The First Profitable Quarter Tempus Didn't Earn: $8M of EBITDA, a $98M Mark, and a Back-Half That Has to Do 92% of the Work

Results vs. Consensus

Q2 2026 Scorecard

A note on the EPS line, because it is easy to get wrong. Tempus printed positive GAAP EPS of $0.03 and a negative non-GAAP loss per share of $(0.04). The two move in opposite directions because the non-GAAP bridge removes $97.4M of fair-value gains that GAAP includes. Street models were built to the non-GAAP line, and that is the line the beat is measured against. No provider published a GAAP EPS consensus for the quarter. Consensus for the non-GAAP figure was not tightly held either: the most widely cited estimate sat at $(0.13), a second provider at $(0.14), and a third at $(0.08). We use $(0.13) and note that the beat is between $0.04 and $0.10 depending on whose book you were reading.

Year-Over-Year Comparisons

Quarter-Over-Quarter Comparisons

First-Half 2026 vs. First-Half 2025

The two filed quarters sum exactly to the filed half-year on every line, so nothing in the quarterly table above is a derivation. Q1 revenue of $348,116 plus Q2 revenue of $382,486 equals the reported $730,602.

Quality of Beat

• Revenue. The beat is 0.4%, which is inside the noise band for a company this size. What is not noise is the mix: Diagnostics grew 19.6% against Data and Applications at 28.0%, and inside Diagnostics the split is stark. Oncology volume accelerated to +31% from +28% in Q1, while Hereditary revenue grew only 5% to $107.4M. Management attributed the Hereditary deceleration to lapping, which is fair, because Q2 2025 was the first quarter with Ambry fully consolidated. But an all-in 21.6% growth rate that decelerated from 36.1% in Q1 is not a comparability quirk you can wave away either; it is what the reported base now looks like without acquisition arithmetic.
• Margins. This is the highest-quality part of the print. The 240bp of consolidated gross-margin expansion came from Diagnostics, where cost of revenue grew 8.5% against 19.6% revenue growth, lifting segment gross margin 380bp to 62.6%. That is operating leverage on a fixed lab-cost base, not mix. It is partly offset by Data and Applications gross margin compressing 250bp to 70.2%, where cost of revenue grew 39.9% against 28.0% revenue growth, which is the cost of provisioning GPUs and platform access for customers building inside Lens. Consolidated margin still went the right way because Diagnostics is 75.6% of revenue.
• EPS. The non-GAAP loss narrowed from $(0.22) to $(0.04). Roughly half of that improvement is operating (non-GAAP operating loss narrowed by $14.3M) and roughly half is below the line: interest expense fell 52.3% to $10.3M as the convertible refinancing retired higher-cost debt, partially offset by an $11.6M loss on extinguishment that the non-GAAP bridge adds back. Share count grew 3.8% year over year, so the per-share improvement is not being manufactured by buybacks; there are none.
• Cash. Operating cash use in the quarter was $7.5M, a genuine improvement, but the half-year figure is $80.8M of cash used against $61.5M a year ago. The balance-sheet build to $820.7M is financing, not operations: $443.1M of net convertible proceeds against $207.7M of term-loan repayment, $100.0M of revolver repayment, $31.2M of capped-call purchases and a $6.4M prepayment premium.

Segment Performance

Diagnostics

Diagnostics generated $289.3M of revenue on 19.6% growth and $181.1M of gross profit on 27.5% growth. The gap between those two numbers is the entire Diagnostics story this quarter: revenue grew 19.6% while cost of revenue grew 8.5%, and the segment converted that into 380bp of margin expansion. Sequentially the segment added $28.2M of revenue and 130bp of margin.

Underneath the aggregate, the two halves of Diagnostics are moving in opposite directions. Oncology volume growth accelerated to 31% from 28% in Q1, which is the number that matters for the long-term thesis because oncology testing is what feeds the multimodal dataset. Hereditary, which arrived with the Ambry acquisition, grew revenue only 5% to $107.4M.

"Hereditary revenue for the quarter was up 5% to $107.4 million, as Q2 of 2025 was a period of abnormally high growth, which we are now lapping."
— Eric Lefkofsky, Founder and CEO

Management framed Hereditary's target as "mid-teens growth by the end of the year," which implies a meaningful reacceleration from 5% over the next two quarters. The company also launched GenomeNext, a whole-genome product aimed at the rare-disease market inside Ambry, and characterized the first month as running 50% ahead of internal expectations on what it conceded were small absolute numbers.

Assessment: The margin expansion is durable because it is volume leverage on a lab cost base that Tempus has already built, and the ADLT repricing described below sits on top of it as pure price. The risk is that Hereditary has to roughly triple its growth rate in two quarters for the segment's aggregate to hold, and the only evidence offered was one month of a new product launch. We would treat the mid-teens exit rate as an aspiration rather than a plan until Q3 shows the inflection.

Data and Applications

Data and Applications generated $93.2M on 28.0% growth, with the Insights data-licensing and modeling sub-business growing 36%. The segment signed approximately $200M of new licenses in the quarter, including BioNTech, Daiichi Sankyo, Level Set Bio and Incyte, and delivered the first version of its oncology foundation model to AstraZeneca.

"Data and Apps revenues were $93.2 million, increasing 28% year-over-year, with our data licensing and modeling business, Insights, growing at 36% in the quarter."
— Eric Lefkofsky, Founder and CEO

Two things in that sentence deserve attention. Insights growing at 36% is very good and it is also slower than the 44.1% it printed in Q1. And segment gross margin fell 250bp to 70.2% because cost of revenue grew 39.9%, well ahead of revenue. On the call management described customers moving beyond data licensing into the Lens platform, uploading their own data, building models in the Tempus environment and having Tempus provision GPUs for them. That is a better business strategically and a structurally lower-margin one mechanically, at least until scale arrives.

Assessment: This is the segment that justifies a technology multiple rather than a lab multiple, and the $200M bookings figure against a segment run-rate of roughly $373M annualized is the strongest single datapoint in the quarter. But the margin trend is the thing to watch, not the growth rate. If GPU provisioning and platform delivery keep cost of revenue growing 12 points faster than revenue, the segment's gross margin advantage over Diagnostics closes inside two years, and with it a good part of the mix-shift argument for the stock.

Key Operating Metrics

Key Topics & Management Commentary

Overall Management Tone: Confident and unusually forward-dated. Management spent most of its airtime on 2027 and 2028 economics rather than on the quarter just delivered, and the prepared remarks moved through the print in under two minutes before pivoting to ADLT pricing, the foundation model and the Personalis rationale. Analyst pushback was narrow and clustered on deal mechanics rather than on operations, and the one genuinely uncomfortable question, whether the acquiree could walk away on the share-price floor, drew a long and voluntary answer rather than a deflection. The posture on near-term numbers was noticeably more restrained than the posture on the pipeline, which is visible in a guidance raise of $5M against roughly $200M of new bookings.

1. The GAAP Profit Is a Mark, Not a Milestone

Tempus reported GAAP net income of $5.6M, its first positive quarter, and led the press release with it. The line that produced it is "Other income, net" of $102.8M, of which the company disclosed $98.5M as unrealized gains on marketable equity securities. The non-GAAP reconciliation removes $97.4M of fair-value changes to get from net income to non-GAAP net loss, which is the company itself telling you the gain is not operating.

"GAAP net income was $5.6 million, and Adjusted EBITDA was $8 million, a $13.6 million year-over-year improvement."
— Eric Lefkofsky, Founder and CEO

The Adjusted EBITDA half of that sentence is the meaningful half. The GAAP half is a function of what the equity portfolio did in the quarter, and marketable equity securities on the balance sheet rose from $150.2M at year-end to $216.4M at June 30. Those marks reverse as readily as they arrive; the Q1 print carried $32.3M of unrealized losses on the same portfolio and contributed to a $125.9M net loss.

Assessment: Anyone building a model to a "Tempus is profitable now" headline is building to a number that swings by $130M between consecutive quarters on portfolio marks alone. The correct GAAP anchor is the operating loss, which widened 22.9% to $75.9M, and the correct non-GAAP anchor is Adjusted EBITDA, which is genuinely positive for the first time. Neither one supports the headline.

2. Gross Margin at 64.4% and Where It Actually Came From

Consolidated gross margin reached 64.4%, up 240bp year over year and 60bp sequentially. The composition is worth separating because the two segments contributed in opposite directions. Diagnostics gross margin rose 380bp to 62.6% on 8.5% cost growth against 19.6% revenue growth. Data and Applications gross margin fell 250bp to 70.2% on 39.9% cost growth against 28.0% revenue growth.

The Diagnostics improvement is the more valuable of the two because it is leverage on a cost base that scales sub-linearly with volume, and because the ADLT repricing described below lands on top of it as incremental price with no incremental cost.

Assessment: A company whose lower-margin segment is expanding margin and whose higher-margin segment is compressing it is converging toward a blended rate somewhere in the mid-60s. That is a fine outcome. It is not the 70%-plus software-like trajectory that the Data segment's growth rate implies to a reader who does not open the segment margin table.

3. Oncology Accelerates to +31% While Hereditary Stalls at +5%

Oncology volume growth accelerated to 31% from 28% in Q1 and management called out June as one of the strongest months in the portfolio's history. Hereditary grew 5%. Both facts live inside a Diagnostics segment that grew 19.6%.

The Hereditary deceleration has a legitimate explanation. Ambry closed in February 2025, so Q2 2025 was the first fully consolidated quarter and included the acquisition ramp. Lapping that produces a hard comparison. The company's stated path back is mid-teens growth by the fourth quarter, supported by the GenomeNext whole-genome launch.

"The launch was great. Great, meaning we had an expectation for the first month, and I think I'm going to say something like two or three weeks in, we were already 50% higher than our expectation."
— Eric Lefkofsky, Founder and CEO

Management was careful to add that these are small numbers, that one month of data is not enough to judge cannibalization of the existing whole-exome business, and that the company is "being conservative about our whole genome estimates."

Assessment: Oncology accelerating is the more important of the two facts, because oncology volume is the input to the data asset and therefore to the highest-margin revenue in the company. But a 5% growth rate in the largest Diagnostics sub-line, with a mid-teens Q4 target attached to a three-week-old product, is a real gap between guidance and demonstrated performance. It is the specific thing to check in the Q3 print.

4. Data and Applications: $200M of Bookings and a Decelerating Insights Line

The quarter's bookings figure of roughly $200M is large relative to a segment annualizing near $373M. New logos included BioNTech, Daiichi Sankyo, Level Set Bio and Incyte, joining AstraZeneca, GlaxoSmithKline, Bristol Myers Squibb and Merck. Management described the pipeline in unusually unguarded terms.

"We've had more momentum in terms of signing deals than we've had in a long time, in years."
— Eric Lefkofsky, Founder and CEO

Against that, Insights growth decelerated from 44.1% in Q1 to 36% in Q2. Management's framing of the long-term algorithm was explicit and lower than either figure: "We really are interested in maintaining long-term growth in that close to 30% range plus," with the stated ambition of sustaining that "for years, three years, five years, seven years."

Assessment: The bookings number is contracted, multi-year and recognized ratably, which means it is a 2027 and 2028 revenue statement more than a 2026 one. Management effectively guided the segment's durable growth rate down to "close to 30% plus" in the same breath as announcing record bookings. That is either admirable conservatism or the first signal that 40%-plus was a pull-forward. We read it as the former, but it should temper any model that extrapolates Q1's 44%.

5. The AstraZeneca Foundation Model Cleared Its Acceptance Criteria

Tempus delivered the first version of its oncology foundation model to AstraZeneca and the model passed blinded acceptance testing, having to match purpose-built models that both parties had tuned for specific tasks.

"The fact that we've met the acceptance criteria means that they're comfortable this model's predictive and can now serve as the foundation, even though it's a foundation model, for all kinds of R&D and development work they're doing."
— Eric Lefkofsky, Founder and CEO

On the commercial question of what happens to the relationship after 2026, management was candid to the point of vagueness about the contract's own terms, saying the existing agreement runs "another couple of years" and then "I think the current agreement goes, I don't even know, through 2028 or something. I have no idea." The substantive commitment was that "They will be a very large client in 2027, no matter what happens."

Assessment: Passing a blinded acceptance test against a large pharma's internal models is the single most credible technical validation Tempus has produced, and it converts the foundation-model narrative from a claim into a delivered artifact. The commercial framing is weaker. An unprompted "I have no idea" about a flagship contract's expiry is not what an investor wants to hear when that contract is the anchor tenant of the segment carrying the multiple.

6. xT CDx Approval Migrates the Tissue Portfolio to ADLT Pricing

The FDA approved the tumor-only xT CDx assay, making Tempus the first laboratory holding companion-diagnostic approval for both tumor-only and tumor-normal comprehensive genomic profiling. The commercial consequence is that the entire solid-tumor DNA portfolio can migrate to Advanced Diagnostic Laboratory Test pricing.

"We expect an estimated $200 uplift in ASP, which equates to approximately $85 million on an annual basis beginning in 2027."
— Eric Lefkofsky, Founder and CEO

Set against a Diagnostics run-rate of roughly $1.16B annualized, $85M is roughly 7 points of segment growth arriving as pure price on already-installed volume, at close to 100% incremental margin.

Assessment: This is the highest-confidence item in the entire 2027 bridge, because the approval is already granted and the pricing pathway is administrative rather than commercial. It is also, precisely because it is 2027, worth nothing to the 2026 guidance the stock is currently being valued against.

7. xF and the $7,500 Assumption: the Bridge to 2028

The liquid-biopsy assay xF is in front of the FDA now, with approval expected in the latter half of 2027. The disclosed ASP uplift assumption moved sharply: at the investor day, using Q4 data, the uplift was framed at $230; on this call it was $550.

"On the xF pricing, as others have gone down the approval for liquid biopsies and indicated the prices that they're going after, our thinking around the ADLT pricing for xF has evolved, and we think that there's additional upside from what we had pegged it for earlier on."
— Jim Rogers, CFO

Pressed for the absolute number, management gave it: "Yeah. For xF, Mark, we're assuming a $7,500 ADLT price." The justification was competitive comparability, that a peer's recently approved liquid-biopsy assay of similar megabase size carries ADLT pricing management put at roughly $8,400, and that "It would be very hard for us to go to the market with a almost identical, at least in terms of size and complexity assay that's radically less expensive."

Combined, management sized the two approvals at approximately $400 million of revenue uplift in 2028.

Assessment: The pricing logic is sound and the comparability argument is how ADLT rates actually get set. The caution is that a 139% increase in a disclosed ASP assumption between one investor event and the next quarterly call, based on watching a competitor's rate, is an assumption and not a rate card. Tempus does not control the Medicare determination, the assay is not approved, and the revenue is two full years out. We would carry the $85M xT uplift in a 2027 model and treat the xF contribution as optionality rather than as a line item.

8. MRD: 9,000 Tests, +38% Sequentially, and a Deliberately Throttled Sales Force

MRD volume reached roughly 9,000 tests, up from roughly 6,500 in Q1. Management sized the addressable market at more than $20 billion and described it as one of the fastest-growing segments in oncology diagnostics. And yet only about 10% of the sales force is permitted to sell it.

"Up until now, we have phased our sales efforts, as only about 10% of our sales force is selling MRD today based on these reimbursed indications."
— Eric Lefkofsky, Founder and CEO

The reason given is unit economics rather than capacity. Management will not unshackle the sales force until the test is at breakeven ASP: "If you're losing money, if your margin's negative and you rush to run an extra 100,000 tests, you're just burning money." The strategic premise is that MRD is not a land-grab market: "We don't believe there's beachfront real estate to be procured. We do believe it's important that we're in market with an offering that is comprehensive and people want."

Assessment: The discipline is admirable and consistent with a company trying to reach free-cash-flow breakeven. It also means the growth rate being celebrated (+38% sequentially) is being produced by a tenth of the commercial organization, which is a genuinely impressive fact and a reminder that MRD contributes very little revenue today. The premise that no land-grab dynamic exists is the load-bearing assumption. If a competitor with deeper reimbursement decides otherwise, Tempus will have chosen margin over share in a $20B market.

9. Personalis: $1.5B, All-Stock, and the Financing Isn't Arranged

Announced July 20, the acquisition of Personalis values the company at $16.25 per share and roughly $1.5B enterprise value. The structure is what matters.

"The transaction is structured as a 100% stock transaction, with Tempus having the option to elect payment in cash, capped at 50% of the consideration paid."
— Eric Lefkofsky, Founder and CEO

Management's stated intent is to lean on debt rather than equity, and was explicit that the arrangement is not yet in place: "We have already begun working with parties to put a debt facility in place as our intention, obviously depending on our stock price, is to finance a large portion of the proceeds with debt to minimize shareholder dilution."

Guidance assumes no contribution. The transaction is expected to close in late Q4 2026 or early 2027.

Assessment: A $1.5B acquisition against an $8.9B equity value, financed by an unarranged debt facility on top of $1.36B of existing convertible obligations, is the largest single risk in the story and it is not reflected anywhere in the 2026 numbers. The strategic logic is sound: Tempus already resells the Personalis assay, the reimbursement pipeline is building, and owning it converts a partner margin into an internal one. The execution question is entirely about the capital structure that emerges, and management has not yet been able to answer it.

10. The Convertible Refinancing and the Claim of Free Cash Flow by Year-End

Tempus completed a $460M offering of 0.0% convertible senior notes due 2032 and used the proceeds in part to repay the Ares term loan. The income-statement effect is already visible: interest expense fell 52.3% year over year to $10.3M, offset in the quarter by an $11.6M loss on extinguishment.

"Importantly, this transaction allows us to save over $30 million annually in interest expense, enabling us to achieve positive free cash flow by year-end."
— Eric Lefkofsky, Founder and CEO

Operating cash use narrowed to $7.5M in the quarter from a first-half total of $80.8M, so the trajectory supports the claim directionally. The balance sheet now carries $1,172.8M of convertible senior notes net and $187.9M of convertible promissory note, against $820.7M of cash, marketable securities and restricted cash, for net debt of roughly $540.0M. The revolver and the term loan are both retired.

Assessment: Trading coupon for conversion optionality at a 0.0% rate is good treasury work and the interest saving is arithmetic rather than forecast. The free-cash-flow claim is a genuine commitment with a date attached, which makes it a clean thing to grade next quarter. It is also worth noting what "positive free cash flow by year-end" means in practice: it is a run-rate statement about a single quarter, not a full-year outcome, and the first half consumed $80.8M.

11. Stock-Based Compensation at 14.1% of Revenue

Stock-based compensation was $54.1M in the quarter, 14.1% of revenue, against $22.5M and 7.1% a year ago. For the half, it was $106.8M and 14.6% of revenue against 8.0%. This is the single largest reconciling item between the GAAP loss and the Adjusted EBITDA profit, and it is growing more than twice as fast as revenue.

It was not raised on the call, by management or by any analyst.

Assessment: Adjusted EBITDA of $8.0M is arrived at after adding back $54.1M of compensation paid in equity. Share count grew 3.8% year over year, which is the actual cost showing up where it belongs. A company that reaches EBITDA breakeven while doubling its equity-compensation intensity has not reached breakeven in the sense most readers assume. This is the number we would want to see flatten before treating the margin story as complete.

Guidance & Outlook

"We are increasing guidance to $1.595 billion-$1.605 billion in 2026, representing approximately 25% growth."
— Eric Lefkofsky, Founder and CEO

Implied second-half ramp. First-half revenue was $730.6M. At the $1.600B midpoint the back half must deliver $869.4M, or $864.4M to $874.4M across the range. Against second-half 2025 revenue of $701.4M that is 23.9% growth at the midpoint, decelerating from the 28.1% the first half just produced. Expressed per quarter, the guide requires an average of $434.7M, which is 13.7% above the $382.5M just reported.

The EBITDA line is the harder one. First-half Adjusted EBITDA was $5.2M. Holding the full year at approximately $65M requires roughly $59.8M in the back half, which is 92.0% of the full-year target, or about $29.9M per quarter against the $8.0M just delivered. That is a 3.7x step-up in quarterly Adjusted EBITDA over two quarters, and management maintained rather than raised it.

Street positioning. Consensus was already inside the prior range, so the raise moves nobody's model materially. No sell-side rating migrated in the two weeks following the print. The debate is not about 2026.

Guidance style. On the evidence of two data points this year, Tempus guides conservatively and raises in small increments: the May print raised the full year and the July print raised it again by $5M. Management's own long-term framing for the Data segment, "close to 30% range plus," is below both its trailing and its current growth rate. That pattern argues the $1.605B high end is achievable. It says nothing about the EBITDA ramp, which is a different kind of promise.

Analyst Q&A Highlights

When the Acquired MRD Economics Actually Flip in Tempus's Favor

The opening question of the call went straight to the acquisition's underlying assumptions rather than to the quarter. The specific concern was average selling price: Personalis has secured coverage in several indications over recent quarters, and the questioner wanted to understand how those economics improve from here. The response tied ASP improvement to two independent levers, incremental reimbursed indications and Tempus's ability to widen the selling motion, and was careful to place the asset early on its own pricing curve rather than claim the improvement had already happened.

Q: "Maybe one for Jim, just on Personalis. We've had quite a lot of questions coming through just on your underlying assumptions on the ASP front and just how those economics could become more favorable to you with time."
— Kallum Titchmarsh, Morgan Stanley

A: "Over time, obviously we would anticipate ASPs to continue to improve as they secure coverage and additional indications. Also from a volume perspective, our ability to expand the sales force that is able to sell that test, which today is around 10%, will help us drive volume. They're early on in the ASP curve, but they've obviously had a tremendous amount of success in getting the first couple indications approved, and we anticipate that continuing."
— Jim Rogers, CFO

The more revealing answer came in the follow-on, where management described the existing reseller arrangement inverting once reimbursement matures: "All of a sudden, the economics that were more favorable for us or that are more favorable for us today because we get paid and don't lose money, will actually flip, and all of a sudden they'll be getting paid, they'll have more margin, and we'll wish we had that deal instead of our deal."

Assessment: That is an unusually frank statement of the deal's timing logic. Tempus is buying the asset specifically because the partnership economics are about to turn against it, which is a coherent reason to acquire and also an admission that the price reflects value that has not yet materialized. It makes the reimbursement pipeline, not the integration, the variable that determines whether $1.5B was the right number.

Whether the Acquiree Can Walk Away at the Share-Price Floor

The most uncomfortable exchange of the call concerned the merger agreement's own mechanics. Because the consideration is stock, the agreement contains a floor below which the target may terminate, and the stock had been trading beneath it entering the print. Management did not deflect, and volunteered both the reasoning behind the floor and its current financing preference.

Q: "Secondly, with your shares trading below $46, there's a possible that Personalis to terminate. Can you just talk about what you can do to avoid that as well as what makes you confident they don't do that, they don't terminate?"
— Kyle Mikson, Canaccord Genuity

A: "In terms of Personalis, I can't see a scenario where they would want to terminate, even if we were slightly below the floor. We established the floor because we weren't willing to have more dilution than X amount. We obviously have cash as a lever. We've got stock as a lever. We don't want to have more than X amount of dilution. Given where we're trading now, obviously my preference would be to fund maybe close to half the transaction in cash and the balance in stock to keep the dilution quite low."
— Eric Lefkofsky, Founder and CEO

The confidence argument offered was the counterparty's own incentives rather than any contractual protection: the two companies are already commercial partners, and management characterized a broken deal as "highly disruptive if this deal didn't get done on their side."

Assessment: The answer is honest and it is also not reassuring in the way it was meant to be. The floor exists because Tempus set a dilution ceiling, which means a lower share price does not simply resize the equity component, it pushes the company toward the cash election and therefore toward the debt facility that is not yet arranged. The deal risk and the financing risk are the same risk, and both are functions of a share price that is currently doing a great deal of moving.

Why the Liquid-Biopsy Price Assumption More Than Doubled in One Quarter

A questioner caught the change directly: the disclosed ASP uplift for the pending xF assay had moved from $230 at the investor day to $550 on this call, without being flagged in the prepared remarks. The explanation was that competitive ADLT rates set in the intervening period had reset the company's own expectation upward.

Q: "First one on the xF FDA clearance tailwind, that looks like it's now $550 using 2Q data compared to $230 that you had at the investor day that was using 4Q data. Just I don't think you called out the reason for the change there. Can you just comment on that?"
— Kyle Mikson, Canaccord Genuity

A: "On the xF pricing, as others have gone down the approval for liquid biopsies and indicated the prices that they're going after, our thinking around the ADLT pricing for xF has evolved, and we think that there's additional upside from what we had pegged it for earlier on. That assay is in front of the FDA now. As Eric mentioned, as we get later into 2027, we would anticipate getting approval and then following the ADLT pathway. That's the rationale behind the change."
— Jim Rogers, CFO

Management added the pricing anchor explicitly, citing a peer assay of comparable megabase size carrying ADLT pricing it put near $8,400, and concluded: "The pricing here is just higher than we expected, and so it's a significant benefit to us."

Assessment: The revision was disclosed only because it was asked about, which is the notable part. A 139% increase in a headline economic assumption for an unapproved product, driven by observing a competitor's rate, belongs in the prepared remarks rather than in the third Q&A answer. The underlying logic is defensible and the number is unverifiable until the FDA acts.

Visibility Into the Implied 2026 Data-Revenue Number

The sharpest modelling question on the call worked backward from disclosed contract values to ask whether the implied data-revenue figure for the year was still the right one, given a mid-year customer expansion and roughly $200M of new bookings layered on top of contract value already earmarked for 2026. The answer never engaged with the specific figure, and pivoted instead to multi-year pipeline visibility.

Q: "As we layer in that BMS expansion and $200 million of bookings in the quarter on top of the $350 million of TCV that was already earmarked for revenue in 2026, how much visibility and confidence do you have in hitting the implied $410 million of data revenue guidance, if that's even still the right number?"
— Matt Shea (for Ryan MacDonald), Needham & Company

A: "The business just feels super healthy, super sticky, and we just have a stronger pipeline and more demand than we've had, which means we have great visibility into our growth rates, not just in 2026, but 2027. That's how we think about the data business. We really are interested in maintaining long-term growth in that close to 30% range plus. We want to plot these things out in a way that we feel like we can grow at that level for years, three years, five years, seven years."
— Eric Lefkofsky, Founder and CEO

Assessment: A precise question about a specific 2026 number received a qualitative answer about 2027 through 2033. The company did not confirm, revise or reject the implied figure. The substantive content, delivered almost in passing, is that management's durable growth algorithm for its highest-margin segment is "close to 30% range plus," which is materially below the 36% just printed and the 44.1% printed in Q1. That is the number to carry forward, and it deserved a clearer statement than it got.

Why the MRD Sales Force Is Still Capped at 10%

Given a market management sized at more than $20 billion and sequential volume growth of 38%, the obvious question is why nine in ten sales representatives are still prohibited from selling the product. The answer was a straightforward unit-economics argument, and it doubled as a statement about how management thinks about competitive dynamics in the category.

Q: "Eric, when do you think you can take that 10% promoting the Personalis test now? Why not take that up faster? Do you think you could take that up sooner rather than later, or are you waiting for the deal to perhaps close?"
— Mark Massaro, BTIG

A: "The same constraints we had when we didn't own Personalis will be the same constraints we'll have even after this transaction closes, which is we just want to time the full unshackling of these efforts to having the tests at an ASP level be basically break even. If you're losing money, if your margin's negative and you rush to run an extra 100,000 tests, you're just burning money."
— Eric Lefkofsky, Founder and CEO

Management then made the strategic premise explicit, arguing that the category has no first-mover lock-in and citing the therapy-selection market as the precedent: "We don't believe there's beachfront real estate to be procured." The stated trigger for ramping is ASP-driven: "At some point, you'll see this pivot where the ASPs will start to climb up and you can see break even in sight, and that's the point where I think you should expect us to ramp up the sales force pretty dramatically."

Assessment: Management gave a clean, falsifiable answer with a trigger attached, which is more than most companies offer on a question like this. The entire position rests on one assumption: that MRD share taken early is not defensible share. If that is wrong, the discipline that protects 2026 EBITDA costs the company the market it just paid $1.5 billion to enter.

Whether Platform Engagement Changes the Revenue Recognition Profile

A recurring line of questioning through the call concerned whether the newer, deeper customer relationships, where pharma clients access GPUs and build models inside the Tempus environment rather than simply licensing files, would accelerate revenue recognition relative to total contract value. Management said no, and used the answer to explain a deliberate structural choice made years earlier.

Q: "Just in terms of economics to Tempus, should we assume that some of the concept of those deals drive potentially better revenue to you all over the course of the partnership? Is it maybe faster recognition of booking revs versus backlog?"
— Brendan Smith, TD Cowen

A: "I don't expect these new deals where people are getting more ingrained with Lens and getting more ingrained with building small or large models in our environment, accessing GPUs at some scale. I don't think they'll change revenue recognition at all. They just are another element of stickiness that locks people into our ecosystem."
— Eric Lefkofsky, Founder and CEO

The historical context offered was that Tempus deliberately abandoned upfront file-licensing revenue years ago in favour of multi-year subscriptions, describing the transition as "tough to swallow back then, but great for the long-term health of the business because we now have awesome visibility multiple years out."

Assessment: The answer is the right one for the business and the wrong one for anyone hoping the $200M bookings number shows up soon. Deeper platform engagement raises retention and raises cost of revenue, which is precisely what the segment's 250bp of gross-margin compression reflects, and it does nothing for near-term revenue. Stickiness is a 2028 argument.

The Algorithm Attach Rate Rose to 45% and Still Earns Nothing

The final question of the call asked whether the previously disclosed 40% algorithm attach rate on solid-tumor testing had held, what drives it, and what rate is embedded in guidance. The answer disclosed an increase and then, unprompted, disclosed that none of it is monetized.

Q: "You guys have cited about a 40% algorithm attach rate on solid tumor. Is that still the case in Q2? I guess, what's the algorithm that drives it and what attach rate is then embedded in your guidance here?"
— Robert Bamberger, Baird

A: "The algorithm attach rate in Q2 is 45%, a slight uptick from the 40% that we had quoted in Q1. It's really broad-based. Then in terms of the guide, many of those algorithms remain not being paid, and so there's no impact on revenue from the number of algorithms. Although it does highlight, again, our advantage in diagnostics are the insights that we provide physicians beyond just the test results."
— Jim Rogers, CFO

Assessment: Five points of attach-rate expansion in a quarter, on a product feature that generates no revenue, is the clearest available evidence that the algorithmic layer is genuinely differentiating the core test rather than decorating it. Management explicitly connected it to the 31% oncology volume acceleration. It is also an unpriced option: nothing in the model assumes these are ever reimbursed, which makes it upside that costs nothing to hold and produces nothing to underwrite.

What They're NOT Saying

1. Stock-based compensation was never mentioned. SBC reached $54.1M, 14.1% of revenue, up from 7.1% a year ago, and it is the largest single item bridging the GAAP operating loss to positive Adjusted EBITDA. Neither management nor any analyst raised it on a call that spent considerable time on margin improvement.
2. No one quantified the back-half EBITDA ramp. Guidance of approximately $65M was reaffirmed against $5.2M delivered in the first half, requiring roughly $59.8M in two quarters. Management restated the number without describing what produces the step-up, and the question was not asked.
3. The full-year operating loss trajectory went undiscussed. Loss from operations widened to $75.9M from $61.8M year over year and to $160.6M from $130.5M for the half. The prepared remarks went from revenue directly to GAAP net income, skipping the operating line entirely.
4. The Personalis debt facility has no terms, no size and no counterparty. Management confirmed it has "already begun working with parties" and did not offer an amount, a rate, a structure or a timeline, on a transaction expected to close within roughly one to two quarters.
5. The AstraZeneca contract's expiry is apparently not known. Asked directly about post-2026 terms for the segment's anchor customer, the answer included "I think the current agreement goes, I don't even know, through 2028 or something. I have no idea."
6. Insights deceleration went unexplained. Growth fell from 44.1% in Q1 to 36% in Q2, in a quarter with record bookings. No analyst asked and management did not volunteer a bridge.
7. Data and Applications gross-margin compression was not addressed. Segment margin fell 250bp year over year as cost of revenue grew 39.9% against 28.0% revenue growth. The GPU-provisioning dynamic was described qualitatively as a positive without ever being connected to the margin line it is producing.
8. No 2027 revenue framing was offered. Management referenced 2027 and 2028 repeatedly, quantified ASP uplifts of $85M and roughly $400M, and declined to place any of it against a revenue base. Only the Data segment received a growth algorithm, and only as "close to 30% range plus."

Market Reaction

• Pre-print setup: The stock closed at $44.29 on July 30, entering the print down 25.0% year to date against an S&P 500 up 8.7%, down 23.4% over twelve months and down 23.5% over the trailing 30 days. The 52-week closing range was $41.55 to $103.25, so the print arrived with the shares near the low end of a badly broken twelve-month chart. The July 30 session itself rose 6.6% from the prior close of $41.55.
• After-hours move: Shares initially fell after the release, trading down roughly 3% to about $42.97, then recovered through the call as management walked through the ADLT repricing path and the Personalis rationale.
• Reaction session (July 31): Opened at $46.24, a 4.4% gap up, traded a $42.30 to $46.30 range and closed at $43.87, down 0.9% on the day. Volume was 7.3M shares against a 30-day average of 5.7M, or 1.3x.
• Since the print: Shares closed at $49.36 on August 18, the last completed session before publication, up 11.4% from the pre-print close and up 12.5% from the reaction-day close. The post-print peak close was $55.69 on August 11, 25.7% above the pre-print level, followed by five consecutive down closes.

The reaction-day pattern is the informative one. A 4.4% gap up that gave back its entire gain and closed slightly red, on only 1.3x volume, is the signature of a print that satisfied nobody's thesis. Buyers had a genuine list to work from: first positive GAAP quarter, first positive Adjusted EBITDA, gross margin at 64.4%, an FDA approval and roughly $200M of bookings. Sellers had a shorter but heavier list: a $5M guidance raise, a widening operating loss and an acquisition whose financing was undefined. On the day, neither side could hold the tape.

What happened afterward is harder to attribute and worth stating plainly rather than rationalizing. Shares rose 12.9% on August 7 and continued to a $55.69 close on August 11, a 25.7% advance from the pre-print level, with no company disclosure to explain it. Tempus filed no 8-K in that window, no sell-side rating changed after August 4, and no company-specific news carried the move. The shares then gave back most of it into the August 18 close of $49.36. This looks like positioning and sentiment in a heavily-shorted, high-beta healthcare AI name rather than a re-rating on the quarter.

For completeness, and because it is material to anyone reading this on the day of publication: the shares are trading sharply higher in the current August 19 session, quoted near $59.61 in the early afternoon on roughly 14 million shares. 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 August 18 close of $49.36.

Street Perspective

Debate: Is the Adjusted EBITDA Inflection Real or a Definition?

Bull view: Adjusted EBITDA turned positive at $8.0M against a $5.6M loss a year ago, gross margin expanded 240bp, non-GAAP operating loss narrowed from $17.0M to $2.7M, and interest expense halved. Every one of those is an operating fact independent of the marks, and the company is now within one quarter of the free-cash-flow crossover it has promised.

Bear view: Adjusted EBITDA is arrived at by adding back $54.1M of stock-based compensation, an item growing at 141% against revenue growth of 22%. Loss from operations widened 22.9%. The "profit" everyone cited is a securities mark. On the only two measures that do not depend on definitional choices, the operating line and cash from operations, the company went backwards year over year in both the quarter and the half.

Our take: The bulls have the better of the gross-margin argument and the bears have the better of the profitability argument, and they are not in conflict. Tempus is genuinely converting scale into unit economics in Diagnostics, and it is simultaneously funding an ever-larger share of its cost base with equity. The honest characterization is that the business is improving and the profitability is not yet real, which is exactly what an approximately $65M Adjusted EBITDA target on $1.6B of revenue describes: a 4% margin, after $200M-plus of annualized SBC add-backs.

Debate: Does the 2027–2028 ASP Bridge De-Risk the Story or Postpone It?

Bull view: The xT CDx approval is granted, not pending, and delivers roughly $85M annualized at close to full incremental margin beginning in 2027. The xF assay is filed, with a comparable competitor rate already establishing the pricing precedent, and the two together are worth roughly $400M of 2028 revenue uplift on volume the company already runs. That is a rare thing in diagnostics: growth that requires no new customers.

Bear view: Both are regulatory and reimbursement events outside the company's control, and the xF assumption already moved 139% in one quarter on the basis of watching a competitor. Meanwhile the near-term algorithm is a 24% second-half growth rate, decelerating from 28%, with a $5M guidance raise. An investor is being asked to underwrite 2028 while paying for 2026.

Our take: The xT uplift belongs in a 2027 model and we would carry it. The xF contribution should be held as optionality until the FDA acts, because an unapproved assay with a self-revised price assumption is not a forecastable line. The deeper point is the one the bears are making about duration: the value creation in this story has migrated out to 2027 and 2028, and the multiple has to be paid today.

Debate: Is Personalis a Strategic Necessity or a Balance-Sheet Risk?

Bull view: MRD is a $20B-plus market growing faster than any other segment in oncology diagnostics. Tempus already resells the assay, so integration risk is low and the commercial motion is proven. Buying it now, before reimbursement matures and the partnership economics invert, is precisely the right timing, and it completes a portfolio that no competitor can match across tissue, liquid and residual disease.

Bear view: A $1.5B acquisition against an $8.9B equity value, financed by a debt facility that has no terms and sits on top of $1.36B of existing convertibles, is a large bet placed by a company that has never generated positive annual free cash flow. The share-price floor in the agreement is already a live issue, and the cheaper the stock gets the more of the consideration has to come from a balance sheet that is not yet self-funding.

Our take: The strategic case is stronger than the market seems to credit, and management's own explanation of why the economics are about to flip is the most persuasive argument for the deal. The financing is the problem, and it is unresolved by design rather than by oversight, because the mix depends on a share price that is currently moving 20% in a session. We would not underwrite this transaction until the debt facility is disclosed, and we do not think a reader should either.

Model Update & Valuation Framework

Valuation, anchored to the August 18, 2026 close of $49.36. With 175,200,077 Class A and 5,043,789 Class B shares outstanding at June 30, the equity value is approximately $8.90B. Net debt is roughly $540.0M, being $1,172.8M of convertible senior notes net and $187.9M of convertible promissory note against $820.7M of cash, marketable securities and restricted cash. Enterprise value is therefore approximately $9.44B, or 5.9x the $1.600B midpoint of FY2026 revenue guidance and 7.4x trailing FY2025 revenue. EV to FY2026 Adjusted EBITDA is not a meaningful figure at a roughly $65M target and we do not present one.

What that multiple has to assume. At 5.9x forward revenue for a business growing 25% with a 4% Adjusted EBITDA margin and a negative GAAP operating margin, the price is neither demanding nor cheap in isolation. It becomes demanding once the composition is examined. Roughly a quarter of revenue carries a genuine technology profile and is decelerating; three quarters is a diagnostics business whose margin gains are real but whose largest sub-line grew 5%. The 2027 and 2028 ASP catalysts are what close the gap, and they are the part of the model that depends on regulators.

Valuation impact. We are initiating without a formal price target. The two variables that would move one are outside management's control in the near term, being the xF determination and the Personalis financing mix, and a target set before either resolves would be a placeholder rather than a view. We would revisit on the disclosure of the debt facility, on a Q3 print that demonstrates the EBITDA ramp is achievable, or on a Hereditary reacceleration toward the mid-teens exit rate management has committed to.

Thesis Scorecard Post-Earnings

This is initiation coverage, so the pillars below are established rather than graded against a prior quarter. Each is stated as we will carry it forward and scored against what this quarter's print and call actually revealed.

Overall: Thesis established. The bull pillars and the bear points are simultaneously confirmed, which is the honest reading of a quarter where the operating business genuinely improved and the reported profitability did not. The asset is differentiated and the 2027 repricing is the most credible catalyst in the story. What is missing is any evidence that the company can convert 25% revenue growth into cash inside the guidance year, and the one number that would have demonstrated intent, the full-year EBITDA guide, was reaffirmed rather than raised on a beat.

Action: Hold. We would not chase the stock here. The post-print recovery has already removed the valuation cushion that existed on July 31, and the two disclosures that would change the view, the Personalis debt facility and a Q3 print consistent with the back-half EBITDA ramp, both arrive before year-end. Owners with a 2028 horizon have a defensible position and no reason to sell into the ASP bridge. New capital has two identifiable catalysts to wait for and is being paid to wait by nothing at all, since the company generates no free cash flow and pays no dividend.