Merck Carried the Quarter, Tempus Carries the Stock: A 30% Beat With Flat First-Half Revenue Underneath

Results vs. Consensus

Two things need separating before any of these figures mean anything. The first is that sell-side coverage of Personalis stopped being maintained on July 20, when the merger was announced, guidance was pulled and the earnings call was cancelled. The second is that the headline revenue figure was not new information on August 4: the joint merger release had already disclosed preliminary second-quarter revenue of $22.4M and the 10,384-test clinical count. What follows is scored against consensus because that is the convention, not because the comparison carries its usual signal.

Consensus providers do not agree on the revenue line. One widely cited estimate implies $17.2M and a 30% surprise; another carries $16.69M and a 34% surprise. The range is $16.7M to $17.2M and the report uses the higher, more widely cited figure. On earnings per share the two agree closely, at $(0.25) and $(0.26) against a reported $(0.30). Only one analyst contributes a revenue estimate to the forward quarters in the data we can see, which is the more useful fact about this consensus than either point value.

Year-over-Year Comparison

Sequential Comparison

Q1 2026 figures above are the arithmetic difference between the six-month and second-quarter columns of the same Exhibit 99.1, and they reconcile exactly to the first-quarter statements as filed.

Revenue: growth that belongs to a customer, not a franchise

Reported growth of 30% is the headline. The composition is the story. Personalis discloses related-party revenue on the face of its income statement, and the Form 10-Q identifies two related parties: Tempus AI and Merck Sharp & Dohme, each holding more than 10% of the common stock. Related-party revenue was $8.4M in the quarter against $1.9M a year ago. Almost all of it is Merck: the 10-Q reports $8.3M invoiced to Merck under a master services agreement that dates to 2017 and was amended in February 2024 to add clinical laboratory services for Merck's clinical studies. Tempus, by contrast, contributed $0.1M of revenue.

Strip Merck out and the picture inverts. Revenue excluding Merck was $14.06M against $15.30M, a decline of 8.1%. Across the first half, revenue excluding Merck fell from $35.41M to $26.63M, a decline of 24.8%. Reported first-half revenue was $37.83M against $37.81M, which is to say the company has grown nothing at all in six months while the market has been marking it on a clinical-volume narrative.

Assessment: A single pharma customer whose spend can move by $6M in a quarter is not a growth engine, it is a timing variable, and the company's own risk language says as much when it flags "the timing of tissue, blood, and other specimen sample receipts from customers, which can materially impact revenue quarter-over-quarter." The clinical business is the one that deserves a multiple, and at $2.5M a quarter it is not yet large enough to carry the enterprise.

Margins: a sequential rebound off an ugly base

Gross margin of 21.2% is a large sequential improvement from the 1.8% Personalis printed in the first quarter, when $15.47M of revenue barely covered $15.19M of cost of revenue. That recovery is what operating leverage looks like in a lab business when volume returns. The year-over-year comparison is the more informative one, and it is negative: 21.2% against 27.6%, with cost of revenue growing 41.5% against 30.0% revenue growth.

The mix explains most of it. Clinical diagnostic testing is being delivered into Medicare rates on a cost base that has not yet scaled, and the revenue streams that carried better margin economics are the ones that shrank. Enterprise sales went to zero from $2.3M, and population sequencing declined 9.3%.

Assessment: Gross profit dollars have now been flat year over year for a quarter in which revenue grew 30%, which means the incremental revenue arrived at roughly zero gross margin. That is defensible during a reimbursement ramp and indefensible as a steady state. Without the merger, this line would be the central question for the next four quarters.

EPS and the shape of the loss

The $(0.30) loss per share is a $(0.07) deterioration year over year and a $(0.01) deterioration sequentially, and the sequential comparison understates the damage because the share count keeps rising. Weighted-average shares grew 19.1% year over year, and shares outstanding rose from 102.48M at year end to 106.37M at June 30, partly through at-the-market issuance.

Operating expenses are where the loss is manufactured. Selling, general and administrative expense grew 55.1% to $22.0M, which is 98.4% of revenue against 82.4% a year ago. Research and development grew 31.9% to $16.3M, or 73.1% of revenue. Combined operating expense of $38.3M ran at 171% of revenue.

Assessment: The $3.8M of fees paid to Tempus sits inside that SG&A line and explains a meaningful part of its growth, which makes the expense at least strategically legible. It does not make it affordable. A company spending 171% of revenue on operating expense while its gross profit dollars are flat is not on a path to profitability on its own balance sheet, and the merger is the resolution of that fact rather than a coincidence alongside it.

The Tempus Transaction: Terms, Mechanics and Spread

Everything above is context for the only variable that now sets the share price. On July 20, 2026 Personalis agreed to be acquired by Tempus AI in a two-step merger, and since that date the stock has traded on deal mechanics rather than results. The terms are unusual enough to be worth setting out precisely, because the phrase "$16.25 per share" describes a ceiling rather than a fixed price.

The exchange ratio is a cap, and the cap is the whole trade

Note that $16.25 divided by 0.3356 equals $48.42 exactly. The two regimes meet at that point by construction. Above $48.42 the ratio floats so that holders receive a fixed $16.25 of value regardless of where Tempus trades, and the security behaves like a cash deal with almost no sensitivity to the acquirer. At or below $48.42 the ratio locks at 0.3356 shares and holders take the acquirer's downside one-for-one.

Stated in the language of options, a Personalis holder is long a cash deal struck at $16.25 and short a put on Tempus struck at $48.42, with the $46.00 walk-right acting as a knock-out on that short put. This is not a collar. A collar would protect the seller's value on the downside; this structure protects the buyer's share count instead, and hands the downside to Personalis holders until the point where the board can walk away entirely.

The practical consequence is that the deal's value to a Personalis shareholder is a function of one number nobody at Personalis controls.

Tempus closed at $46.12 on August 6, which is 0.3% above the level that would hand the Personalis board a termination right. That proximity, reached within three weeks of signing, is the single best explanation for why the spread has stayed wide.

Where the spread sits

Against the August 18 close of $13.91, with Tempus at $49.36 and therefore above the Floor Price, the deal is worth $16.25 and the gross spread is 16.8%. Annualized, that is roughly 52% to a December 31 close and roughly 29% to a March 31 close. Those are not the numbers a market puts on a signed, board-approved, fully financed transaction it expects to complete.

Solving for the implied probability of completion requires an assumption about where the stock trades on a break. Using the unaffected 30-day VWAP of roughly $12.70 that the stated 28% premium implies, the market is pricing a 34% chance of closing. Using $11.00, which we think is closer to fair given how the standalone fundamentals have developed since, the implied probability is 55%. Either way the market is pricing something well short of confidence.

Personalis currently trades at about 84% of the value of the 0.3356 shares it would deliver at the cap. That 16% discount, over a deal expected to close within five months, is the compensation on offer for the combination of Tempus share-price risk, the walk-right scenario, and the ordinary tail of a shareholder vote.

The vote is the least of the risks

Closing requires the affirmative vote of a majority of outstanding shares, which is a demanding standard because abstentions count as votes against. It is nonetheless well covered.

Merck's Voting Agreement carries a termination trigger worth noting: it falls away on any amendment that reduces or adversely changes the consideration, extends the outside date, or adds conditions. A renegotiation downward, which is precisely what a sustained Tempus decline below $46.00 would invite, releases the largest committed vote at the moment it would be most needed.

Assessment: Regulatory risk looks modest. The only antitrust gate disclosed is the ordinary Hart-Scott-Rodino waiting period, and the two companies are already commercial partners rather than head-to-head competitors, which is the fact pattern that usually clears. Vote risk is manageable with a quarter of the register committed. The real exposure is that Personalis holders own a capped claim on a volatile acquirer, and that the contract gives their own board an exit but gives them no floor.

The process deserves scrutiny

Tempus's Schedule 13D/A sets out a sequence that is unusually compressed. Following confidential preliminary discussions, and "in response to the Issuer's request for Tempus to submit a proposal," Tempus submitted a non-binding proposal on July 19, 2026. The merger agreement was signed on July 20, 2026. One day separates the non-binding indication from the executed contract.

Against that, the Personalis chief executive described the outcome this way:

"After conducting an exhaustive process, we are confident Tempus' offer provides the most value to our shareholders and the fastest path to bringing Personalis' industry-leading tests to patients suffering from cancer."
— Chris Hall, Chief Executive Officer, Personalis

Both statements can be true if the exhaustive process preceded the request and Tempus was asked to submit at its conclusion. That is the ordinary reading. It still leaves a transaction in which a 12.5% holder and commercial partner acquires the company at a 6% premium to the last close, with no majority-of-the-minority condition among the closing conditions, and where the parties have said they intend to file a Schedule 13E-3, the going-private form used when an affiliate is on the buy side. At least one plaintiffs' firm announced an investigation on the day of the announcement, which is routine for a transaction with these features and is noted here as a governance datapoint rather than a legal view.

Assessment: The 28% premium to the unaffected 30-day VWAP is the more honest framing than the 6% premium to the last close, because the stock had run hard into the announcement. But shareholders who bought that run were told, in effect, that the run was the deal. The proxy and the Schedule 13E-3 will contain the background-of-the-merger section that settles whether the process was genuinely competitive, and that document is the one to read when it lands.

Revenue Composition

Personalis reports four named revenue streams plus an "other" line. Three of the four went the wrong way this quarter.

Pharma testing services

The largest stream at 75% of revenue, and the one that produced the quarter. Growth of 52.3% looks like commercial momentum until the related-party disclosure is applied: Merck accounted for $8.3M of the $16.8M. A year ago Merck was $1.9M of $11.0M. Excluding Merck, pharma testing services declined from $9.1M to $8.5M, a fall of 6.9%.

The company frames this stream slightly differently in its release, combining pharma testing with the "other" line to report "$16.8 million compared with $11.1 million, a 51% increase." That arithmetic is correct on its own terms. It does not change the concentration underneath.

Assessment: Pharma testing revenue recognized on sample receipt is lumpy by nature, and a single customer moving from $1.9M to $8.3M in a year is as likely to reflect the timing of Merck's clinical study enrollment as any durable share gain. The base business here is shrinking modestly. Treat the $8.3M as a contract milestone, not a run rate.

Clinical diagnostic

The stream the equity story rests on, and the only one where the growth is unambiguously franchise-driven. Revenue of $2.541M grew 441.8% year over year and 77.6% sequentially. Volume reached 10,384 tests against 3,478, and more than 1,400 physicians have now ordered NeXT Personal. Note that the release headline rounds this line to "$2.6 million" while the supplemental revenue table in the same exhibit reports $2,541 thousand; the table figure is used throughout this report.

The more interesting number is price. Revenue per clinical test was roughly $245 in the quarter against roughly $135 a year ago, an increase of about 82%. That is what Medicare coverage does to a diagnostic: it converts tests that were previously delivered for research or at negligible realization into reimbursed volume.

Assessment: This is a real inflection and it is the asset Tempus is buying. It is also still only 11.4% of revenue, and it is being bought at a moment when the reimbursement ramp has begun but the operating leverage has not. Whoever owns it in 2028 is likely to own something considerably more valuable than $2.5M a quarter. That is an argument about the price of the deal, not about whether the deal makes sense.

Population sequencing

Revenue of $3.0M declined 9.3% year over year while recovering 20.0% sequentially. This is the VA Million Veteran Program contract, a mature government relationship rather than a growth vector.

Assessment: Directionally down and structurally finite. The first-half decline is steeper than the quarterly one, from $7.5M to $5.5M, and nothing in the disclosure suggests a reversal. Model it as a decaying annuity.

Enterprise sales

Zero, against $2.3M in the year-ago quarter and $0.4M in the first quarter. An entire revenue line has been extinguished within twelve months, and the release does not mention it.

Assessment: A stream going to exactly zero is either a discontinued offering or a lost customer, and shareholders were told which by neither the release nor a call. On its own the $2.3M is not material to the thesis. As an indicator of disclosure posture in a quarter with no call and no guidance, it is worth more than its dollar value.

Key KPIs

The Concentration Problem

The related-party footnotes are the most informative disclosure in this filing set, and they describe a business whose two largest counterparties both sit on its share register.

Merck became a related party in December 2024 by buying 14,044,943 shares for $50.0M at $3.56 per share. Its commercial relationship is older, running to a 2017 master services agreement amended in February 2024 to cover clinical laboratory services for Merck's studies. Merck was invoiced $8.3M in the quarter and $11.2M in the half, and owed $7.4M at June 30 against $2.0M at year end. That receivable is 49% of the company's entire $15.1M accounts receivable balance.

Tempus acquired its stake in August 2024 by exercising two warrants for 9,218,800 shares at an average exercise price of $2.00, issued in November 2023 in consideration of its obligations under the commercialization agreement, and buying a further 3,500,000 shares at $5.07. Under that agreement Tempus markets NeXT Personal Dx in the United States, Personalis performs and bills the tests, and Personalis compensates Tempus per test for orders obtained and results delivered, plus co-promotion fees running through the end of 2026.

The economics of that arrangement are worth stating plainly. In the second quarter Personalis paid Tempus $3.805M in order, results-delivery and net promotional fees, and recognized $2.541M of clinical diagnostic revenue. The channel cost 1.5 times what the channel produced. The fee line is not purely variable, since it includes co-promotion payments that run to a fixed end date, so this is not a per-test gross margin. It is still the case that the flagship product's route to market is currently value-destructive at the reported scale, and that the fee grew 307% year over year while clinical revenue grew 442% off a much smaller base.

One further term deserves attention. Tempus receives the initial and longitudinal genomic data generated by these tests and holds the right to use it, and if Tempus licenses that data onward Personalis is entitled to 10% to 20% of Tempus's gross license revenue, during the six-year term and for ten years after. No revenue share has been recognized in any of the four periods presented.

Assessment: A company that sells 37% of its revenue to one shareholder and buys its primary distribution from another has limited independent bargaining power, and that is the structural condition in which a 6% premium becomes acceptable to a board. The data-licensing royalty is the clearest example: it is an asset of real potential value that has never once produced a dollar, and it now transfers to the counterparty that controls whether it ever does. None of this makes the merger improper. It does explain why the standalone alternative was unattractive enough that the board asked Tempus to bid.

Key Topics & Management Commentary

Overall Management Tone: With no call to read, the posture has to be inferred from what the company chose to publish, and the choice was narrow. The release leads with the three strongest operational facts, carries one prepared executive quote, discloses the guidance and call withdrawal in a single sentence under a heading that still reads "Full Year 2026 Outlook," and says nothing about the revenue line that went to zero, the margin decline, or the burn. That is a company in deal-pendency mode rather than one hiding a problem, but the practical effect on shareholders is the same: the disclosures that would have been pressed on a call were not volunteered in print.

Guidance and the earnings call were withdrawn together

The most consequential sentence in the release is administrative. Personalis eliminated both of the mechanisms through which shareholders normally interrogate a quarter, in one line, under a heading that promises an outlook and then declines to give one.

"As a result of the announcement on July 20, 2026 that Personalis and Tempus AI, Inc. have entered into an Agreement and Plan of Merger, Personalis will no longer provide financial guidance or conduct a quarterly earnings conference call."
— Personalis, Inc., Form 8-K Exhibit 99.1, August 4, 2026

This is common practice for a company under a signed merger agreement and is not a governance failure in itself. It does change what a shareholder can know. There will be no management framing of the margin trajectory, no opportunity to ask why enterprise sales stopped, and no forward revenue anchor between now and closing, which on the company's own timetable could be as late as April 2027 before extensions.

Assessment: Withdrawing guidance is defensible; withdrawing the call as well removes the one venue where the related-party concentration would almost certainly have been raised. For an investor underwriting the break scenario, this quarter is the last relatively complete picture of the standalone business they will get.

Medicare coverage reached four indications

Two coverage decisions landed in and around the quarter, taking NeXT Personal from two reimbursed indications to four. The company received approval for immunotherapy monitoring in late-stage solid tumors during the quarter, and for neoadjuvant therapy monitoring in Stage II-III triple-negative and HER2-positive breast cancer shortly after it, which reconciles with the July 20 merger release describing coverage in three indications at that date.

"Received Medicare coverage approval for NeXT Personal for monitoring treatment response to neoadjuvant therapy (NAT) in patients diagnosed with Stage II-III Triple-Negative Breast Cancer (TNBC) or HER2-positive (HER2+) breast cancer."
— Personalis, Inc., Form 8-K Exhibit 99.1, August 4, 2026

Assessment: This is the highest-value operational development of the quarter and it flows directly into realized price, which roughly doubled per test. Each incremental indication expands the addressable ordering population without requiring a new assay, which is the leverage that makes MRD diagnostics attractive. It is also the clearest evidence that Tempus is buying the franchise at the beginning of its reimbursement curve rather than the end.

Clinical volume inflected, and so did realized price

Volume of 10,384 tests represents 199% year-over-year and 33% sequential growth, and the ordering base has passed 1,400 physicians. The price effect is the part that has not been widely discussed: revenue per test moved from roughly $135 to roughly $245.

"We delivered another exceptional quarter of growth, highlighted by a 199% year-over-year and 33% sequential increase in clinical volume, with more than 1,400 physicians ordering NeXT Personal,"
— Chris Hall, Chief Executive Officer, Personalis

Assessment: Volume and price rising together is the signature of a diagnostic crossing from early adoption into reimbursed routine use, and it is genuinely rare. If this stream were standalone and growing at this rate, it would command a valuation that makes $16.25 look inadequate. It is not standalone, it is 11.4% of revenue, and it is attached to a cost base losing $33.6M a quarter.

The clinical revenue commitment survived the deal

Management repeated a full-year target even while withdrawing formal guidance, which is a meaningful distinction: this is the one forward number shareholders were given.

"We remain firmly on-track to achieve more than a 500% increase in our clinical revenue over last year."
— Chris Hall, Chief Executive Officer, Personalis

The first-half evidence supports it. Clinical diagnostic revenue of $3.972M in the six months against $0.777M a year earlier is already a 411% increase, so the full-year commitment does not require heroics from here, only continuation.

Assessment: Credible, and worth crediting. It is also the narrowest possible forward disclosure: a growth rate on the smallest of four revenue streams, with nothing said about the 75% of revenue that is pharma testing or about consolidated revenue, margin or burn. Take the target at face value and it still tells you almost nothing about 2026 revenue.

The VICTORI colorectal readout

The strongest clinical datapoint disclosed was from a prospective investigator-led study in colorectal cancer.

"The prospective VICTORI study led by the University of British Columbia showed NeXT Personal detected 100% of all patient relapses, including all distant metastases in historically difficult-to-detect regions like the lung. Notably, just four weeks after surgery, NeXT Personal detected over 80% of patients who later relapsed, providing clinicians with an early signal of cancer to inform treatment pathways."
— Personalis, Inc., Form 8-K Exhibit 99.1, August 4, 2026

Assessment: 100% sensitivity for relapse is the number that gets a test into guidelines, and detection at the four-week landmark is what makes it clinically actionable rather than merely prognostic. The release does not give the cohort size, the specificity, or the false-positive rate, and a prospective study led by a single academic centre is not a registrational dataset. Directionally strong, insufficiently specified to underwrite.

Sub-10 ppm sensitivity as competitive positioning

The second data disclosure is aimed squarely at the competitive question of whether ultrasensitive detection matters.

"Approximately 21% of pre-operative adenocarcinoma and 18% of post-operative landmark detections in the TRACERx study were below 10 ppm—thresholds frequently missed by less sensitive assays. Patients detected in this range experienced a three-fold increased risk of recurrence compared to patients with undetectable ctDNA, potentially enabling much earlier clinical intervention."
— Personalis, Inc., Form 8-K Exhibit 99.1, August 4, 2026

Assessment: This is the most strategically important claim in the release, because it is the argument that NeXT Personal is not substitutable by cheaper MRD assays. If roughly a fifth of true positives sit below the detection floor of competing tests, and those patients carry three times the recurrence risk, then sensitivity is not a specification contest but a clinical one. It is also the claim most in need of independent replication, and the company's own forward-looking language concedes the risk that the assay may not remain "unique in its ability to detect traces of cancer in the ultrasensitive range."

Operating expense outran revenue by a wide margin

Selling, general and administrative expense grew 55.1% to $22.0M against 30.0% revenue growth, and research and development grew 31.9% to $16.3M. Combined, operating expense reached $38.3M against $22.4M of revenue. SG&A alone was 98.4% of revenue against 82.4% a year ago.

Part of the SG&A increase is explained: fees paid to Tempus rose from $0.935M to $3.805M, contributing roughly $2.9M of the $7.8M increase. The balance is not broken out, and with no call there was no opportunity to ask.

Assessment: Roughly 37% of the SG&A growth traces to the commercialization agreement, which is at least a deliberate investment in the clinical channel. The remaining $4.9M is unexplained in a quarter where the company also disclosed it would stop explaining things. Growing SG&A by 55% into a flat first-half revenue base is the decision a management team makes when it expects to be someone else's cost line before the bill comes due.

Cash burn accelerated and the equity facility stayed open

The cash flow statement is the disclosure that best contradicts the release's "Strong Cash Position" framing.

Cash and short-term investments ended at $212.66M against $239.95M at December 31, a decline of $27.3M that is materially smaller than the $55.2M of free cash burn because $29.4M of it was replaced with equity. Stock-based compensation of $6.5M against a $61.7M net loss means this loss is almost entirely cash. Annualizing the first half implies roughly $110M of yearly free cash burn, or about 1.9 years of runway on the current balance.

Assessment: The company describes $212.7M as a strong cash position and in absolute terms it is. Against a burn that grew 56% while revenue grew 0.1%, it is under two years. That is the number that makes the standalone alternative unattractive and, in our view, the number that best explains why the board solicited a bid at a 6% premium rather than continuing alone.

Gross margin: sequential recovery, structural decline

Gross margin of 21.2% against 1.8% in the first quarter is the headline improvement. Against 27.6% a year ago it is a 643 basis point decline, and gross profit dollars were flat at $4.74M versus $4.76M despite $5.15M of additional revenue. The incremental revenue in the quarter carried essentially no gross margin.

Assessment: Two forces are working against each other. Clinical mix is dilutive today because reimbursed volume is being delivered on an unscaled cost base, and it should be accretive later as sequencing cost per test falls against a fixed Medicare rate. Meanwhile the higher-margin streams are the ones disappearing. Whether the crossover arrives is the single most important operational question about this asset, and it is now a question for Tempus's shareholders rather than these ones.

Two revenue streams kept shrinking, and neither was addressed

Enterprise sales went from $2.3M a year ago to $0.4M in the first quarter to zero in the second. Population sequencing, the VA Million Veteran Program work, declined 9.3% year over year and 26.9% across the first half, from $7.5M to $5.5M. Neither decline appears in the release's operational highlights or the CEO commentary.

Assessment: Together these two streams have removed roughly $6.4M of first-half revenue year over year, which is most of the reason first-half revenue was flat despite the Merck ramp and the clinical inflection. A shareholder reading only the bullets would conclude the business grew 30%. A shareholder reading the supplemental revenue table would conclude two of five lines are in terminal decline and a third depends on one customer's study calendar.

Guidance & Outlook

There is no guidance. Personalis withdrew it on August 4 in connection with the merger, and the "Full Year 2026 Outlook" heading in the release contains only the withdrawal sentence.

What the Street is left with: the one externally maintained set of estimates we can see carries roughly $22.9M for the third quarter and roughly $79.1M for the full year, against $37.8M delivered in the first half. That full-year figure requires $41.2M in the second half, or 9% growth over the first half. Given that the first half was flat year over year and that $8.3M of the second quarter came from a single customer's study timing, the risk around that number is wider than a point estimate suggests.

Guidance style: not assessable. The company has removed the instrument.

What They're NOT Saying

1. That the growth is one customer. The release attributes revenue growth to "higher pharma testing services and growth in clinical tests." It never mentions that Merck accounted for $8.3M of the $16.8M pharma line, or that revenue excluding Merck declined 8% year over year. The disclosure exists, in a footnote on the face of the income statement and in Note 8 of the Form 10-Q, but it is not in the narrative.
2. That an entire revenue line went to zero. Enterprise sales fell from $2.3M to nil. There is no mention of it in the highlights, the financial results bullets, or the CEO commentary. No explanation was offered for whether the offering was discontinued or a customer was lost.
3. That gross margin fell 643 basis points. The word margin does not appear in the release. In a quarter where gross profit dollars were flat on 30% revenue growth, the omission is conspicuous.
4. That operating cash burn grew 56% on flat first-half revenue. The release says the company "ended the quarter with approximately $212.7 million in cash, cash equivalents, and short-term investments" under the heading "Strong Cash Position." It does not say that $48.2M of operating cash left the business in six months, that capex nearly tripled, or that $25.5M of the gap was funded by selling stock.
5. What the Tempus channel actually costs. The $3.8M paid to Tempus this quarter, against $2.5M of clinical revenue, is disclosed only in Note 8 of the Form 10-Q. It is the single most important unit-economic fact about the flagship product and it appears nowhere in the earnings release.
6. Why the board asked Tempus to bid. The 13D/A discloses that the proposal came "in response to the Issuer's request," one day before signing. The joint release describes "an exhaustive process." Neither document reconciles the two, and the background-of-the-merger section that would is not yet filed.
7. What happens if Tempus trades below $46.00. The company has not said whether it would exercise its termination right, renegotiate, or close anyway. With Tempus having closed within 0.3% of that level on August 6, this is not a hypothetical, and shareholders have no stated board policy to price against.
8. Anything about the data-licensing royalty. The 10% to 20% share of Tempus's gross data-license revenue has produced nothing in any period presented. Whether that reflects no licensing activity, or licensing that has not been attributed, is not addressed.

Market Reaction

• Pre-print setup: Personalis closed at $13.52 on August 4, the day of the release. The stock entered the print up 69.8% year to date and 142.7% over twelve months, against a 52-week closing range of $4.41 to $15.79, but down 0.7% over the trailing 30 days. The S&P 500 was up 13.0% year to date.
• Release timing: after the close. The Form 8-K was accepted at 4:04 PM ET on August 4, making August 5 the reaction session.
• Reaction session: the stock opened at $13.44, traded between $13.35 and $13.73, and closed at $13.52. That is a move of exactly zero, against an S&P 500 that fell 0.2%.
• Volume: 1.5M shares against a 30-day average of 5.0M, or 0.3x. Turnover collapsed rather than spiked.
• Subsequent drift: $13.03 on August 6, then a recovery to $14.50 by August 10 and $13.91 by August 18, tracking Tempus rather than anything company-specific.

A stock that closes a post-earnings session unchanged to the cent, on one-third of normal volume, after reporting 30% revenue growth and a 20% EPS miss, is telling you the print carried no information. In this case that is literally true. Preliminary second-quarter revenue of $22.4M and the 10,384-test count had been published on July 20 in the joint merger release, sixteen days earlier. By August 5 there was nothing left to discover except the margin and cash detail, and the market plainly did not re-underwrite on it.

The move that mattered happened three weeks earlier. On July 20, the day the merger was announced, Personalis fell 13.3% from $15.39 to $13.34 while Tempus fell 7.7%. A target falling on its own takeover is unusual and here it is fully explained by the structure: the consideration is Tempus stock at a capped ratio, so a decline in the acquirer is a decline in the deal, and the 6% premium was too thin to absorb it. Shareholders who had bid the stock to a 52-week high of $15.79 on July 15 discovered that the takeout they were anticipating was worth less than the price they had paid.

Since then the correlation has been close to mechanical. Tempus bottomed at $41.55 on July 29 and Personalis bottomed at $11.73 the same day. Tempus recovered to $52.05 on August 7 and Personalis to $14.16 on the same session. Tempus closed at $46.12 on August 6, within 0.3% of the $46.00 level that would give the Personalis board a termination right, which is the closest this deal has come to breaking and is the event that reset the spread wider. Personalis has not traded on Personalis news since July 20.

The August 19 session was still in progress at publication with both names sharply higher, and no intraday level is relied on anywhere in this note. All spread math above is struck at the August 18 close.

Street Perspective

Debate: Is a 17% spread compensation or a warning?

Bull view: The bull case on the Street is that this is a mispriced arbitrage. The transaction is signed, board-approved on both sides, carries no financing condition, faces only an ordinary Hart-Scott-Rodino waiting period between two firms that are partners rather than competitors, and has roughly a quarter of the share register effectively committed to vote for it. A 17% gross spread on that fact pattern annualizes above 50% to a year-end close, which is several times what comparable signed strategic deals pay.

Bear view: The bear camp argues the spread is priced correctly because the risk is not completion but consideration. Holders do not own a claim on $16.25; they own a claim on 0.3356 Tempus shares once Tempus trades below $48.42, and Tempus has moved more than 12% in a single session twice in the last month. The wide spread reflects the cost of hedging an acquirer with that volatility, plus the genuine possibility that a sustained decline below $46.00 triggers a renegotiation that also releases the largest committed vote.

Our take: The bears have the better of it on mechanics and the bulls on outcome. The spread is not mispriced so much as mis-described: it is not deal-break compensation, it is the premium for being short a put on Tempus struck at $48.42. If you want the $16.25 you must either hedge the acquirer or hold a view on it, and most long-only holders are doing neither.

Debate: Is the MRD franchise worth more than $16.25?

Bull view: The optimistic case holds that the board sold at the wrong moment. Clinical volume tripled year over year, realized price per test roughly doubled, Medicare coverage went from two indications to four inside a quarter, and the TRACERx analysis suggests the assay's ultrasensitivity is a durable clinical differentiator rather than a specification. Selling a franchise at that inflection, at a 6% premium to the last close, transfers the entire reimbursement ramp to the buyer.

Bear view: The skeptical case is that the franchise cannot be valued independently of the company carrying it. Clinical diagnostics is 11.4% of revenue, the channel that delivers it costs $3.8M a quarter against $2.5M of revenue, consolidated gross profit dollars are flat, and free cash burn of roughly $110M annualized against $212.7M of liquidity leaves under two years. A franchise that requires another two capital raises to reach scale is worth less in the hands of a seller than a buyer.

Our take: Both are right about different time horizons, and the bear case governs the decision. The MRD asset is probably worth more than $16.25 in 2029 and demonstrably not fundable to 2029 on this balance sheet without material dilution. That gap is the acquirer's profit, and it was created by the burn rate, not by the process.

Debate: Where does the stock trade if the deal breaks?

Bull view: One camp argues the downside is contained near the unaffected 30-day VWAP of roughly $12.70, because the operational news since has been good: two new Medicare indications, record volume, and the strongest clinical dataset the company has published. On this view a break costs about 9% from the current price, which makes the risk-reward strongly favourable.

Bear view: The other camp notes that the unaffected VWAP already embedded takeover speculation, that the standalone disclosures released since are materially worse than what was known then, and that a company whose acquirer walked away, whose largest customer relationship is a study-timing artifact, and which must return to the equity market within two years would not re-clear at $12.70. This camp underwrites $9 to $11.

Our take: We are with the second camp, and it is the reason for the rating. The break scenario is not a return to the status quo ante; it is a return to a worse company with a failed process behind it and a depleted register in front of it. Using $11.00 as the break price, the market is discounting a 55% chance of completion, which is roughly where we would put it once the Tempus share-price condition is treated as a real condition rather than a formality.

Model Framework

We do not carry a published model on Personalis and are not initiating one, because a company under a signed merger agreement with no guidance and no call does not reward a projection exercise. The parameters below are the working assumptions behind the rating rather than a forecast, and the valuation that matters is contractual rather than fundamental.

Valuation: the fair value of the equity is the probability-weighted deal outcome, not a multiple. At a 55% chance of completion at $16.25 and a $11.00 break, expected value is $13.89, which is where the stock closed on August 18. The market is efficiently priced on our own assumptions, which is precisely why this is a Hold rather than a call in either direction. Moving to Outperform requires either conviction that completion odds exceed 70%, or that Tempus itself is undervalued, since holders who do not sell into the close will own Tempus shares.

Thesis Scorecard Post-Earnings

This is our initiation on Personalis, so the pillars below are established here rather than carried forward, and this quarter is their first scoring.

Overall: Thesis established. Every pillar scored as confirmed this quarter, which is unusual and reflects that the merger agreement resolved most of the uncertainty that would normally sit in a first read. The bull pillars support the price floor; the bear pillars cap what the equity can be worth and explain why the spread has not compressed.

Action: Hold. The contract caps the upside at roughly 17% and the fundamentals set the downside near $11. That is a fair coin, correctly priced, in a security whose terminal state is ownership of a different company's stock. We would revisit toward Outperform on either a sustained move in Tempus well clear of the Floor Price, which converts this into a near-cash-equivalent spread, or a widening of the discount toward 25% without a change in the documents. We would revisit toward Underperform if Tempus trades below $46.00 into the measurement window, since a renegotiation would reset the price and simultaneously release Merck's voting commitment.