Tempus AI: A Conference Transcript, a 'Frenzy of Demand', and a Deal Whose Economics Just Flipped

Our view

the transcript contains one fact and one mood. The fact is the deal logic: Tempus distributed Personalis' whole-genome MRD assay for years at about $400 per test while Personalis lacked reimbursement, and Medicare coverage for lung, breast and immunotherapy monitoring is now lifting the assay's price, which is why Tempus chose to own it before that value accrued to the seller [1]. The mood is the demand language, which is not a number. The second quarter was the company's first profitable one, $5.6 million of net income on $382.5 million of revenue [3][4]; a 10.7% move on tone puts the burden on the third-quarter report to show it in bookings.

Starting market state

At the September 15 close TEM was $68.85, +10.7% on the day on 8.2 million shares, in the memory's state "up/above/off/clear/ordinary/strong/crowd_in": a strong move above reference levels on ordinary volume with crowded participation [6][7]. Tuesday's move is inside the state being scored. The pre-market print at 11:31 UTC on September 16 was $68.70, -0.2%, an indication rather than a completed session [6]. Attention: TEM ranked fifth on the Yahoo trending list at 07:15 ET and did not appear on the Stocktwits list [8].

What changed

The primary source is a Rule 425 filing accepted at 21:25 ET on September 15 containing a transcript of CEO Eric Lefkofsky's conference discussion, filed because it concerns the pending Personalis acquisition [1]. In it he describes a frenzy of demand as AI labs and health systems strike compute deals, with Tempus as the fuel that makes that spending intelligent, and explains the acquisition: Personalis' assay is whole-genome based with sensitivity and specificity he calls extraordinary; Tempus had been its exclusive US distributor under an arrangement that paid Tempus about $400 per test regardless of reimbursement, which favored Tempus while Personalis went unpaid; approvals for lung, breast and immunotherapy monitoring made it apparent that Personalis' average selling prices would rise and the arrangement would flip [1]. The merger agreement was signed July 20: Personalis holders receive Tempus Class A shares at an exchange ratio, with Tempus able to elect cash of $16.25 per share for up to half the shares, and a termination fee of about $76.8 million payable by Personalis in specified circumstances [2]. Date learned: the transcript overnight, the deal on July 20; economic event: the merger close, subject to the Personalis vote.

Expectations

Second-quarter 2026 revenue was $382.5 million and net income $5.6 million, the first profitable quarter after a $125.9 million first-quarter loss; cash and equivalents were $599.6 million at June 30 [3][4]. The first-half loss was $120.3 million [4]. We did not retrieve the share count from the cover page in time for this note and did not retrieve a dated sell-side consensus; the market value implied by Tuesday's close depends on that count and is left to the reader with the caveat. Our labelled reading of what the move priced: the transcript's demand language shifted the expected slope of genomics and data revenue for the second half, and the deal explanation removed a worry that Tempus was buying a distributor relationship it already had. Neither is a number the company has committed to; third-quarter guidance in the July 30 8-K and any update on the third-quarter call are what will test it [3].

Observed response

Tuesday's +10.7% on 8.2 million shares is observed and is the event that defines the current state. The completed September 16 session is not yet observed; it will be scored after the close against the memory's 1-session band, with the Federal Reserve decision at 14:00 ET as a confounder, then the 5-session excess return by September 22.

Close update, session 1 (September 16, written after the close): TEM closed at $69.97, +1.6% versus the $68.85 close of 2026-09-15, on 7.5 million shares, about 0.9 times the 20-day average; the session's range was $68.42 to $71.48 after a $68.79 open. Against the market (SPY -0.4%), the excess move was +2.1 points. Against the 1-day band from the pre-news analogs (p10 -3.8, median -0.2, p90 +4.2), that is inside the band.

Comparable history and its limits

The memory's analogs describe states like Tuesday's close, after the rise: 300 analogous states across 200 symbols, with the 5-day excess band -9.6 / +0.2 / +9.3 and up 51% of the time [7]. That is a symmetric, wide distribution: for strong crowded moves up, the memory has seen follow-through and give-back in nearly equal measure over five sessions, and the band is not conditioned on the cause. Among names we have written about, the nearest comparison is VERA, a biotech whose readout produced a one-day move that the close-to-close number then hid; both are cases where the intraday path will matter more than the five-session number.

The countercase

The case for the market is that a founder-CEO describing a frenzy of demand, on the record in an SEC filing, after a first profitable quarter, is exactly the moment expectations should reset, and that owning the MRD assay ahead of reimbursement-driven price increases is the kind of timing that produces durable revenue. The case against is that the words are not bookings, that the stock has moved 10.7% on a transcript rather than a filing with numbers, that insiders filed sales notices in early September, and that the deal adds shares to a company that has one profitable quarter behind it [1][2][5].

What would change our view

Our view is wrong if the third-quarter report shows revenue acceleration and raised guidance that match the transcript's language, or if the Personalis deal closes with the assay's reimbursed volumes visibly rising. The market's move is wrong if the five-session window gives back most of Tuesday's gain and the third-quarter numbers do not show the demand the transcript describes.

What we will revisit

Session 1: the completed September 16 session, written up after the close (Federal Reserve decision at 14:00 ET is a confounder). Session 5: September 22. Session 10: September 29. Company: the Personalis stockholder vote and closing; third-quarter results in early November. The company's share count and a full read of the July 30 10-Q were not completed for this note and are disclosed as gaps; the issuer's press-release index was not machine-readable by our scanner at check time. The issuer's press-release index and presentations page were not machine-readable by our scanner at the check time; issuer statements that morning beyond the cited filings are a coverage gap.