Tempus AI (TEM) Q2 2026: The First Profitable Quarter Was a $97 Million Mark-to-Market Gain
Tempus AI reported GAAP net income of $5.6 million for the quarter ended June 30, the first positive bottom line in its public history, and the number is almost entirely an accounting event. Loss from operations widened to $75.9 million from $61.8 million a year earlier. The line that closed the gap was $97.4 million of fair value changes on marketable equity securities, non-cash and non-recurring by construction. Strip it out and Q2 was a larger operating loss on slower growth. Management did not hide this; the press release discloses the $98.5 million of unrealized gains in the same bullet as the net income figure. Coverage that leads with profitability is reading the wrong line.
The number that actually describes the business is the eleven-point spread between oncology volume growth of 31% and Diagnostics revenue growth of 20%. Volume is outrunning revenue by a wide margin, and the drag is visible in the segment detail: Hereditary, the Ambry Genetics business acquired in early 2025, grew 5%. Total revenue of $382.5 million was up 22%, against 36.1% in Q1 and 83% for full-year 2025. Guidance of $1.595 to $1.605 billion implies roughly 25% for the year and about 24% in the second half. The company is settling into a growth rate less than a third of what it printed in 2025, and the comparison base gets no easier once Ambry annualizes fully.
The most consequential item in the release is also the least discussed. FDA approval of the tumor-only xT CDx assay makes Tempus the only laboratory holding companion diagnostic approval for both tumor-only and tumor-normal comprehensive genomic profiling, and management flagged that it migrates tissue testing to ADLT pricing. Under PAMA, Advanced Diagnostic Laboratory Test status lets a lab bill Medicare at its own list price for three quarters, after which the rate resets to the median private payer rate. The approval is therefore not the value event. The value event is what commercial contracts Tempus signs during those three quarters, because those contracts set the rate that persists afterward. A lab that lists aggressively and negotiates poorly gets a temporary revenue bump and a permanent reset lower. This is the single largest ASP lever in the model and it will be decided in negotiating rooms over the next four quarters, not by the FDA letter.
The moat is not in the sequencing. Chemistry commoditizes, gross margin of 62.6% in Diagnostics is lab-services economics rather than software economics, and Guardant, Foundation Medicine, Caris and Natera all run comparable assays. What Tempus owns that cannot be purchased is the longitudinal linkage between molecular results and downstream clinical outcomes, an asset that accrues only with elapsed time and only through the Hub integration sitting inside physician workflow. Data and Applications generated $93.2 million, up 28%, with Insights up 36% and roughly $200 million of bookings signed with BioNTech, Daiichi Sankyo, Incyte and Level Set Bio. Delivering a working oncology foundation model to AstraZeneca is a harder credential than any assay approval. One qualification belongs on that number: $22.0 million of the $93.2 million was related-party revenue, up from $15.9 million. Excluding the SB Tempus relationship, the segment grew 25% rather than 28%, and segment gross margin fell to 70.2% from 72.7%. The software business is becoming less software-like as delivery and inference costs rise.
The cost structure explains why adjusted EBITDA of $8.0 million should be treated as a presentation choice. Stock compensation and related payroll tax ran $55.6 million in the quarter, 14% of revenue, against 7% a year earlier. The entire positive EBITDA figure is manufactured by adding back a cost that increased $32 million year over year. Operating cash flow was negative $80.8 million for the first half; trailing twelve-month free cash flow is negative $263 million. Full-year adjusted EBITDA guidance of roughly $65 million is unchanged, which means the second half is expected to carry $60 million of it.
The genuinely good news went unremarked. Tempus placed $460 million of zero-coupon convertible senior notes due 2032, retired the $207.7 million term loan and the $100 million revolver, took an $11.6 million extinguishment loss, and bought $31.2 million of capped calls. Quarterly interest expense fell to $10.3 million from $21.6 million. That is a well-executed refinancing that removed expensive secured debt at the cost of future dilution priced above the current stock. Convertible notes now stand at $1.17 billion against $820.7 million of cash and marketable securities and $444.9 million of book equity, a debt-to-equity ratio of 3.25 and a net debt position of $628 million.
The Personalis terms deserve closer reading than they have received. Consideration is $16.25 per share against a floating exchange ratio, fixed at 0.3356 once the Tempus stock price falls to or below the $48.42 floor. TEM closed at $43.87 on July 31, which puts the ratio at its cap. Applied to roughly 106.8 million Personalis shares, that is about 35.8 million new Tempus shares, near 20% of the current 180.4 million count before assumed options. Two asymmetries follow. Personalis holders now absorb every dollar of Tempus decline below $48.42, and they simultaneously hold a termination right if the fifteen-day VWAP into closing is below $46.00, exercisable in a two-business-day window before the scheduled close. Tempus has capped its dilution and given away deal certainty. Termination fees run about $76.8 million in each direction, and the outside date is April 20, 2027 with automatic extensions. Tempus retains the option to fund up to half in cash, which would require borrowing against a $820.7 million balance and is constrained by the 40% stock minimum needed to preserve reorganization treatment.
The stock told the story better than the print. Adjusted EPS of negative $0.04 beat the negative $0.15 consensus, revenue beat $379.5 million, guidance rose above the $1.59 billion street number, and shares opened at $46.24 and closed at $43.87, down 0.95% on the session. A beat-and-raise that fades intraday is a multiple problem, not an execution problem. TEM is down 24% over twelve months against a 52-week range of $40.77 to $104.32, trading below both the $51.03 fifty-day and the $59.40 two-hundred-day averages with an RSI near 38 and a beta of 3.61. Market capitalization is $7.92 billion, enterprise value $8.54 billion. Short interest sits at 30.4 million shares, 16.8% of shares outstanding and 26.5% of float, down from 33.6 million the prior month. Baird trimmed to $59 from $67 while keeping Outperform; the fifteen-analyst average target is $66.21.
Valuation is clearer when the two businesses are separated. Annualizing Q2, Diagnostics runs at $1.16 billion and Data and Applications at $373 million. Marking the diagnostics business at a generous two times revenue for a growing lab leaves roughly $6.2 billion of enterprise value against $373 million of data revenue, close to seventeen times. That is the actual bet in the stock: not that Tempus sequences tumors profitably, but that the data and modeling franchise compounds at software multiples while the lab funds customer acquisition.
Base case sits at $45 to $60. The deal closes, ADLT migration lifts diagnostics ASPs through 2027, growth stabilizes in the low twenties, and the multiple holds near five times forward revenue. Bull case is $70 to $85 and requires three things together: ADLT contracts negotiated well enough to push diagnostics gross margin toward 70%, Insights sustaining mid-thirties growth without further related-party concentration, and MRD volumes compounding off the Personalis base against Natera’s entrenched Signatera position. A short position at 26.5% of float sitting on a stock near its 52-week low supplies the fuel if those land. Bear case is $28 to $35 and does not require Tempus to stumble. It requires only cohort derating, healthcare-AI and diagnostics names compressing toward lab-services multiples. At three and a half times an estimated $2.0 billion of 2027 revenue, less $628 million of net debt, spread over roughly 220 million post-deal shares, the math lands near $28 without any change to the operating forecast.
One number decides more than the rest between now and closing. If the fifteen-day volume-weighted average price into the close sits below $46.00, Personalis can walk, and the entire MRD strategy resets to a partnership Tempus no longer controls.