Third-quarter earnings call, November 2, 2026: revenue up 23% to $411.5 million, 2026 guidance raised to $1.60 billion to $1.605 billion

Tempus AI reported third-quarter revenue of $411.5 million, up 23% year over year, and raised full-year 2026 revenue guidance to $1.60 billion to $1.605 billion. The more consequential development was profitability. Adjusted EBITDA reached $25.6 million, which CEO Eric Lefkofsky described as the point where gross profit growth finally outran reinvestment. "We had long suspected that with gross profit dollar growth outpacing our investments, we would eventually turn the corner and be able to generate meaningful EBITDA. This quarter, we made the turn." Full-year adjusted EBITDA guidance was held at approximately $65 million, so the guide implies a materially lower fourth quarter run rate than the third, a point management did not address directly and one that will likely draw questions given the Personalis burn expected to join the P&L. Cash and marketable securities ended the quarter at $854.4 million.

Data Business Is Shifting From Transactional Licensing to Hosted AI Infrastructure

The most useful new detail on the call was how the Insights data licensing business has changed its customer relationships. Data and Applications revenue rose 33% to $108.1 million, with Insights up 29%. Lefkofsky explained that the model has moved well beyond the original practice of selling de-identified files that customers downloaded into their own environments. "We didn't know what you were doing with the data. We didn't know what kind of value you're getting from it. We would hope that you would call back 6 months later, a year later, want more data, but it was very transactional."

Customers now build cohorts inside Tempus' Lens platform, pull the relevant data into a workspace, and increasingly connect it to dedicated GPU clusters, typically 8, 16, 32 or 64 GPUs, on multiyear commitments. Models built there either stay in Tempus' environment or incur an egress fee to leave. Lefkofsky was blunt about the implications: "It's gone from like not sticky to crazy sticky. We just don't know how they would in mass get off unless somebody else builds the same set of products we've built, which we don't see." The company did not disclose what share of Insights customers use the AI modeling offering, a question Goldman Sachs raised, nor how much revenue the GPU commitments contribute. Lock-in is a strong claim, but the economics of this layer remain unquantified.

BMS Deal Extends the Data Franchise Beyond Oncology

Tempus signed a collaboration with Bristol Myers Squibb to license BMS oncology data and models for up to $137.5 million over 5 years. Separately, BMS, Pathos and Tempus agreed to build Alzheimer's data sets and models, with Pathos paying Tempus $55 million over 5 years. Together the agreements are expected to generate $192.5 million of additional data and applications revenue over roughly 5 years, or an average of about $38 million a year. The "up to" language on the BMS oncology license means not all of the $137.5 million is committed.

Lefkofskyidentified the Alzheimer's work as the first truly large non-oncology data investment by a customer, targeting a data set of roughly 0.5 million patients with Alzheimer's and late-stage dementia. He called it "the start of a whole new round of growth for our data business." Management noted that more than 80% of the roughly 50 million patients in the Tempus data set are outside oncology. Three major strategic collaborations have been signed this year, which Lefkofsky described as more than the company would have been content with from just one. CFO Jim Rogers argued that migrating customers from episodic purchases to 5- to 6-year contracts improves visibility into 2027 through 2029. Ryan MacDonald of Needham noted bookings above $550 million so far this year.

Oncology Pricing Provides a Large, Mechanical 2027 Tailwind

Oncology diagnostics revenue grew 28% to $172.6 million on 32% volume growth, an acceleration from the prior quarter. Rogers quantified the pricing setup for next year: with FDA approval extended to tumor-only use for the xT assay, updated pricing takes effect January 1 and allows the remaining xT volume to migrate, producing roughly $80 million of oncology ASP uplift in 2027. The xF liquid assay remains in the FDA process, and approval would bring a further step-up. Rogers said average realized oncology pricing should rise by almost $750 over the next couple of years, and that these dollars can help fund investment elsewhere.

Liquid biopsy appears to be gaining relative momentum. xF order acquisitions rose 25% year over year, and Lefkofsky pointed to a 300 to 400 basis point uptick in liquid growth in September. He attributed the trend to growing use of concurrent solid and liquid testing in metastatic patients, citing research that 10% to 20% of therapeutically relevant mutations are missed if only one modality is used, as well as serial testing. If the September pace persists into the fourth quarter, he said, "you will start to see the first signs of real differentiation between liquid and solid." He also noted that Guardant has posted strong volumes.

Hereditary Stabilizing, but Growth Is Well Below the Company Average

Hereditary cancer revenue was $111.3 million. This remains the slowest-growing piece of the portfolio, and the business is still lapping an outsized prior-year surge, when volume jumped to roughly 30% growth after a competitor's financial difficulties pushed testing to Tempus, versus a normal rate near 15%. Lefkofsky said unit growth returned to low double digits in September and October. On Canaccord's question about proposed clinical lab fee schedule rates reflecting PAMA cuts, management said it does not expect a material impact on hereditary cancer pricing. Rogers added that hereditary ASPs could drift up as the small rare disease business grows. Lefkofsky framed the opportunity as underpenetrated, with about 70 million lives covered under current guidelines against roughly 2 million tests a year in the US.

Personalis Closes Next, but MRD Scale-Up Is Gated by Unit Economics

Tempus has cleared regulatory review of its pending Personalis acquisition, with a shareholder vote scheduled for November 30. Tempus has distributed Personalis' tumor-informed MRD assay, NeXT, since 2023. BTIG's Mark Massaro noted volumes are rising by about 3,000 units per quarter with only about 15% of the sales force selling the test.

Lefkofsky was candid that the constraint is profitability rather than deal closing: "At the present moment when Personalis runs these tests, they lose money. And once we acquire them, we'll be in the same boat." He expects per-test gross profit to turn positive as reimbursement rises, with colorectal approvals the key catalyst, likely around mid-2027, possibly in the second or third quarter. Only then will Tempus "put the pedal to the metal" on the broader sales force, implying meaningful MRD growth in the second half of 2027 and into 2028. In response to Mizuho on competition, Lefkofsky said he sees "no competitive impediment" and compared Natera's position today to Foundation Medicine's in solid tumor profiling when Tempus entered that market. That is a confident assertion for a market where the leader has substantial scale, and the company offered no market share target.

Margin Framework: Two-Thirds Reinvestment Continues, Personalis Absorbs Some of It

Responding to TD Cowen's question on consensus margins of about 7% in 2027 and 11% in 2028, Rogers reiterated the company's framework of reinvesting roughly two-thirds of incremental gross profit and letting one-third reach EBITDA. He said that ratio would flip after a couple more years. Personalis' losses will consume part of the reinvestment budget. "We're not in a position where we want to maximize EBITDA in the short term because we think the opportunity set across our entire portfolio is so large that, that would be a poor choice." Lefkofsky said EBITDA should still improve even after absorbing Personalis. Management did not endorse the consensus figures, and the framework suggests upside from pricing may be partly reinvested rather than dropped through.

Long-Term Growth Target Stays at 25%, Not Higher

Cantor Fitzgerald asked whether 2027 and 2028 could exceed a 25% growth rate given oncology volume in the low 20s to 30s and ASP lifts. Lefkofsky declined to raise expectations. "We've always said we're more interested in long-term 25% growth than we are in kind of short-term 30% growth that then decelerates." He allowed that individual periods could run faster but said he would respond the same way. With slower hereditary growth acting as a drag, 25% plus is the company's stated aggregate level.

Clinical Trial Matching Is Not a Scale Business; CRO Work Deliberately Deprioritized

On trial matching, Lefkofsky acknowledged that a marketplace aimed at small biotechs did not gain traction. About 12 to 18 months ago Tempus refocused the TIME network, which spans more than 1,000 oncologists, on large enterprise data clients, which must load most of their oncology trials. A wait list now exists. Even so, he said "I don't think it's going to scale into a $1 billion business." Tempus is also deemphasizing pure CRO work, which he said is in flux because of AI, and does not intend to compete in that area.

Liquidity and Data Access Questions From Retail Investors

In its first Robinhood retail question segment, management said it was "super confident" in funding growth and the Personalis deal if AI investor appetite weakens, pointing to diagnostics ASP gains, long-term data contracts and the ability to scale back investment. On the risk that Tempus does not own patient data, Lefkofsky said more than 80% of oncology patients in the data set were not sequenced by Tempus. Data flows from more than 8,000 providers through multiple channels, which he said have been diversified "so that we never have to worry about is the data going to dry up." The company now holds more than 600 petabytes of data, with over 50 petabytes added in the quarter.