Tempus Made the Turn — Now Comes the Hard Part of Proving It
A first real adjusted-EBITDA quarter, an ASP guide that quietly jumped from $500 to ~$750, and an AI story that stopped being about foundation models
TEM · Earnings Call · 2026-11-02
The Turn, and the Asterisk Attached to It
Tempus did something in Q3 that it had been promising for roughly a decade: it produced an adjusted-EBITDA number that isn't an apology. Revenue of $412 million (+23%), adjusted EBITDA of $25.6 million, $854 million of cash and securities, and a 2026 revenue guide lifted to roughly the $1.6 billion line.
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. Despite our increasing investments in diagnostics, we have always been a tech company.
Fair enough — and the market has already voted, with the stock up more than 50% over the last ninety trading days into the print. But two things complicate the headline. First, full-year adjusted EBITDA guidance sat at ~$65 million while Q3 alone delivered $25.6 million — implying a step back in Q4 as the pending Personalis acquisition starts absorbing burn and the company keeps pouring two-thirds of incremental gross profit back into the business, the reinvestment framework Jim Rogers has now restated on three consecutive calls. Second, the number is adjusted. Stock-based compensation ran $54 million in the last reported quarter against $382 million of quarterly revenue, and free cash flow less SBC was negative $68 million. Receivables sat at 94.4% of revenue, roughly ten points higher year over year. The turn is real in the narrative; it is not yet real in the checkbook.
The Number That Actually Moved Was a Reimbursement Number
Strip out the AI framing and the biggest single delta on this call was pricing. In February, Rogers framed the oncology ASP runway as “about $500 or greater than $500 of upside to ASP” — James Rogers, Chief Financial Officer · 2026-02-24. In May, he explicitly told investors the pending liquid-biopsy filing would “not expect that to have any impact on pricing or ASPs in 2026” — James Rogers, Chief Financial Officer · 2026-05-05. On Monday, the runway was re-cut upward and given a date:
“That updated pricing will kick in on January 1 and allow us to move the remaining XT over. That gives us about $80 million of uplift in oncology ASP in 2027.” — James Rogers, CFO · 2026-11-02
And in the retail Q&A, Rogers put the aggregate at “almost $750 over the next couple years” — James Rogers, CFO · 2026-11-02 — a materially larger figure than the number management was anchoring to nine months ago. The mechanism is unglamorous: the recent FDA approval expansion to tumor-only for XT opens the ADLT reimbursement rate, and the migration of remaining volume happens automatically on January 1. Nothing about that requires a new model, a new GPU cluster, or a new disease. It is the single most underappreciated part of the quarter, and it is why oncology revenue (+28%) grew slower than oncology volume (+32%) — the price step-up hasn't landed yet.
The AI Story Changed Shape Without Anyone Saying So
What's striking is what faded. In February, the foundation model built with AstraZeneca was the marquee topic — Eric Lefkofsky described “a particular cluster we set up of, about a little over 1,000 H200s dedicated to that Oncology foundation model” — Eric Lefkofsky, Chief Executive Officer · 2026-02-24, plus a second GB200 cluster. It ranked as the company's single most prominent keyword in Q2 2026. This quarter the compute conversation is entirely commercial: Tempus sells three-year slices of 32 or 64 GPUs attached to its Lens platform, and the CEO's framing is customer lock-in rather than model capability.
“It is gone from, like, not sticky to crazy sticky and so we just — you know, in light of how people are using our data today, we just do not know how they would unmask it off unless somebody else builds the same set of products we have built.” — Eric Lefkofsky, CEO · 2026-11-02
That is a quieter, better business. Data and Apps grew 33% to $108 million with Insights at 29%, and the BMS oncology data licensing arrangement (up to $138 million over five years) plus a $55 million Pathos-funded Alzheimer's dataset gets Tempus to $193 million of incremental contracted revenue. More important than the dollar figure: it is the first genuinely large non-oncology data deal. Lefkofsky calls it “the start of a whole new round of growth for our data business” — Eric Lefkofsky, CEO · 2026-11-02, and backs it with a striking disclosure — of the roughly 50 million patients in the dataset, “more than 80% of those patients are outside oncology” — Eric Lefkofsky, CEO · 2026-11-02. That is the answer to the "what if AI data licensing in oncology saturates" question, and it arrived via a partner paying Tempus, not the reverse.
What Went Quiet Matters as Much as What Got Loud
The trial matching business, a fixture of the data segment narrative for years, was effectively written down in plain English: the Thyme marketplace experiment was abandoned for an enterprise model, and “I do not think it will scale into a billion dollar business” — Eric Lefkofsky, CEO · 2026-11-02. That is an unusually candid admission for a company that has spent years teaching the street to think of data as a portfolio of large option-value bets.
At the other end, the Personalis big unlock — MRD — finally got a date, but the gate isn't the deal close, it's per-test unit economics. With roughly 3,000 net-new MRD units added per quarter and only ~15% of the salesforce selling, Lefkofsky now guides the inflection to “midyear. It could be Q2. It could be Q3” — Eric Lefkofsky, CEO · 2026-11-02 of 2027 — a year later than the "next several quarters" language that surrounded the partnership in May.
Finally, the tape. Tempus is up 50.8% over 90 days and sits 15% below its late-September high, while the AI data center complex — the forty-plus tickers that move with capex and HBM — is down more than 20% on the month with fifty-plus decliners. Tempus has decoupled from that trade. Price-to-revenue is 7.3x, 44% below its 2025 peak, and the diagnostic peer group that reported alongside it — non small cell lung cancer exposure from Merck, Bristol, AstraZeneca, Regeneron, Exelixis — all told the same demand story in oncology tissue testing. Tempus isn't differentiated by the market being good. It's differentiated by the reimbursement step function that starts January 1, and by a data business that just found a new organ to sell into.