Tempus is a cancer-testing lab with a data library
Tumour tests
read a tumour's DNA and RNA
Hereditary tests
inherited cancer risk
Data library
drugmakers pay to use it
Tempus reads a tumour's genes and reports the drugs and trials it may respond to. Insurers pay per test, and drugmakers pay to use the resulting data.
The company · 2 · What exactly does it sell?
Testing is three-quarters, data a quarter
Each $1 of sales, 2025
47%28%25%
Cancer and other testsHereditary testsData for drugmakers
Insurers pay for the tests; drugmakers pay to use the data they produce.
The company · 3 · How does it make money?
Paid per test and per data deal
$1.51Bsales in the past year
About 55%of tests it was paid for, 2023 and 2024
20.4%of data sales from Pathos, the founder's firm, in Q3
Insurers often refuse to pay for a test, so many are run for nothing.
Is it a good business? · 4 · What gives it an edge?
Tests feed data that pays for more tests
About 367,500
cancer samples read in 2025
More than 5,000
hospitals and clinics linked in
87%
of regular doctors still order a year later
Data customers spent 26% more in 2025, on the company's own measures.
Is it a good business? · 5 · What is the catch?
Roche owns the two biggest rivals
Rocheowns Foundation Medicine and Flatiron
Carissells the same tissue report
AstraZenecagot an AI model Tempus built
Listed rivals: ROP.SW · CAI · AZN · NTRA
Tempus's own AI model, delivered to a customer, may become a competitor.
Is it a good business? · 6 · What does its record show?
Revenue quadrupled; losses narrowed
Sales, $B
0.3
0.5
0.7
1.3
1.5
Dec 2022Dec 2023Dec 2024Dec 2025last 12 mo
Sales$321M → $1.51Bsales, 2022 to the past year
Losses−83% → −18%operating profit per $1
Growth+83%sales growth in 2025, much bought
In the last year selling costs rose per dollar, pay in shares doubled and the operating loss widened.
Is it a good business? · 7 · How much room is there to grow?
Testing is still rare
About 2.11 millionAmericans diagnosed with cancer in 2026
About $3.4Ba year at one test each
91.4%of Medicare cancer patients never tested
Tempus was paid on only about 55% of the tests it ran in 2023 and 2024.
Is it a good business? · 8 · Who is in charge?
The founder controls the vote
2015Eric Lefkofsky founded Tempus
57.9%of the vote he holds, with 21.4% of the shares
Pathoshis other firm, a big data customer
He also chairs Pathos, which supplies about a fifth of data sales.
The story right now · 9 · Why do drugmakers buy data?
Drugmakers pay for real patients' data
26%more spent by data customers in 2025
$42M → $14MRecursion's yearly data fee
21%share jump on another firm's result
Drugmakers use genetic data from real patients to pick which drugs to develop and whom to test them on. Tempus sells that data, and AI makes it more useful.
The story right now · 10 · How fast does Tempus really grow?
Much growth was bought or related
83.0% reported growth2025.
33.5% without the purchaseOne February sentence.
A fifth of data from PathosThe founder's firm.
A big data deal endsUnless AstraZeneca extends.
Data licensing above 30% without Pathos, and AstraZeneca extending, would prove us wrong.
The story right now · 11 · What changed lately?
Sales up 23%, loss narrowed
$411.5Msales, up 23.1%
$56.9Moperating loss, narrower
$22.1Mof data sales from Pathos
24.7% to 26.1%fourth-quarter growth the forecast needs
Mixed quarter
The full-year forecast rose again. Personalis holders vote on 30 November.
The story right now · 12 · What is everyone arguing about?
One February sentence separates 83.0% reported growth from 33.5% without the purchase.
The price and our view · 13 · What is a share worth?
The price sits far above our best case
$3.22what we think a share is worth · the last close is of that
$3.22our value
Attractive below $2.58
Neutral
Unattractive above $3.86
$1.53gloomy: growth halves, less kept per $1
$3.22our estimate: experts' path, then slowing
$4.74sunny: growth beats the forecast by a quarter
Value per share in three futures. None is a prediction of the share price.
We estimate what a share is worth from the cash the business should earn over many years. Our rule: buy only when the price is at least a fifth below our estimate, as a cushion for being wrong.
The price and our view · 14 · What if it all works?
As a bet on the data platform, it is fragile
Priced for successthe kind of bet this is
1.2×if it works
54%kept if it fails
Multiples of the company's value when the bet was last struck, 30 Sep 2026.
It is worth a little more if the plan works and about half if it fails.
The price and our view · 15 · How could we be wrong?
A renewal that decides the case
AstraZeneca renewalBiggest
Ends at the buyer's choice.
Revenue from the founder's firmShowing now
Pathos.
Getting paidWatch
About half of tests.
Shares for PersonalisUnknown
Depends on the price.
Its own AI model, delivered to a drugmaker, could become a competitor.
The price and our view · 16 · How will we know?
February tests growth without Pathos
Data sales above 30% without PathosGood sign
AstraZeneca extendsGood sign
Another data customer cuts its feeBad sign
FEB 2027-
Fourth-quarter and full-year 2026 results, February. Companies report every three months; each report is a test of the story.
Our view also changes on price alone: if the shares fall to within a fifth of our estimate, we move to Neutral.
The price and our view · 17 · So, what do we think?
A large cancer lab, priced as the platform it hopes to be
$3.22our value
Attractive below $2.58
Neutral
Unattractive above $3.86
UnattractiveHigh conviction
Tempus runs one of America's largest cancer-testing labs, and every test feeds a data library drugmakers pay for. But much growth was bought, a big data deal ends soon and a fifth of data sales come from a related company.
↑ ↓ to move · C for contents
What it does
More about the company
Who it is
When someone is diagnosed with cancer, the oncologist has to choose a treatment. Tempus reads the DNA and RNA of the tumour and sends back a report naming the drugs and trials that tumour is likely to respond to. Its main test won FDA approval in 2023.
It sequenced about 367,500 cancer samples in 2025, against 170,300 cases at Caris, a rival. About 205 sales representatives sell into more than 5,000 hospitals and clinics.
How it makes money
Two ways. Medicare or an insurer pays for each test, about $1,600 on average in 2025, although Tempus was paid on only about 55% of the tests it ran in 2023 and 2024.
Each test also yields an anonymised record linking the tumour to the patient's treatment and outcome, which drugmakers license in bulk. Data and applications brought in $93.2M in the June quarter.
Growth and profit
What growth costs
For years, growing made Tempus less loss-making: research and overhead costs grew slower than sales, helped by a large purchase. In the past year that stopped. Overhead and pay in shares grew as fast as sales or faster, so the loss from running the business got slightly wider rather than narrower.
Its record
The numbers
Is it growing?Fast
Sales grew 37% over the twelve months to Sep 2026, slower than the 83% in the year to Dec 2025.
Is it profitable?Losing
Operating margin is −18%, up from −37% in the year to Dec 2023.
How steady is it?Swingy
Yearly sales growth has ranged from +30% to +83% over the last 3 years.
Does it earn good returns?Poor
Shareholders' money lost 38% a year, up from −89% in the year to Dec 2025.
How long can it keep going?Long runway
It used $137M of cash over the last twelve months, so the $590M it holds lasts about 4.3 years at that pace.
Is the balance sheet safe?Net cash
It holds $590M of cash against $176M of debt, $415M more cash than it owes.
Are shares being diluted?Diluting
The share count rose 188% in 3.8 years, and pay in shares costs 13% of sales.
Why it wins
The evidence
The two halves feed each other. Every test adds a record; links into more than 5,000 hospitals and clinics let Tempus join it to what happened to the patient; drugmakers pay for that joined record; and their money funds the software in the report that brings doctors back.
On the company's own measures, data customers spent 26% more in 2025 than in 2024, and 87% of doctors who have ordered more than five tests are still ordering a year later.
Who could beat it
The rivals in full
Roche owns the two biggest rivals, Foundation Medicine in testing and Flatiron Health in cancer data, and neither publishes accounts. Caris sells the same tissue report. The subtler threat is Tempus's own making: it delivered an AI cancer model to AstraZeneca, and its risk factors warn that a drugmaker with a model may stop licensing data.
Each rival
Roche's Foundation Medicine, FoundationOne CDxROP.SW
The tissue sequencing report the oncologist orders to choose a therapy - the same order, placed with a different laboratory.
Caris Life Sciences, MI Cancer SeekCAI
The same tissue report, sold to the same oncologist, with its own algorithms attached to it.
The oncology foundation model Tempus itself built for AstraZenecaAZN
The recurring data licence a drugmaker buys each year, once it owns a model trained on that data.
Natera's SignateraNTRA
The recurrence-monitoring test Tempus is buying Personalis in order to sell.
How big it could get
The market and who has signed
The market
About 2.11 million Americans will be diagnosed with cancer in 2026. At about $1,600 a test, one test each would be about $3.4B a year. Testing is still rare: 91.4% of Medicare cancer patients from 2016 to 2023 had no genomic test at all.
Who has signed
AstraZeneca has committed at least $220M through December 2026, rising to $320M through 2028 if it chooses. Pathos AI, chaired by Tempus's founder, owes $200M over three years and may pay half in its own private shares. Other signed data contracts held more than $1.1B at the end of 2025, but customers can cancel for convenience.
Who runs it
Leaders, pay and the cash
Who runs it
Eric Lefkofsky founded Tempus in 2015 and still runs it; James Rogers has run its finances since 2021. Lefkofsky owns about a fifth of the shares but controls the vote through special shares. Most executive pay rides on sales growth, counting companies bought, and on the share price; nothing rewards cash or profit per share. In 2025 the board paid most of a bonus whose profit goal was missed.
What they did with the money
It is buying Personalis for about $1.5B, mostly in its own shares: about 17.9 million new shares at the 25 September price, up to 31.5 million if its price falls, adding 9.9% to 17.3% to the count. Staff were paid $106.8M in stock in the first half, and $1.17B of convertible debt sits against $599.6M of cash.
The last report
The latest report
Tempus's sales grew 23% to $411.5M and its operating loss narrowed to $56.9M
What it did to the thesis Mixed quarter
It did not move the Unattractive view. Growth and the loss both improved, but the loss is still far from the profit margin the price needs, and Pathos still paid more than the line we set.
The report
The September quarter showed sales up 23.1% to $411.5M, faster than June's 21.6%. Data and applications grew 32.9%, and Pathos, a company whose executive chairman is Tempus's own chief executive, supplied $22.1M of it, 20.4%, down from 24.2% in the first half.
What's happening now
The background
The shares jumped 21% on 19 August on a vaccine result from Merck and Moderna, not on anything Tempus published. Since then two filings have cut the other way: Recursion, a data customer with no tie to management, cut its yearly fee from $42M to $14M, and Tempus's own projections have Personalis losing money until 2030.
The September-quarter results of November 2 raised the forecast and narrowed the loss without changing what the price needs. The next test is the fourth-quarter and full-year results, due about February; Personalis holders vote on November 30.
Where we disagree
Both sides in full
The market
The AI-native precision-medicine platform: data licensing compounds, the network widens the lead, and 25% growth runs for years.
We believe
The growth is slower and lower in quality than the headline: bought in 2025, partly related-party in the data line, slowing elsewhere.
The argument
Seeing that takes reading: a related-party note, and one sentence in a February release separating 83.0% reported growth from 33.5% without the acquisition.
We are wrong if data licensing grows above 30% without Pathos, test volume holds above 30% growth, and AstraZeneca extends to 2028.
What it's worth
How we value it
Valued on the analysts' revenue path for three years, each later year closing 30% of the gap to 4%, on official operating profit with stock pay a cost and tax paid in full; the thesis's own evidence is the check on that path.
The three cases
Bear case$1.53
Growth runs at half the analysts' path, 12%, 11% and 7%, the margin settles a fifth lower at 1.6%, and new capital earns only the 8% rate: the downturn test.
Our value$3.22
Revenue grows 25%, 22% and 13% over three years, the analysts' path (C73), then fades toward 4%, closing 30% of the gap each year; the margin moves to 2.0%, the analysts' year-three figure after stock pay and on the filed basis (7.5% less).
Bull case$4.74
Growth runs a quarter above the analysts' path, 31%, 28% and 16%, the margin reaches 2.3%, and new capital earns 15%.
The bet
Seen as a bet
Even if this works, the shares look expensive. The best case we can defend is management's own plan: sales growing by about a quarter each year to 2028, helped by higher test prices and faster data sales. At what rival labs are worth for their sales, that would still be worth only a little more than today's price.
If it fails, a real testing business remains, but its growth looks like its rivals' and a quarter of its data sales come from a company its own chief executive chairs.
What could go wrong
Every risk
The risks
A renewal that decides the case
AstraZeneca's agreement ends on 31 December 2026, and extending it is AstraZeneca's choice.
Revenue from the founder's company
Pathos supplied 24.2% of the data line in the first half, and may pay up to half its fees in private shares.
Its own model as competitor
A drugmaker that owns a model trained on Tempus data may stop licensing the data.
Getting paid
Tempus was paid on only about 55% of its cancer tests, and Medicare pays nationally only for FDA-approved uses.
What we still do not know
Whether AstraZeneca extends its agreement to 2028.
How many new shares the Personalis deal creates; that depends on Tempus's price before it closes.
Whether the $80M to $100M the chief executive expects from Medicare pricing arrives. It is in no filing and not in the forecast.
What changes our view
Every route to a new view
Every route to a new view
→ Neutral
The share price falls to about $3.86, back within 20% of what we think the business is worth.
→ Attractive
The share price falls to about $2.58, 80% of what we think the business is worth.
→ Attractive
The operating loss narrows four quarters running while sales grow above 20%, with growth excluding acquisitions above 30% in two of them. Pathos falls below a tenth of the data line while that line still grows above 25%.
Research reopens
AstraZeneca extends, taking its committed spend to $320M through December 2028. Tempus reports a full year of positive free cash flow after counting stock pay.
The next test
It supports the case if
Q4 revenue above $470M with data and applications growing above 30% and Pathos revenue below $20M in the quarter.
It pressures the case if
Q4 revenue below $458M, or the guide met only because Pathos revenue rose again.
The verdict
Scores and how we weighed it
The scores
Growth
3 of 4
Business quality
3 of 4
Financial position
3 of 4
Execution record
3 of 4
Per share capture
2 of 4
Potential
2 of 4
Risk profile
2 of 4
Valuation
1 of 4
Evidence quality
3 of 4
How we weighed it
ValueNegative
The price sits far above our value and above the bull case.
BusinessMixed
A mixed business: real testing volume, but much growth was bought.
GrowthMixed
Growth and the loss improved, still far from the margin the price needs.
RisksNegative
A key data agreement ends soon, the founder's company supplies a large share and its own model may compete.
RivalsNegative
Dearer than the middle of its named rivals while keeping less of each dollar.
The groupMixed
It sits in none of our ranked groups; drug research budgets move it.
Latest newsMixed
The last report improved growth and the loss without moving the view.
In plain words
Our estimate of value
We forecast how much cash the business can earn over the next ten years and beyond, then ask what that stream is worth today to an investor who wants an 8% yearly return. We do it three times: a gloomy future, an expected one and a sunny one. We call a share Attractive only when its price is at least 20% below the expected estimate, and Unattractive when it is 20% above.