has occurred on the Personalis side. Additional indications are also working their way through the pipeline. As time goes on, we'd certainly expect ASPs to keep getting better as they lock in coverage and further indications. And on the volume side, being able to grow the sales team capable of selling that test — which sits at roughly 10% right now — is going to help us push volume. They're at an early point on the ASP curve, yet they've clearly enjoyed a huge amount of success getting the first few indications approved, and we expect that to keep going.
Eric Lefkofsky
executiveTo add a little more context before I dive into the numbers: what makes their situation so persuasive, I think, is that they've got a genuinely strong set of trials underway, and we—along with everyone else—are tracking those readouts and responding as they become publications and, ultimately, approvals. Securing 3 approvals already is a real achievement, and there's an entire pipeline of additional ones on the way. What has really become clear during the past 30, 60, 90 days is that you can begin to foresee how the ASP narrative will shift for them in 2027. Suddenly, the economics that favor us—or that favor us right now, since we get paid and avoid losses—are going to reverse, and just like that they'll be the ones getting paid, they'll enjoy greater margin, and we'll be kicking ourselves for not having their deal rather than ours. That's definitely a strong element of the story, as Jim pointed out. Beyond that, I'd say there are compelling dimensions to their data as well. Nearly every large biopharma customer of ours that runs big trials is working to pin down the endpoint of those trials. Traditionally, we focus heavily on scans as a key endpoint for determining whether disease is coming back, whether there's progression, or what's going on. Increasingly, though, you're picking up earlier indications from these MRD tests that reveal signs of cancer returning 6 months or 12 months ahead of a scan. Naturally, if you're a pharmaceutical company, having visibility into when patients relapse and being able to start them on a drug sooner matters enormously; we have a steady flow of parties asking us to incorporate MRD data alongside the existing data they rely on for licensing and modeling. My expectation is that, in time, it turns into a truly compelling piece of our broader data portfolio.
Brad Bowers
analystHi there. I appreciate you taking the question. To start, congratulations on the big deals you closed during this quarter. I'd like to zero in on the AstraZeneca component specifically. Also, congrats once more on getting the first iteration of the model delivered. Perhaps you could elaborate a bit on how that shapes up, and then I believe there's an aspect of the elephant in the room regarding what the agreement will look like for 2027 and beyond. From my perspective, it appears the foundation model clearly plays a major role in that. Maybe you could shed some light on how that contracting piece fits in. Just a reminder about the sorts of escalators that might be present, and whether the foundation model catalysts arrive at some point following this year, meaning the contract would need to already be in place. Thank you.
Eric Lefkofsky
executiveYeah. AZ accepted the foundation model. That mattered a great deal because specific criteria had to be satisfied by us. What's neat about that is you're training this enormous multimodal model built on billions of parameters, extremely complex, and it needed to match the performance of particular models that both our team and their team had created, which were highly tuned for specific use cases, including predicting response to both public and private trials. We'd send those models over to them, and in a blinded manner they would essentially evaluate how our large model did against their own internal models. We lacked access to a good portion of that data. The fact that the acceptance criteria were met means they feel confident this model is predictive and can now act as the foundation — even though it is a foundation model — for every kind of R&D and development effort they're pursuing. That's an enormous hurdle, we're thrilled, and we remain committed to investing in it. Aside from our foundation model work, they clearly are a licensee of our data along with a whole bunch of our products. As I recall, the current agreement we have with AZ runs for another couple of years; it doesn't conclude at the end of this year. I believe the existing agreement lasts, I'm not even certain, through 2028 or thereabouts. I really don't know. Several years remain on it, even once this year ends. There are specific criteria under which they can opt into preferential pricing; if they don't, they'd simply pay more for the data they license. For starters, there's a bunch of projects they've already committed to that will carry into 2027. Regardless of what happens, they'll be a very large client in 2027. I literally cannot picture a scenario in which they wouldn't want to lock in for a longer period to take advantage of discounts. That would simply make no sense. They've given us no signal that they won't want to lock in for a long period and take advantage of discounts. My suspicion is that we'll deliver a similar amount of data and revenue to them next year. My guess is that eventually there will be a long-term extension in place, or they will simply rely on the existing contract and agree to comparable dollar amounts for data. Based on everything we are seeing, such as their CEO discussing it, I believe on CNBC or somewhere similar, they are extremely pleased and plan to remain our long-term partner.
Kyle Mikson
analystHi, everyone. I appreciate you taking the questions. Nice work on this quarter — congratulations. My first one concerns the xF FDA clearance tailwind: it appears to now sit at $550 based on 2Q figures, whereas at the investor day you cited $230 drawn from 4Q figures. I don't believe you explained what drove that shift. Could you address it? Second, since your shares are changing hands under $46, there's a chance Personalis could walk away. Could you discuss steps you can take to prevent that, and also why you're confident they won't pull the plug — that termination won't happen? Thanks.
Jim Rogers
executiveYes. Regarding xF pricing, now that other players have moved through the approval process for liquid biopsies and have signaled the prices they intend to pursue, how we think about ADLT pricing for xF has shifted, and we believe there's more upside than the figure we had estimated previously. That assay currently sits before the FDA. Per Eric's remarks, as we move deeper into 2027, we'd expect to secure approval and then pursue the ADLT pathway; that explains the revision.
Eric Lefkofsky
executiveWell, the market's always changing. When you look at our assay, the closest comparison by size—literally in terms of total megabases—is the assay Guardant recently got cleared, and I think their ADLT price sits around $83 or $8,400. For us to enter the market with something nearly the same, at least in size and complexity, but priced far lower would be really tough. Since we've got products of similar complexity and size, we need to align with those who came before and established ADLT pricing. The price point here exceeds what we'd anticipated, so that's a meaningful upside for us. Once it's approved and on the market, it'll be a meaningful upside. That's where the big lift comes from; as for Personalis, I really can't picture them wanting to walk away, even if we dipped a bit under the floor; the floor exists because we weren't prepared to accept dilution beyond a certain amount. Clearly, cash is one lever we hold. Stock is another lever we hold. We're not looking to accept dilution beyond a certain amount. Considering our current trading level, my clear preference would be to cover roughly half the deal with cash and the rest with stock, which keeps dilution pretty minimal. I think we'll get that chance, and I really can't picture any situation where this fails to close. As you'd expect, they're eager to complete this deal; they're already a partner of ours. On their end, it would be extremely disruptive if this deal fell apart, and I simply can't picture any situation where it doesn't move ahead, even if they wind up with somewhat fewer shares.
Matt Shea
analystHi, Matt Shea here, filling in for Ryan. Appreciate you taking my question. Eric, the data and Insights segment has had genuinely strong momentum across the opening six months of 2026 — that includes the BMS expansion back in May plus quite a few deals announced earlier today. Could you speak to how much momentum you're carrying into the second half? And then maybe for Jim: once we factor in that BMS expansion alongside $200 million of bookings this quarter, layered onto the $350 million of TCV already set aside for 2026 revenue, what kind of visibility and confidence do you have around reaching that implied $410 million figure for data revenue guidance — assuming that's still the correct number? With the raise, it could be somewhat higher. Where do you see the levers for upside?
Eric Lefkofsky
executiveYeah. Jim, feel free to jump in, though I think my answer will address both parts. Given the deals we've been putting pen to paper on — first off, momentum has picked up. I believe I said this on the prior call, or the one before, but the data business is absolutely on fire. The pace at which we're signing deals is greater than it's been in quite a while, many years. Apart from the Foundation model, this is likely the strongest stretch of three or four quarters we've ever recorded when it comes to momentum. We're simply having an incredible moment. A growing number of customers want our data — and even more significantly, what's truly thrilling is that it isn't only our data they're after; they want Lens access, they want us wiring up and provisioning GPUs inside Lens on their behalf, they're putting their data up there. They're developing models that stay within Lens. The whole business simply feels extremely healthy, extremely sticky, and our pipeline is stronger with greater demand than ever before — meaning we enjoy excellent visibility into our growth rates, not only for 2026 but for 2027 as well. That's the lens through which we view the data business. We're genuinely focused on sustaining long-term growth in that neighborhood of 30% or better. Our aim is to map these things out so that we're confident about growing at that pace across many years — three years, five years, seven years. We feel terrific; we're positioned wonderfully for 2026. We're positioned wonderfully for 2027, and these days a lot of our thinking goes toward 2028.
Mark Massaro
analystHi everyone, thanks for fielding our questions, and congratulations. Perhaps I could begin by asking for some clarity regarding the greater pricing estimates for xT CDx — sorry, I mean the xF. Could you describe which rates or price points you're projecting for the Medicare portion? Since you referenced Guardant, being more specific would really help. Eric, at what point do you expect you'll be able to raise that 10% toward promoting the Personalis test? What's stopping you from increasing it more quickly? Do you believe you might do so sooner instead of waiting, or is the plan to hold off until the deal potentially closes?
Jim Rogers
executiveSure; regarding xF, Mark, our assumption is an ADLT price of $7,500.
Eric Lefkofsky
executiveWhen it comes to ramping MRD more quickly, the restrictions that applied while Personalis wasn't ours are going to remain in place once this deal is done. Specifically, we want to schedule the complete removal of the restraints on these programs around the moment when test pricing reaches roughly the break-even point. When you're underwater, when your margin is in negative territory, and you hurry to process another 100,000 tests, all you're doing is torching cash. If we believed this market amounted to prime waterfront property that had to be acquired, we'd go ahead. We'd announce to everyone, "We intend to torch a large pile of cash, and here's our rationale for believing it's worthwhile." That isn't our view. We held the same view regarding therapy selection, and were that true, Foundation Medicine would be the one ruling this space, not Tempus and Caris. Our view is that there's no prime waterfront property available for acquisition. Our view is that it does matter for us to be present in the market with a comprehensive offering that customers actually want. We're confident we can dole this out gradually without surrendering the market opportunity. Clearly, our growth is extremely rapid. The sequential rate is 38%, and we're reaching meaningful scale, with considerably greater scale coming in 2027. Eventually there will be a turning point when ASPs begin rising and break-even comes into view, and at that moment, my expectation is that you'll see us expand the sales force quite aggressively. Even so, it won't be visible to you, because of how our business's underlying economics work on the gross margin growth side. The discretionary investments we make are so large that if we chose to put an additional $50 million into the sales force, we'd simply allocate $50 million less toward cloud or other items that are invisible to you, and we'd remain EBITDA-positive and cash-flow-positive regardless. Our situation is excellent because the core business is now producing so much gross margin and gross profit dollar expansion, and we're layering in so many long-duration incremental investments, that we're able to fund things such as sales force expansion without slipping into negative EBITDA, negative cash flow, or regression. Our position, I'd say, is strong.
Ricki Levitus
analystHello, Ricki here, filling in for Subbu. We appreciate you letting us ask questions. After GenomeNext came out, is there anything new regarding how you see the rare disease ramp at Ambry? Your letter said you anticipate that accelerating during the back half. Could you put a number on that for us? What would a strong second half for rare disease at Ambry look like?
Eric Lefkofsky
executiveYes. I'll handle the first part. The launch went really well. By that I mean we had a target for month one, and roughly 2 or 3 weeks in, we were already exceeding it by 50%. Still, those are modest figures. Ultimately, this is a brand-new offering for us, so reaching the market and moving 500 or 1,000 tests represents a solid beginning. I do believe there's some upside coming in the latter half of the year tied to whole genome. We simply lack sufficient visibility to know yet. At present, it isn't taking business from our whole exome operation. It won't stay non-cannibalistic forever. It isn't now, but we've only got 1 month of information, clearly, we're working to lift our hereditary growth rates, we see that business reaching mid-teens growth by year-end; we're staying conservative with our whole genome projections, even though it will accelerate. I think we're somewhat in a wait-and-see mode about how this will play out, again, lucky that we don't depend on it since our 2 core businesses, oncology testing and data, are beating expectations, we'll be okay.
Brendan Smith
analystHey everyone, thanks for fielding this. I'd like to piggyback once more regarding the data and Insights segment. Building on what you said about the traction there. You brought up certain recent arrangements that sound like they could be broader in scope, with a few of those pharmaceutical players aiming to utilize Lens, and you also referenced additional data and app products. From Tempus's financial standpoint, is it fair for us to think that the structure of such arrangements could translate into stronger top-line results for you throughout the life of the collaboration? Or perhaps it means booked revenue gets recognized sooner rather than sitting in backlog? I'm mainly asking to get a clearer picture of how those different drivers play out, and what the right way is to think about the difference between reported revenue and TCV as an increasing number of those customers come aboard and start actively using the platform.
Eric Lefkofsky
executiveSure, I'll begin, and then Jim can add anything. Some background might help here. There was a time when our model involved approaching clients and telling them, "We've got de-identified data available. If you'd like to license it, we'll deliver 5,000 files to you, and you pay us." That made for very uneven revenue, though we booked it all at once. Several years ago, we made a change: we halted taking all that money upfront and instead transitioned customers onto licenses lasting one, two, three, or five years, deferring a significant portion of those funds. That was hard to accept at the time, yet it has proven excellent for the company's long-term strength, since we now enjoy tremendous visibility many years ahead. Regarding these fresh agreements—where customers become further embedded with Lens and deepen their work constructing either compact or extensive models within our platform, while tapping GPUs at a certain magnitude—I don't anticipate any alteration to revenue recognition. They simply represent one more source of stickiness that binds users to our ecosystem. There are two factors keeping them committed. First, they commit to lengthy agreements with set durations, and cancellation isn't possible until whatever that set duration happens to be runs out. Second, they're now tied in since the models they're developing live inside our environment—those models can't be moved elsewhere. That being said, what we believe is the primary driver of their stickiness is simply that the quality of our data is excellent and the tooling delivers genuine value. Were that not true, our data business wouldn't be as robust as it is, nor would it keep expanding at such a rapid pace.
Robert Bamberger
analystYes, I appreciate you answering my question. You previously mentioned an algorithm attach rate of roughly 40% for solid tumor. Does that figure still hold for Q2? Also, which algorithm is responsible for that, and what attach rate has been factored into the guidance you're providing?
Jim Rogers
executiveFor Q2, the algorithm attach rate came in at 45%, which is a modest increase over the 40% figure we cited in Q1. It's genuinely spread across the board. Naturally, we offer a collection of algorithms targeting various questions or insights that doctors might request, so the mix of algorithms being ordered is quite diverse. As for the guidance, a lot of those algorithms are still not getting reimbursed, so the quantity of algorithms has no effect on revenue. That said, it does once again underscore that our strength in diagnostics lies in the insights we deliver to physicians past merely the test results. This supports that growth in core volume, which rose to 31% during Q2, and represents yet another element of the data edge we possess.
Elizabeth Krutoholow
executiveAppreciate you all being here. Should anything come up, feel free to get in touch with the investor relations team. Enjoy the rest of your day.
Operator
operatorEveryone, that wraps up the session for today. Thanks to each of you for being here. You're free to drop off now.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Tempus AI, Inc. transcript — plus 132,000+ transcripts from 7,000+ companies, speaker segments, AI summaries and full-text search — through the EarningsCalls.dev API.
Get the API View API docs →This call discussed
For developers and AI pipelines
Programmatic access to Tempus AI, Inc. earnings transcripts and 132,000+ others is available through the
EarningsCalls.dev REST API. Plans from $39.99/month — full transcripts, speaker segments,
full-text search, and the recently-added /api/v1/transcripts/recent polling endpoint for ETL pipelines.
More tools for stock research
OptionsBell: an unusual options activity scanner that tracks options flow across all US stocks and emails you when large trades appear in Tempus AI, Inc. or any ticker on your watchlist.
StockMarketScan: a stock screener with alerts: 24 pre-built screens refreshed every 5 minutes, watchlists and email notifications. Both offer API and MCP access and fair monthly pricing.