Real Leverage at Last: Record EBITDA and Positive Operating Cash, Against a Multiple That Re-Rated 51% in Eleven Weeks

Results vs. Consensus

Q3 2026 Scorecard

A note on the consensus lines, because the two numbers that matter diverge sharply in quality. The revenue beat is 0.5% against the most widely cited estimate and only 0.05% against a second book that sat at roughly $411.3M. On either number the revenue result is inside the noise band for a company this size, and nobody should mistake it for the story. The EPS line is a different matter. The Street modelled a non-GAAP loss of $(0.09) and Tempus delivered non-GAAP income of $0.07, a $0.16 gap on a $411M revenue base. That is not a rounding difference. It is the Street carrying an operating-leverage assumption that the company beat by a wide margin for the second quarter running, after a $0.09 gap on the same line in Q2.

Note also that the GAAP and non-GAAP per-share figures point in opposite directions this quarter, and for a reason worth holding onto. GAAP diluted EPS is $(0.08) and non-GAAP EPS is $0.07. The bridge between them runs two ways at once: it removes $41.5M of fair-value gains that GAAP counts, and it adds back $50.8M of stock-based compensation and related payroll tax that GAAP also counts. Both adjustments are large, they roughly offset, and the stock-compensation add-back alone would flip the sign.

Year-Over-Year Comparisons

Quarter-Over-Quarter Comparisons

Nine Months 2026 vs. Nine Months 2025

The three reported quarters of 2026 sum exactly to the filed nine-month figures, so nothing in the quarterly tables above is a derivation: $348,116 plus $382,486 plus $411,537 equals the reported $1,142,139. The same arithmetic applied to Adjusted EBITDA puts first-half 2026 at $5,211, which is the number against which the back-half requirement is measured below.

Quality of Beat

• Revenue: Growth of 23.1% is organic in the sense that matters, with no acquisition closing in or around the quarter and the Ambry comparison now fully lapped. The composition improved rather than deteriorated for the first time in this coverage. Data and Applications accelerated to 32.9% from 28.0% in Q2, and within Diagnostics, oncology volume growth accelerated again to 32% from 31%. The offset is Hereditary at 3%, examined in detail below. A finer point that cuts in the company's favour: stripping related-party revenue, Data and Applications grew 71.4% rather than 32.9%, because the Pathos and SB Tempus contribution fell from 38.3% of the product line to 20.4%.
• Margins: The highest-quality element of the print, and better than Q2 because both product lines contributed. Consolidated gross margin reached 64.8%, up 200bp, on Diagnostics cost of revenue growing 15.6% against 20.0% revenue growth and Data and Applications cost of revenue growing 20.2% against 32.9% revenue growth. The 250bp of Data and Applications margin compression we flagged as the thing to watch in Q2 reversed into 300bp of expansion. Operating expenses grew 19.4% against revenue at 23.1%, which is the first quarter in this coverage where the operating cost base grew slower than the top line.
• EPS: Non-GAAP income per share of $0.07 against $(0.11) a year ago. Unlike Q2, where roughly half the improvement sat below the operating line, the bulk of this one is operating: non-GAAP income from operations swung $23.1M to a positive $14.4M. Interest expense fell a further 56.5% as the 0.0% convertible refinancing annualised, and there was no debt-extinguishment charge this quarter against $12.0M last year. Share count grew 3.6%, so the per-share figure is not being flattered by buybacks, of which there are none.
• Cash: The line that separates this quarter from the last one. Operations generated $9.2M against a $7.5M use in Q2, and the nine-month outflow narrowed to $71.6M from $181.3M. The caveats are that quarterly capital spending is not disclosed, and that nine-month purchases of property and equipment plus capitalised software consumed $35.9M of cash, roughly $12.0M a quarter on average, which makes genuinely positive free cash flow in the quarter unlikely on our estimate. Working capital also moved against the company: receivables rose to $380.7M from $311.2M at year-end, extending days sales outstanding to roughly 85 days from 78, and contract assets inside that balance nearly quintupled to $40.5M from $8.1M.

Product Line Performance

Tempus reports as a single operating segment under ASC 280, with the chief executive reviewing results on a consolidated basis. Diagnostics and Data and Applications are product lines rather than reportable segments, and the margin disclosures below come from the company's own non-GAAP reconciliation rather than a segment footnote.

Diagnostics

Diagnostics generated $303.5M on 20.0% growth and $189.4M of gross profit on 22.8% growth, lifting segment gross margin 140bp year over year to 62.4%. The sequential picture is less flattering and more interesting: margin slipped 20bp from 62.6% in Q2, and the reason is mix rather than cost discipline. Minimal residual disease testing is growing at triple digits off a low base, it carries a lower price than comprehensive genomic profiling, and it is now large enough to pull the blended number down.

The product line splits into two halves that continue to move in opposite directions, and the gap between them widened again this quarter.

Oncology

Oncology revenue of $172.6M grew 28% on 32% volume growth, running roughly 101,500 clinical tests. Volume growth accelerated for the second consecutive quarter, from 28% in Q1 to 31% in Q2 to 32% here, which is the single most important operating number in the company because oncology volume is the input to the dataset that the Data and Applications product line sells.

"In Oncology, we continued to see exceptionally strong and accelerating volume growth in Q3, running ~101,500 clinical Oncology tests representing 32% volume growth year-over-year. Every segment of our oncology portfolio is performing well, from solid tumor profiling to liquid biopsies for CGP and MRD monitoring." — Eric Lefkofsky, Founder and CEO

Two disclosures underneath that headline deserve more attention than they received. First, comprehensive genomic profiling volume across solid and liquid grew approximately 22%, not 32%. The ten-point difference is MRD, which ran roughly 12,000 tests, up 300% year over year. The headline oncology volume number is therefore being carried by the smallest and least profitable part of the portfolio. Second, and consequently, blended oncology reimbursement fell sequentially to roughly $1,700. Excluding MRD it rose, from $1,850 in Q2 to $1,880 in Q3, which is the number that will step up on January 1.

Assessment: The volume acceleration is genuine and it is the right thing to be accelerating, but investors should read the 32% and the $1,700 together rather than separately. Tempus is deliberately trading price for volume in MRD while it waits for reimbursement to mature, and that trade shows up as ASP and margin dilution today in exchange for installed base later. The January ADLT step is what makes the trade work. If that step lands as described, the ex-MRD ASP of $1,880 becomes the base for a repricing on volume the company already runs.

Hereditary

Hereditary revenue grew 3% to $111.3M on roughly 143,000 tests, with volume up 2% and average reimbursement of about $780 against $760 in Q2. This is the quarter's clear miss, and it is a miss against a specific commitment management made on the last call rather than against a Street number.

"We remain excited about the opportunity in hereditary profiling and anticipate increased growth rates that reach low to mid double digits by year end, which is a bit slower than we had expected last quarter. Our hope is to return that business to stable growth in 2027, and we’re seeing some early encouraging signs of that." — Jim Rogers, CFO

The Q2 framing was "mid-teens growth by the end of the year." The Q3 framing is "low to mid double digits," and the letter concedes the downgrade in the same sentence that delivers it. The supporting evidence offered has also changed character. Last quarter it was three weeks of GenomeNext launch data; this quarter it is order requisitions per day, up roughly 13% year over year in September and 12% in October. Requisitions are a reasonable leading indicator of units, but they are not revenue, and the gap between a 12% requisition trend and 3% reported revenue growth is wide enough that the bridge matters.

Management's explanation for the deceleration is unchanged and remains credible: 2025 included an abnormal share gain when a competitor hit financial difficulty, producing roughly 30% growth against a normal mid-teens rate, and that comparison is now being lapped. The forward case rests on the underpenetrated market, with reimbursement guidelines covering roughly 70 million patients against roughly 2 million tests run annually in the United States, and on the GenomeNext whole-genome product inside the Rare business, where whole-genome orders now represent nearly half of Rare and management expects triple-digit growth in 2027.

Assessment: Hereditary is now 36.7% of Diagnostics and growing at 3%. Strip it out and the rest of the product line grew roughly 32%, against 20.0% for Diagnostics as reported, which is the size of the anchor. We are marking the Q2 watch item as challenged rather than open. A target that moves from mid-teens to low-double-digits in one quarter, while revenue runs at 3%, is a forecast that has not yet found its floor. The encouraging element is that the requisition data is a real leading indicator rather than an anecdote, and two consecutive months of low-double-digit requisition growth is the first evidence of a turn. We would want a third.

Data and Applications

Data and Applications generated $108.1M on 32.9% growth, accelerating 500bp from Q2's 28.0%, with gross margin expanding 300bp to 71.5%. The acceleration came despite the company now lapping the full contribution of the AstraZeneca and Pathos foundation-model arrangement, which makes the comparison harder than the headline suggests.

The most underappreciated number in the release is not in the release at all. Related-party revenue inside this product line fell from $31.2M in Q3 2025, split between Pathos and the SB Tempus joint venture, to $22.1M this quarter, all of it Pathos. Excluding related parties, Data and Applications revenue grew from $50.2M to $86.0M, or 71.4%, and the related-party share of the product line fell from 38.3% to 20.4%. Revenue concentration in parties connected to the chief executive has been a standing concern in this coverage since the Q1 note. It materially improved this quarter, and nobody mentioned it.

"Our Data business, which we historically have referred to as Insights, is made up of data licensing (people who want our data for analytic purposes) and AI modeling (people who want our data to build their own models). The latter is increasingly becoming an important part of our business, where clients want access to our data connected to dedicated compute capacity (GPUs), so they can build models for their use that remain in our environment, or that they pay to egress." — Eric Lefkofsky, Founder and CEO

Insights

Insights, the data licensing and modelling business, grew 29%. That is the third consecutive quarterly deceleration, from 44.1% in Q1 to 36% in Q2 to 29% here, and it has now arrived exactly at the "close to 30% range plus" long-term algorithm management articulated on the Q2 call. The question for next quarter is whether 29% is the floor the algorithm describes or a waypoint on the way lower. Against the deceleration, data licensing bookings exceed $550M year to date, a figure that includes the October BMS and Pathos agreements and therefore runs past the quarter end, with most of the fourth quarter still to come.

Applications

The Applications product line contributed $9.7M against $3.2M a year ago, tripling off a small base. The underlying products are Next, a care-gap closure tool now deployed at 18 health systems and scanning roughly 900,000 cancer patients daily; TIME, the clinical trial matching network; and Algos, the algorithmic diagnostics suite, whose genomic algorithms carry a 45% attach rate across Tempus assays, unchanged from Q2. Two new FDA clearances landed in the quarter, for ECG-PH covering pulmonary hypertension and ECG-MR covering mitral regulation, joining existing clearances in atrial fibrillation and low ejection fraction.

Trials

Trials remains the drag inside the product line, and management was blunter about it than in prior quarters. The marketplace model for just-in-time trial matching was abandoned a year to eighteen months ago in favour of an enterprise structure where large pharma clients load most or all of their oncology trials into the network. Management says there is now a waiting list, and also that the business will not scale into a billion-dollar line. The contract research component is explicitly described as neither strategic nor an area of investment.

Assessment: This is the product line that justifies a technology multiple rather than a laboratory multiple, and it did three things this quarter that strengthen that claim: it accelerated against a harder comparison, it expanded margin by 300bp after compressing 250bp in Q2, and it grew its arm's-length revenue base at 71%. The caution we raised in Q2 about GPU provisioning structurally compressing margin was wrong for this quarter, and management now guides the full-year product-line margin to the mid-70s. The remaining watch item is Insights at 29%, which has converged on the stated algorithm with no margin for further deceleration before it falls through it.

Key Operating Metrics

Key Topics & Management Commentary

Overall Management Tone: Confident and noticeably shorter than prior quarters, with prepared remarks that ran a few minutes and a Q&A of nine analyst questions followed by three retail submissions. Management spent its airtime on 2027 and 2028 rather than defending the quarter, which it has done before, but this time the quarter did not need defending and the forward focus read as momentum rather than deflection. Analyst pushback was almost entirely absent: not one of the nine questions challenged a number, and the only question touching the balance sheet or the acquisition financing came from the retail queue.

1. The Operating Inflection Is Real This Time, and the Cash Statement Proves It

A quarter ago we wrote that Tempus had not earned a profit in any operating sense, and that the correct GAAP anchor was an operating loss that had widened 22.9%. This quarter that anchor moved the other way. The GAAP loss from operations narrowed 6.8% to $56.9M, the first year-over-year narrowing in 2026 after Q1 widened 23.3% and Q2 widened 22.9%. Non-GAAP income from operations turned positive at $14.4M against a $8.7M loss. Most importantly, operations generated cash.

"As expected, cash provided by operating activities improved significantly to $9.2 million in the quarter, compared to $(7.5) million in Q2 2026." — Jim Rogers, CFO

The mechanism is unglamorous and durable: gross profit grew $56.8M year over year while operating expenses grew $52.6M, and gross profit is growing faster than the cost base for the first time in this coverage. Interest expense fell a further 56.5% as the 0.0% convertible refinancing annualised, so the capital structure is no longer consuming the operating improvement on its way to the bottom line.

Assessment: This is the evidence Q2 lacked. Adjusted EBITDA can be engineered with add-backs; cash from operations cannot, at least not for long and not without a visible working-capital tell. The tell here is mild rather than absent, since receivables grew faster than revenue and contract assets quintupled, so we would not extrapolate $9.2M into a clean run-rate. But the direction is no longer in dispute.

2. The GAAP Improvement Is Still a Mark, and Now It Is a Mark on the Takeover Target

Net loss narrowed to $14.9M from $80.0M, an improvement of $65.1M. Of that, $42.8M is an unrealized gain on marketable equity securities, and the company's own letter identifies where the majority of it came from.

"Additionally, we experienced an unrealized gain on marketable securities of $42.8 million, a majority of which is associated with our ownership stakes in Personalis." — Jim Rogers, CFO

The portfolio consists of exactly two securities. Recursion closed the second quarter at $3.67 and the third at $4.13; against the $46.9M current balance that implies roughly 11.3M shares and a gain near $5.2M. The residual of roughly $37.6M, approximately 88% of the quarter's total, falls to a Personalis stake of about 13.0M shares as that stock moved from $13.40 to $16.28, which values back to the $212.3M carried in non-current marketable equity securities within a rounding difference. The arithmetic ties, and the company's qualitative statement agrees with it.

The circularity is the point. Personalis traded up to $16.28 because Tempus offered $16.25 a share for it. Tempus then booked the convergence as income. When the deal closes, that stake is eliminated in consolidation and the gain never becomes cash. Strip the fair-value line entirely and the quarter is a $56.3M net loss against a comparably adjusted $78.7M a year ago, which is still a $22.4M improvement and still the right direction. It is simply about a third of the improvement the headline advertises.

Assessment: This is the second consecutive quarter in which the most-quoted GAAP number is a securities mark, and the second consecutive quarter in which the underlying operating trend is better described by other lines. The difference is that in Q2 the ex-mark comparison went backwards and this quarter it goes forwards. Anyone modelling to the net-loss line should use the ex-mark figures, and should expect the Personalis contribution to reverse out of the balance sheet rather than into the bank.

3. Adjusted EBITDA of $25.6M, and the Q4 Bar That Fell from 3.7x to 1.34x

Adjusted EBITDA reached $25.6M against $1.5M a year ago. Nine-month Adjusted EBITDA is $30.8M, which means the first half contributed $5.2M and the third quarter contributed the rest.

"We generated $25.6 million in Adjusted EBITDA in Q3 and ended the quarter with $854.4 million in cash and marketable securities. We have 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." — Eric Lefkofsky, Founder and CEO

Graded against the specific bar we set a quarter ago, the result is a near miss that nonetheless transforms the problem. We said the back half needed roughly $59.8M, or about $29.9M a quarter. Q3 delivered $25.6M, falling $4.3M short of that pace. But the remaining fourth-quarter requirement to reach approximately $65M is $34.2M, which is 1.34 times the quarter just delivered, against the 3.7 times step-up we flagged in August. For context, Q4 2025 Adjusted EBITDA was $12.9M, so the implied fourth quarter also requires 165% year-over-year growth on that line.

Management hedged the number in the letter in a way it had not previously: "the word approximately implies a modest range." That is a small sentence doing meaningful work eight weeks before the year ends.

Assessment: We were too bearish in August. We modelled $50M to $60M of full-year Adjusted EBITDA against the approximately $65M guide, and Q3 makes the guide reachable. A 34% sequential step into a quarter that is seasonally the strongest, with the BMS contribution beginning to recognise ratably and Data and Applications accelerating, is a demanding but ordinary ask rather than the implausible one it was. We now carry $62M to $68M. The hedge in the letter is the reason we do not simply write $65M.

4. Stock-Based Compensation Falls Sequentially

Stock-based compensation was $48.1M in the quarter, down from $54.1M in Q2 and $52.7M in Q1, and down to 11.7% of revenue from 14.1%. Including related employer payroll tax the figure is $50.8M, or 12.3% of revenue. Year over year the comparison is still unflattering, with the expense up 41.5% against revenue up 23.1%, and the nine-month ratio stands at 13.6% of revenue against 8.8%.

This is the single largest reconciling item between the GAAP operating loss and positive Adjusted EBITDA, and a quarter ago we named it as the number we most wanted to see flatten before treating the margin story as complete. It did not merely flatten; it declined in absolute dollars, below both Q1 and Q2. Neither management nor any analyst mentioned it, for the second consecutive quarter.

Assessment: A $6.0M sequential decline in the largest add-back does not move Adjusted EBITDA, which adds stock compensation back in full. It shows up in the GAAP operating loss instead, where it supplied roughly a third of the $19.0M sequential narrowing. That is a legitimate improvement, not an accounting one, because less equity issued is less dilution, but it is worth separating from the operating gain. The structural test is whether 11.7% holds when revenue growth decelerates toward the company's own 25% algorithm, and one quarter does not settle that.

5. Hereditary Misses, and Management Walks Down Its Own Target

Treated in full in the product-line section above. The short version for a reader skipping ahead: revenue grew 3% on 2% volume growth, and the commitment to mid-teens growth by year end became low to mid double digits with an explicit acknowledgement that the forecast had slipped. The requisition data, up roughly 13% in September and 12% in October, is the only forward-looking support offered.

"So I think the overall hereditary business seems to be poised to do kind of much better as we move forward than it has over the last few quarters. In large part because it is both lapping excessive growth last year, which we have talked about relative to the some of the market dynamics where you had a big player in the market that was having some financial issues, a bunch of volume moved over. ... We have seen some good signs of that in September and October." — Eric Lefkofsky, Founder and CEO

Assessment: The lapping explanation is sound and we accept it. What we do not accept is the pattern: a specific numerical commitment made on one call, restated lower on the next, with the supporting evidence shifting from product launch data to requisition trends. Hereditary is now 36.7% of the larger product line, and the rest of Diagnostics grew roughly 32% against the 20.0% the product line reported. The 2027 case rests on the Rare business, where whole-genome orders are approaching half the sub-line and management expects triple-digit growth. That is a plausible path and it is not yet a demonstrated one.

6. Oncology Volume Accelerates to 32% While Blended Price Goes Backwards

Oncology volume growth accelerated for a second consecutive quarter while blended oncology reimbursement declined sequentially to roughly $1,700. Both facts have the same cause.

"Oncology average reimbursement was approximately $1,700 in the quarter, down slightly from Q2, the result of increased MRD volumes. Excluding MRD, reimbursement for Oncology ASP increased from $1,850 in Q2 to $1,880 in Q3." — Jim Rogers, CFO

Comprehensive genomic profiling volume grew approximately 22% across solid and liquid combined, so MRD supplies the gap to the 32% headline. Within CGP, management flagged the first divergence between the two modalities it has been willing to call out: September order requisitions for the xF liquid assay rose 25% year over year, and the company saw a 300 to 400 basis point uptick in liquid during the month.

"Up until now, our growth rates have been relatively consistent. Solid's been growing really strong as has liquid. We have historically called out that it is not meaningful to talk about the difference between the 2 of them. that is still the case for Q3. Our solid portfolio grew really quickly. We did see a sign here in September, which we called out that we saw this kind of uptick of, whatever, 300, 400 basis points. If that continues into Q4, you will start to see the first signs of really between liquid and solid where even with really strong growth rates in solid, liquid's outperforming." — Eric Lefkofsky, Founder and CEO

Assessment: The mix shift toward liquid matters more than it sounds, because xF is the assay awaiting FDA approval and therefore the one carrying the larger of the two pending price steps. Volume accumulating in liquid ahead of an approval that reprices liquid is the right sequencing if the approval arrives. If it does not, Tempus will have built share in its lowest-priced CGP modality. The honest read on the blended $1,700 is that it is a deliberate, disclosed and temporary cost of building the MRD and liquid franchises.

7. The January 1 ADLT Step, Trimmed from $85M to $80M

The tumor-only xT CDx approval received earlier this year allows Tempus to migrate the remainder of its solid-tumor DNA portfolio to Advanced Diagnostic Laboratory Test pricing. The conversion completes by the end of this year and the pricing takes effect on a date, not a guess.

"And then on the oncology side, as we have talked about, you know, given the expansion of our FDA approval to include tumor-only for XT test, 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. Then we still have XF kind of going through the FDA process" — Jim Rogers, CFO

The figure given in Q2 for the same item was approximately $85M. It is now approximately $80M, a quiet $5M reduction that nobody asked about. Beyond xT, the letter says the company is now working on a premarket approval submission for xR, its whole-transcriptome RNA sequencing panel, alongside the pending xF filing. Management sizes the combined xT and xF path at roughly $750 of incremental ASP over several years, taking comprehensive genomic profiling reimbursement to parity with competitors "in the $2,600 range."

Assessment: This remains the highest-confidence item in the 2027 bridge and the only one with a date attached. On roughly 101,500 quarterly oncology tests, $80M across a year is real money arriving at close to full incremental margin on volume already being run. Two cautions. The figure has now moved once, downward, without explanation, which is a reminder that it is an estimate rather than a rate card. And an uplift that begins on January 1 contributes nothing to the quarter the stock is currently being valued against.

8. Data and Applications Reaccelerates, and the Related-Party Crutch Falls Away

Treated in the product-line section above. The two numbers worth repeating here are that the product line accelerated 500 basis points to 32.9% while lapping the full AstraZeneca and Pathos foundation-model contribution, and that arm's-length revenue inside it grew 71.4% as related-party revenue fell from 38.3% of the line to 20.4%.

The stickiness argument management made is the most substantive strategic disclosure of the call. Clients no longer download files; they sign up for three-year commitments of dedicated GPU capacity inside the Tempus environment and pay an egress fee to remove models they build there.

"But now we are connecting Lens to, you know, 8 GPUs, 16 GPUs, 32 GPUs, some kind of cluster like that. Typically denominations of 8. Where clients are signing up for, like, 3 years of 32 GPUs. Or 3 years of 64 GPUs. ... And so it is gone from, like, not sticky to crazy sticky ..." — Eric Lefkofsky, Founder and CEO

Assessment: Multi-year GPU commitments with egress friction are a materially better revenue structure than episodic file licensing, and they explain both the margin behaviour and the bookings. They also explain the Q2 margin compression we worried about, which turns out to have been provisioning costs landing ahead of the revenue rather than a structural change. The falling related-party concentration is the quieter and more important development: a data business growing 71% at arm's length is a different asset from one growing 33% with a third of it coming from a company the chief executive chairs.

9. BMS and the First Large Dataset Outside Oncology

The quarter's largest commercial announcement came after it ended. In October, Tempus signed a direct licence with Bristol Myers Squibb worth up to $137.5M over five years, and a second arrangement under which Pathos pays Tempus $55M over five years to build an Alzheimer's and late-stage dementia dataset for BMS. Together the two are worth $192.5M.

"As I have said before, if we signed just 1 of these big collaborations, we would be content. We have signed 3 this year alone." — Eric Lefkofsky, Founder and CEO

The letter quantifies the revenue effect precisely, which is unusual and welcome: approximately $36M a year recognised largely ratably over five years, excluding any cloud or compute purchases BMS may make on top. Against a Data and Applications run-rate of roughly $432M annualised, that is about eight points of product-line growth arriving under contract.

Assessment: The strategic significance exceeds the dollars. This is the first large commitment to a Tempus dataset outside oncology, and it validates a claim the company has made for several quarters without evidence: that more than 80% of the 50 million patient records in the platform sit outside cancer, and that the data has value there. Note the structure, though. The Alzheimer's half flows through Pathos, the related party whose executive chairman is Tempus's chief executive, which is precisely the arrangement whose declining share we just praised. It is disclosed and it is not improper, but it is worth tracking rather than assuming away.

10. Personalis: Cleared, Scheduled, and Still Unfinanced

The transaction advanced on every axis except the one that matters to shareholders. Hart-Scott-Rodino has expired, UK Competition and Markets Authority clearance is obtained, the S-4 was declared effective on October 21, the shareholder vote is set for November 30, and closing is expected in December. A reciprocal termination fee of approximately $76.8M is in place, and $3.7M of transaction costs ran through the quarter.

What has not advanced is the financing, and in three months the language has not moved forward.

"We've begun discussions with parties on a potential debt facility, as our intention is to finance a large portion of the consideration in cash to minimize shareholder dilution. Our ultimate decision will obviously depend on our stock price, but we are putting the pieces in place to ensure maximum flexibility." — Jim Rogers, CFO

In July the phrasing was "we have already begun working with parties to put a debt facility in place." In November it is "we've begun discussions with parties on a potential debt facility." No size, no rate, no structure, no counterparty, with the vote under four weeks away. The 10-Q discloses nothing further.

The share price has done the company a considerable favour here. At roughly $1.5B of consideration, an all-stock transaction at the November 4 close of $75.61 issues about 19.8M shares, or 10.9% dilution against 182.2M outstanding. At the $49.36 level that anchored our August note it would have issued about 30.4M shares, or 16.7%. The share-price floor that was a live termination risk in July is no longer in play. The cash election is the constraint that remains: taking the full 50% would require roughly $750M against $590.5M of actual cash on the balance sheet, which is why the facility exists and why its absence is conspicuous.

On what the deal is for, management was clear and usefully specific about the timing of the economics. Only about 15% of the sales force currently sells MRD, on the figure an analyst put to management, and the gate is reimbursement rather than the closing.

"I think it is going to be largely a function of... reimbursement than it is a function of closing the deal. We have to be sensitive that at the present moment when Personalis runs these tests, they lose money. And once we acquire them, we will be in the same boat." — Eric Lefkofsky, Founder and CEO

Assessment: Our August position was that we would not underwrite this transaction until the debt facility was disclosed, and nothing has happened to change that. The deal risk has genuinely fallen, because regulatory clearance is obtained and the dilution arithmetic improved by roughly a third with the share price. The financing risk has not fallen at all. A company closing a $1.5B acquisition in December that has not disclosed terms for the facility funding it in November is carrying an avoidable question into the close. One useful detail for modellers: Personalis was paying Tempus roughly $400 per test under the existing reseller arrangement, and that intercompany revenue is eliminated on consolidation, so reported revenue gains less than Personalis's standalone revenue would suggest.

11. The Capitalized-Software Line Nobody Mentioned

Capitalized software, net, on the balance sheet stands at $40.6M against $6.1M at year-end 2025. The cash flow statement shows $12.3M of capitalised software costs paid in the nine months, and the supplemental non-cash disclosure shows a further $21.5M accrued but not yet paid, for roughly $33.8M of additions. The comparable figure for the first nine months of 2025 was $4.6M, with nothing accrued.

Costs that are capitalised do not appear in operating expense. The subsequent amortisation does appear, and Adjusted EBITDA then adds it back. The effect runs in the company's favour twice. For scale, nine-month Adjusted EBITDA improved $51.1M year over year, and the increase in capitalised software additions over the same comparison is roughly $29.2M, or about 57% of that improvement.

Assessment: We are not calling this aggressive accounting, and we have no basis to. Internal-use software capitalisation is routine, Tempus is visibly building platform infrastructure, and the change is fully disclosed in the filings. What we will say is that a seven-fold increase in capitalisation intensity, in the same nine months that the company crossed into positive Adjusted EBITDA, is a conspicuous coincidence that went entirely unremarked in the release, the shareholder letter and the call. It belongs in any model that treats the EBITDA inflection as a clean operating signal.

Guidance & Outlook

"We are increasing guidance to $1.60 to $1.605 billion in 2026. We continue to expect 2026 Adjusted EBITDA to be approximately $65 million. As always, given the unique nature of our business, it’s difficult to predict these numbers with complete accuracy; as such, the word approximately implies a modest range." — Jim Rogers, CFO

Implied fourth-quarter ramp. Nine-month revenue is $1,142.1M, so the guide implies a fourth quarter of $457.9M to $462.9M, with a midpoint of $460.4M. That is 24.7% to 26.0% year-over-year growth, or 25.4% at the midpoint, and it requires 11.9% sequential growth against the 7.6% just delivered. For reference, the seasonal pattern is close: Q4 2025 grew 9.9% over Q3 2025 and Q4 2024 grew 10.9% over Q3 2024. The guide therefore asks for about one and a half points more than the two-year seasonal norm, which the BMS recognition beginning in the quarter and a Data and Applications line accelerating 500 basis points should supply.

The EBITDA line is now the easier of the two. Nine-month Adjusted EBITDA of $30.8M against an approximately $65M target leaves $34.2M for the fourth quarter. That is 1.34 times the $25.6M just delivered and 165% above the $12.9M of Q4 2025. A quarter ago the same calculation produced 3.7 times. The bar did not move; the company moved under it.

Street at: consensus sat at $409.4M for the quarter and the company delivered $411.5M, so the $2.5M midpoint raise brings the full-year guide roughly into line with where models already sat rather than above them. The guidance change is not where the Street's revision is coming from. The EPS line is, and it moved $0.16.

Guidance style. Three data points this year, and the pattern is unmistakable. The February guide was approximately $1.59B. May took it to $1.590 to $1.600B. July took it to $1.595 to $1.605B. November takes it to $1.600 to $1.605B. Each move is a raise, each is small, and the cumulative effect is that the high end has not changed since July despite two beats. This is a management team that raises the floor and leaves the ceiling alone. On that pattern the $1.605B high end is comfortably achievable and the company will probably print just above it.

One omission worth flagging in the guidance itself. The July release stated that guidance assumed no impact from the Personalis transaction. The November release contains no such sentence, while the letter says closing is expected in December. Either a partial-quarter contribution is now contemplated inside the range, or the carve-out simply was not restated. Nobody asked, and the difference is material to how the fourth-quarter revenue figure should be read.

Analyst Q&A Highlights

Whether 2027 Reimbursement Tailwinds Survive Fee-Schedule Pressure

The opening question of the call paired the hereditary outlook with proposed clinician laboratory fee-schedule rates, asking whether pricing pressure on one side offsets the ADLT tailwind on the other. The answer split cleanly: the chief executive took the volume half and attributed hereditary's weakness to lapping, while the finance chief took the pricing half and quantified the 2027 benefit for the first time since the figure was revised.

Q: "So 1 question on Hereditary. You are hoping that return to stable growth next year. But just given these proposed clinician lab fee schedule rates, that [Inaudible], I am just wondering how you are thinking about potential headwinds there. Given you then obviously have ASP tailwinds on the oncology side. So maybe just talk about diagnostics outlook for 2027."
— Kyle Mikson, Canaccord Genuity

A: "So on ASPs for hereditary as a reminder, obviously, cancer makes up the majority of that revenue, but we do have a small but growing rare business. ... And then on the oncology side, as we have talked about, you know, given the expansion of our FDA approval to include tumor-only for XT test, 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."
— Jim Rogers, CFO

Assessment: The pricing answer is the substantive one and it contains the quarter's quiet downward revision, from approximately $85M to approximately $80M, delivered without being flagged and without a follow-up. The fee-schedule concern was dismissed rather than addressed with numbers. On a call where no question challenged a figure, this was the closest anyone came to one, and the revision it surfaced went unexamined.

Whether the Margin Inflection Drops Through or Gets Reinvested

The most consequential exchange of the call, and the only one that engaged directly with the economics the stock just re-rated on. The question laid out the published consensus margin path for the next two years and asked how much of the coming oncology price increase would be allowed to reach the bottom line. The response restated the company's standing reinvestment ratio and declined to endorse the consensus numbers, in either direction.

Q: "Maybe just on EBITDA. Nice print. Consensus has margins expanding next year, I think, to 7% and then 11% in 2028 And there is always just push and pull for diagnostic companies how much you reinvest back in the business versus let it drop through. ... I know you have the $2.6 thousand realized oncology price that you also pointed to in the next few years to dream meaningfully above what we have today. So I am just wondering how much of that increase in oncology pricing you guys would be inclined to drop through to the bottom line."
— Daniel Brennan, TD Cowen

A: "So, you know, I think our you know, view on EBITDA kind of remains the same. You know, we are we are still relatively early days on a lot of the offerings. And so we have had this perspective of kind of reinvesting 2/3 or roughly 2 thirds of any incremental gross profit dollars that we generate back in the business and then letting 1/3 drop down to the bottom. Bottom line. So, you know, obviously, with Personalis is coming on board, we are already going to invest in MRD. We will utilize some of that reinvestment dollars to absorb the burn from personalis. But then after another couple years, then that flips, right? Then you are you are generating, you know, meaningful gross profit dollars such that you really only need to invest 1/3, and kind of 2/3 drop down to the bottom of it. ... But we are not in a position where we wanna maximize EBITDA in the short term."
— Jim Rogers, CFO

Assessment: The one-third rule is the most useful forward number management gives, and it sets a near-term ceiling that the market may not be applying. If roughly $80M of ASP uplift arrives in 2027 at close to full incremental margin, the rule implies roughly $27M reaches Adjusted EBITDA and the rest is spent, before the acquired MRD losses are absorbed. Management also said explicitly that it is not trying to maximise near-term EBITDA, and said the ratio flips to two-thirds drop-through only "after another couple years" from here. A model extrapolating this quarter's 6.2% Adjusted EBITDA margin into a steep curve before then is working against a stated policy.

When the Acquired MRD Sales Force Gets Unshackled

A recurring theme across two separate questions concerned the gap between MRD's volume growth and the small fraction of the sales organisation permitted to sell it. The response moved the gate from the transaction closing, where the questioner placed it, to reimbursement economics, and attached a loose and candidly uncertain timeline.

Q: "Nice continued growth in MRD. You know, you are tracking at about up 3 thousand units per quarter. Recognizing that only 15 of your Salesforce is selling today, and the shareholder meeting is sort of around the corner after Thanksgiving, you just give us a sense for once Personalis is in your portfolio to what extent you plan to ungate and let the remaining 85% of reps sell the test. Is that something that we could expect in the first half of 27?"
— Mark Massaro, BTIG

A: "I think it is going to be largely a function of it is more a function of reimbursement than it is a function of closing the deal. We have to be sensitive that at the present moment when Personalis runs these tests, they lose money. And once we acquire them, we will be in the same boat. ... But the big unlock, which I suspect will be I do not know, could be midyear. It could be Q2. It could be Q3. Who knows? But somewhere, I suspect, midyear, you will be in this boat where they will turn the corner and be kind of positive gross profit dollar contribution on a per test basis."
— Eric Lefkofsky, Founder and CEO

Assessment: An honest answer that pushes the MRD revenue inflection out to the back half of 2027 at the earliest. The discipline is correct, and it is the same discipline described last quarter. What changed is the admission that the acquired business loses money per test and will continue to do so inside Tempus until around the middle of 2027, a drag on the 2027 EBITDA path that management says it can absorb while still improving Adjusted EBITDA but has not quantified.

Whether the Long-Term Growth Algorithm Is 25% or Something Higher

A question about 2027 construction, built from component assumptions on CGP volume and data growth, invited management to endorse a higher long-term rate. It declined, and did so in a way that reads as a deliberate expectation reset rather than modesty.

Q: "So can we assume it is CGP volume holds in the low 20s and D&A might be closer to 30% than 25%. You have still got the incremental gross profit, 2 thirds of it, even after Personalis being reinvested Can we get to the 25% long term range? And thinking out further once we get the benefit of XT and XF uplift 2028 be a year above that 25% CAGR."
— Parth Talsania, Cantor Fitzgerald

A: "so first of all, we feel like a business of our size growing at 25% a year over a long term is incredibly healthy. So we have always said we are more interested in long term 25% growth than we are in kind of you know, short term 30% growth that then decelerates. ... Is it possible that you get into 2027 or 2028 and have moments where the business is growing significantly faster either because of ASP lift or because the data business just is running really hot. Yes. ... But, you know, at the end of the day, if you can sustain kind of 25% growth over a decade or another decade, will build a really large business."
— Eric Lefkofsky, Founder and CEO

Assessment: This is the second consecutive quarter in which management has declined to endorse growth above its stated long-term rate, after framing the data business at "close to 30% range plus" in Q2. Reported growth is 23.1% and the algorithm is 25%, so the gap has closed from the other direction and the company is now growing at roughly its own stated durable rate. Anyone paying nine times forward revenue is paying for the moments management says may happen rather than the rate management says it is managing to.

What Actually Makes the Data Contracts Sticky

A two-part question on customer stickiness and the share of Insights customers using the AI modelling capability produced a long and genuinely informative answer to the first part, and no answer at all to the second. The description of how the business migrated from file delivery to in-environment model building is the clearest articulation of the moat management has given.

Q: "So given this strength, think you mentioned in the AI modeling portion of the Insights business, how does this change the stickiness of the relationships with your customers? And then maybe touch on what percentage of the insights data customers also utilize AI modeling and then how you expect this to evolve into next year?"
— Ricki Levitus, Goldman Sachs

A: "Over time, we basically said, look, we built this very large environment. You know, we have something like 700 software engineers that have been working here for almost a decade. So we built this very large multimodal technology environment to interrogate this kind of data. We started saying to our clients, why do not you do that work in our environment? ... Where clients are signing up for, like, 3 years of 32 GPUs. Or 3 years of 64 GPUs. ... And so it is gone from, like, not sticky to crazy sticky ..."
— Eric Lefkofsky, Founder and CEO

Assessment: The architecture argument is persuasive and it is corroborated by the numbers, with 300 basis points of margin expansion in the product line and three large strategic collaborations signed this year. The unanswered half is the one that would let an outsider verify it. A penetration rate for AI modelling within the Insights base would convert "crazy sticky" from a characterisation into a metric, and it was requested directly and not supplied.

What Is Driving the Liquid-Biopsy Requisition Surge

A question about the 25% year-over-year increase in liquid-assay order requisitions drew the most forward-looking operational answer of the call, including the first explicit signal that the two comprehensive genomic profiling modalities are beginning to separate.

Q: "In your letter, you specifically call out strength in excess of order requisitions being up 25% year over year. Could you give us any additional color on what is driving strength? Is there any specific indication or academic setting, community setting EMR integration, any reason?"
— Subbu Nambi, Guggenheim

A: "And we published on this as well, but it can be 10 to 20 percent of mutations that would be therapeutically relevant that you will not find if you only do solid tumor profiling or you only do liquid biopsies. So you have this movement to kind of do both more often for later stage patients that are at high risk of mortality. ... We did see a sign here in September, which we called out that we saw this kind of uptick of, whatever, 300, 400 basis points. If that continues into Q4, you will start to see the first signs of really between liquid and solid where even with really strong growth rates in solid, liquid's outperforming."
— Eric Lefkofsky, Founder and CEO

Assessment: Concurrent and serial testing is a secular tailwind that benefits the whole category rather than Tempus specifically, and management said as much by name-checking a competitor's strong unit volumes. The company-specific read is that volume is accumulating in the modality awaiting FDA approval, which is favourable sequencing if the approval arrives and a lower-priced mix shift if it does not. One month of a 300 to 400 basis point divergence is a signal, not a trend.

Whether the Balance Sheet Funds the Plan if AI Sentiment Turns

The only question of the entire call touching the capital structure or the acquisition financing came from the retail submission queue rather than from any of the nine covering analysts who spoke. It asked directly whether operating cash flow and existing liquidity can fund both the growth plan and the Personalis acquisition if investor appetite for AI-linked companies weakens.

Q: "if investor appetite for AI related companies weakens, how confident are you that operating cash flow in existing liquidity can fund your growth plans and the Personalis acquisition?"
— Retail investor submission via the Robinhood platform

A: "I am super confident. ... I mean, I would say that the nice thing, in part, of the reason our business model is different is we are not overly reliant on 1 product. ... We are entering this period of kind of increasing ASPs pretty dramatically. You know, almost $750 over the next couple years. And a lot of those dollars can be used to kind of fund investments, you know, elsewhere. ... And lastly, as we mentioned on kind of previous calls, you know, we are we are reinvesting, you know, heavily today. And, certainly, those investments are meant to drive long term growth. So you can always scale that level of investment as necessary as well."
— Eric Lefkofsky, Founder and CEO; Jim Rogers, CFO

Assessment: The answer is reasonable and it does not address the question that was asked. The funding risk is not whether a growing business can support itself over several years; it is whether a specific December closing can be funded on the terms management prefers without a facility that has no disclosed terms in November. That the question came from a retail investor and not from any of nine sell-side analysts on a call four weeks before the shareholder vote is itself the finding. The sell side asked about 2027 growth rates, MRD sales coverage and trial matching. Nobody asked about the debt.

What They're NOT Saying

1. The debt facility still has no size, rate, structure or counterparty. The language moved from "already begun working with parties to put a debt facility in place" in July to "begun discussions with parties on a potential debt facility" in November, which if anything reads as less advanced. The quarterly filing adds nothing. The shareholder vote is November 30 and closing is expected in December.
2. The guidance carve-out for Personalis was dropped without comment. The July release said guidance assumed no impact from the transaction. The November release says nothing on the subject while the letter says closing is expected in December. The fourth-quarter revenue figure cannot be read cleanly without knowing which it is.
3. Capitalised software additions rose roughly seven-fold and nobody mentioned it. Nine-month additions of approximately $33.8M, being $12.3M paid and $21.5M accrued, against $4.6M a year ago. The costs leave operating expense and the later amortisation is added back in Adjusted EBITDA. Absent from the release, the letter and the call.
4. Total Remaining Contract Value and Net Revenue Retention have not been updated since the FY2025 print. Both were disclosed at year-end 2025, at over $1.1B and 126% respectively. The Q2 release carried the definitions without values. The Q3 release drops the "Other Key Metrics" section entirely. With more than $550M of bookings claimed year to date, an updated contract-value figure is the single cleanest proof available, and it is the one not given.
5. The free-cash-flow commitment was never graded. On the Q2 call management said the refinancing would enable "positive free cash flow by year-end." Operating cash flow was discussed this quarter and free cash flow was not, quarterly capital spending is not disclosed, and nobody asked whether the commitment still stands.
6. No 2027 revenue or EBITDA figure was offered, only a direction. Management referenced 2027 repeatedly, quantified an $80M ASP uplift and a $750 multi-year ASP path, and committed to "continued improvement in Adjusted EBITDA and cash from operations" without placing any of it against a base. When handed published consensus margins of 7% and 11% for 2027 and 2028, management neither endorsed nor corrected them.
7. Roughly $19.6M of Diagnostics revenue is unreconciled. The letter describes Diagnostics as having "two main components: Oncology and Hereditary" and gives them as $172.6M and $111.3M, which sum to $283.9M against a $303.5M product line. The difference is 6.5% of the segment and is not identified.
8. Working-capital quality deteriorated quietly. Receivables rose to $380.7M from $311.2M at year-end, extending days sales outstanding to roughly 85 from 78, and contract assets inside that balance rose to $40.5M from $8.1M. In the quarter that operating cash flow turned positive, revenue recognised ahead of billing rights nearly quintupled.
9. Insights deceleration went unexplained for a second consecutive quarter. Growth fell from 44.1% to 36% to 29% across three quarters, while data licensing bookings for the year passed $550M. No analyst asked for the bridge and management did not volunteer one.

Market Reaction

• Pre-print setup: The stock closed at $72.22 on October 30, entering the print up 22.3% year to date against an S&P 500 up 10.7%, but down 19.6% over twelve months and down 5.8% over the trailing 30 days. The 52-week closing range was $41.55 to $89.85, so the print arrived with the shares in the upper half of a wide band and recovering from a weak month.
• Reaction session, November 2: Results and the shareholder letter came before the open, with the call at 8:45 AM ET. Shares opened at $74.43, a 3.1% gap, traded a $70.30 to $83.42 range and closed at $78.48, up 8.7%. Volume was 17.0M shares against a 6.5M 30-day average, or 2.6 times.
• Peer and benchmark context: The S&P 500 fell 0.5% on the session. The three closest listed comparables all fell, by 4.0% to 8.5%. Tempus was the only one of the four to rise, and it rose by more than any of them fell.
• Follow-through: Shares closed at $77.80 on November 3, down 0.9%, and $75.61 on November 4, down 2.8%. Net of the full move, the stock is up 4.7% from the pre-print close, having given back 46% of the reaction-day gain over two sessions on declining volume.

The intraday shape of the reaction session is the most informative part. Shares opened 3.1% higher, reached as much as 15.5% above the pre-print close, then surrendered more than a third of the move before the close. The release and the letter came out before the open and the call began at 8:45 AM ET, so the opening gap already reflected the written disclosure and at least part of the call. The daily price data does not show when in the session the high was set. What the tape then did, fading from $83.42 to $78.48 and onward to $75.61 over two more sessions, says the marginal buyer was not willing to pay the top of the range for it.

The divergence from the diagnostics complex deserves a flag rather than a theory. On a day the S&P 500 fell and three comparable names fell 4% to 8.5%, Tempus gained 8.7% on 2.6 times normal volume. That is an idiosyncratic response to the company's own disclosure, not a sector rotation, and it is the cleanest evidence available that the print genuinely changed minds. Several published price-target increases followed over the next two sessions, including one upgrade out of the most negative rating on the Street, whose new target nonetheless sits below where the shares were already trading. That combination, a capitulating bear setting a target under the market price, is usually a sign that positioning rather than valuation drove the move.

For completeness, and because it is material to anyone reading this on the day of publication: the shares are trading sharply lower in the current November 5 session, quoted near $67.20 in early-afternoon dealing on roughly 4.8 million shares, against a broad market down about 4% on the day. That is an incomplete session, no company disclosure accompanies it, and no valuation figure in this note is anchored to it. Every level and multiple in the Model Update section below uses the November 4 close of $75.61.

Street Perspective

Debate: Is Q3 the Start of a Margin Curve or the Best Quarter of the Cycle?

Bull view: The inflection showed up simultaneously in four independent places, which is hard to fake: a GAAP operating loss that narrowed, positive non-GAAP operating income, positive cash from operations, and stock compensation falling in absolute dollars. Gross profit grew $56.8M while operating expenses grew $52.6M, and that gap widens mechanically as the January ASP step lands on a cost base that does not move with price.

Bear view: Adjusted EBITDA still excludes $48.1M of quarterly stock compensation, up 41.5% year over year, roughly 57% of the nine-month improvement is matched by an unexplained seven-fold increase in software capitalisation, and management stated plainly that it intends to reinvest two-thirds of incremental gross profit and is "not in a position where we wanna maximize EBITDA in the short term." The acquired MRD business loses money per test and will be absorbed in December.

Our take: The bulls have the better of this one and the bears have the better of the magnitude. The inflection is real, because cash from operations is not an adjusted metric and it turned. But the slope implied by a 51% multiple expansion is steeper than the company's own near-term reinvestment policy permits, and management places the flip to two-thirds drop-through a couple of years out. A one-third drop-through rule applied to roughly $80M of 2027 ASP uplift implies about $27M of incremental Adjusted EBITDA before Personalis losses, which is a fine outcome and not a software margin curve.

Debate: Does the January ADLT Step Justify a Multiple That Already Moved?

Bull view: The approval is granted, the conversion completes this year, the date is January 1, and the uplift lands on roughly 101,500 quarterly tests already being run at close to full incremental margin. That is growth requiring no new customers, no new capacity and no new approval, and it is eight weeks away rather than two years away. A second and larger step follows if the pending liquid-biopsy filing clears, with a whole-transcriptome submission now also in preparation.

Bear view: The uplift estimate has already moved once, downward from approximately $85M to approximately $80M, with no explanation offered and none requested. Against a revenue base guided to roughly $1.6B, $80M is five points of growth, which the shares have already paid for several times over in a move from 5.9 times forward revenue to 8.9 times on a $2.5M increase in the full-year midpoint. The liquid-biopsy approval remains unapproved and its price assumption was self-revised 139% in a single quarter earlier this year.

Our take: We carry the $80M and treat the liquid and transcriptome steps as optionality, which is unchanged from our August position except for the $5M trim. The disagreement is not about whether the uplift is real. It is about what fraction of it is still unowned at the current price, and on our arithmetic the answer is most of it was unowned at $49 and much less of it is unowned at $76.

Debate: Is Personalis a TAM Expansion or a Financing Event?

Bull view: Regulatory clearance is complete on both sides of the Atlantic, the registration statement is effective, the vote is scheduled and the share-price rally has cut maximum dilution from roughly 17% to roughly 11%. Tempus has distributed the asset since 2023, so integration risk is minimal, and owning it converts a partner margin into an internal one in a market management sizes above $20B and describes as one of the fastest-growing segments in oncology diagnostics.

Bear view: Three months after saying it had begun arranging a facility, the company has disclosed no size, no rate, no structure and no counterparty, with the closing expected within weeks. Taking the full cash election would require roughly $750M against $590M of actual cash. The acquired business loses money on every test it runs, management expects that to persist until the middle of 2027, and no guidance quantifies the drag.

Our take: Both are true and they resolve on different timelines. The strategic case has strengthened: MRD volume rose 300%, reimbursement widened to breast, non-small-cell lung and immuno-oncology monitoring, and the asset is visibly closer to per-test breakeven than it was in July. The financing case has not moved at all. We said in August that we would not underwrite the transaction until the facility was disclosed, and that remains our position with the closing weeks away. That the only question about it came from a retail investor suggests the Street has simply stopped asking.

Model Update & Valuation Framework

Valuation, anchored to the November 4, 2026 close of $75.61. With 177,182,062 Class A and 5,043,789 Class B shares outstanding at September 30, the equity value is approximately $13.78B. Debt is $1,174.9M of convertible senior notes net plus $176.0M of convertible promissory note, or $1,350.8M, against $854.4M of cash, marketable securities and restricted cash, for net debt of roughly $496.5M. Enterprise value is therefore approximately $14.27B, or 8.9 times the $1,602.5M midpoint of FY2026 revenue guidance and 11.2 times trailing FY2025 revenue. Enterprise value to FY2026 Adjusted EBITDA at an approximately $65M target is not a meaningful figure and we do not present one.

One adjustment to that net-debt calculation matters. Of the $854.4M, $259.2M is marketable equity securities and $212.3M of that is the Personalis stake, which is deal consideration rather than a source of funds; it extinguishes on closing. Counting only genuinely liquid resources of $595.2M, net debt is $755.6M and enterprise value is $14.53B, or 9.1 times the guidance midpoint. Either figure is a full multiple.

What has actually changed since August. Our Q2 note put enterprise value at $9.44B and 5.9 times forward revenue against a $49.36 share price. The same calculation today gives 8.9 times, a 51% expansion in eleven weeks. Over that period the full-year revenue guidance midpoint rose by $2.5M, or 0.16%, and the full-year Adjusted EBITDA guidance did not move at all. The re-rating is therefore almost entirely a change in what the market will pay for the 2027 and 2028 story, not a change in the 2026 numbers. Some of that is deserved, because the quarter removed real doubt about whether the operating leverage exists. The question a buyer faces now is narrower than it was: not whether Tempus can inflect, but whether nine times forward revenue for a 25% grower that plans to let only a third of incremental gross profit through for the next couple of years leaves anything on the table.

Valuation impact. We are maintaining coverage without a formal price target, for the same reason given in August and with the reason now dated. The two variables that would set one are the Personalis financing mix and the pending liquid-biopsy determination. The first resolves at or shortly after the December closing. The ADLT step resolves on January 1. Both fall inside the next reporting period, which makes the Q4 print the first occasion on which a target can be set against known rather than assumed inputs. We would move to Outperform on a combination of the facility being disclosed on non-dilutive terms, a Q4 Adjusted EBITDA print at or above $34M, and a Hereditary exit rate that reaches the low double digits management has now committed to. We would move to Underperform if the cash election forces an equity raise, or if the January repricing slips.

Thesis Scorecard Post-Earnings

The pillars below are the ones carried in the standing thesis since initiation, graded against what this quarter's print and call revealed. Status tags that moved are marked.

Overall: Thesis strengthened. Two of three bear points moved from emerging to contained on hard evidence rather than on management assurance, and all three bull pillars were confirmed with the data product line beating our own bearish margin call. The two things that did not improve are the two that were always going to require someone other than the operating business to fix: the acquisition financing, which needs a lender, and Hereditary, which needs a market. Set against that, the price moved 53% while the 2026 numbers moved 0.16%.

Action: Hold. This is a valuation call rather than a business call, and the distinction matters. Owners should not sell into an inflection that has just been demonstrated on the cash statement with a dated repricing eight weeks away. New capital is being asked to pay nine times forward revenue for a company whose own long-term algorithm is 25% growth, with only a third of incremental gross profit reaching the bottom line for the next couple of years, and with the acquisition financing undisclosed under four weeks before the vote. We raise conviction to 6 from 5 and would rather own this at a lower multiple or with the facility disclosed, and both of those questions answer themselves before the next print.