TEM — Tempus AI, Inc.

Tempus has assembled the largest clinically annotated AI dataset in oncology — 45 million patient records, 400 petabytes of molecular and clinical data, and ordering relationships with 50% of U.S. oncologists — creating a data moat that compounds with every additional biopsy sent and every additional pharma partnership signed. The business generated $1.265 billion in revenue in 2025, growing 80% year-over-year, and just crossed adjusted EBITDA breakeven for the first time. The catalyst needed to make this actionable is straightforward: FDA approval of the xT CDx as an Advanced Diagnostic Laboratory Test, which would trigger a pricing step-up from roughly $1,640 to above $2,100 per test; without it, the path to meaningful GAAP profitability is slower than the current valuation requires.

---

The premise underlying precision medicine investment is that the most important variable in cancer treatment is molecular information — that two patients with the same tumor type will respond differently to the same drug, and that knowing the genomic signature of a tumor before prescribing a treatment will produce materially better outcomes. This premise is no longer contested in oncology; it is standard of care. The American Society of Clinical Oncology, the National Comprehensive Cancer Network, and every major cancer center in the country have integrated genomic sequencing into frontline treatment protocols. The question for investors is not whether precision medicine is real — it is — but who captures the economic value of operating at the intersection of molecular data and clinical decision-making.

The answer to that question is contested in a way it was not five years ago. The genomics testing market has attracted serious capital and produced serious competitors. Foundation Medicine, owned by Roche, has been sequencing tumors at scale since 2012 and built a dataset that predates Tempus by several years. Guardant Health focuses on liquid biopsy — blood-based genomic testing — and is approaching $1 billion in annual revenue. Caris Life Sciences, still private, competes directly in solid tumor profiling. The AI healthcare market, broadly construed, drew $3.97 billion in revenue in 2026 and is projected to reach $125 billion by 2040 on a 28% compound annual growth rate. The growth is real; so is the competition.

What distinguishes Tempus from a genomics testing company is the architecture of what it built around the testing capability. The founding insight — that the data generated by clinical genomic sequencing is more valuable than the test itself — led to an intentional strategy of building a data platform alongside the diagnostic laboratory. Every tumor sequenced creates a molecularly annotated patient record. Connected to clinical outcomes, imaging, pathology slides, and electronic health record data from 5,000 partnered institutions, these records form a longitudinal dataset that pharmaceutical companies, academic medical centers, and biotech developers need and cannot acquire elsewhere. This is the business Tempus is actually building, and the diagnostics laboratory is both the engine and the marketing function for building it.

Tempus reports two segments: Diagnostics (which includes genomics testing and, since the Ambry Genetics acquisition, hereditary cancer screening) and Data and Applications (which includes Insights — data licensing to pharma and biotech — and AI-powered clinical decision tools). In fiscal 2024, Diagnostics generated $451.7 million and Data and Applications generated $241.6 million. In fiscal 2025, total revenue reached $1.265 billion, with Diagnostics roughly tripling through a combination of organic growth and the Ambry acquisition, which added hereditary volume on top of the existing oncology testing base.

The data moat is the correct starting point for evaluating competitive position. The 45 million patient records Tempus holds are not interchangeable with smaller datasets — the value of clinically annotated molecular data is superlinear in size, because rare variants, rare drug responses, and rare outcomes only become statistically analyzable at scale. A dataset of one million patients cannot tell you how BRCA2-variant triple-negative breast cancer responds to pembrolizumab in patients over seventy with co-occurring PTEN loss. A dataset of forty-five million might. This scale creates a structural barrier to entry that compounds: Foundation Medicine has had longer to build its dataset, but Tempus grew faster in the 2020s by building physician relationships rather than purely pharma relationships. The resulting penetration figures are striking: 50% of U.S. oncologists are now ordering Tempus tests or contributing data, up from 30% five years ago, and approximately 95% of the top 20 pharmaceutical companies use Tempus data or trial matching services.

Metric	Tempus AI	Guardant Health	Foundation Medicine (Roche)
Patient Records / Dataset	45M+ records, 400 PB	~2M+ liquid biopsy tests	Millions (proprietary, undisclosed)
Oncologist Reach	50% of U.S. oncologists	Solid oncology focus	Academic center focus
Pharma Customer Penetration	~95% of top 20 pharma	Major pharma relationships	Roche-integrated
Net Revenue Retention	126%	Not disclosed	Private / part of Roche
FY2025 Revenue (est.)	$1.265B	~$750M	~$1B+ (Roche segment)

The 126% net revenue retention is the most important single number in the table. It means that the average Tempus customer — a pharmaceutical company or research institution — spends 26% more in year two than in year one, without the company acquiring a single new customer. This metric is the functional proof that the data is generating real value: customers expand their data access, order additional analyses, and bring new programs to the platform as they validate the insight quality. A dataset that produced marginal returns would produce net revenue retention below 100%, as customers let agreements lapse rather than expand them. Tempus's data is not marginal to its customers — it is becoming infrastructure.

The financial picture is improving rapidly, though it requires careful reading. Full-year 2025 revenue of $1.265 billion represents 80% growth over 2024's $693 million, but the headline growth is inflated by the Ambry Genetics acquisition. Organic growth excluding Ambry was approximately 33.5% in Q4 2025 — still exceptional, but meaningfully different from the reported figure. Non-GAAP gross margins reached 61.7% for Diagnostics and 69.7% for Data and Applications in Q3 2025, improving approximately 400 basis points year-over-year at the company level. Adjusted EBITDA for full-year 2025 was $5 million — the company's first year of adjusted profitability after years of operating losses.

The gap between adjusted and GAAP results requires specific attention. GAAP operating margin was negative 19.88% in Q4 2025 despite a 62.74% GAAP gross margin, reflecting significant stock-based compensation, amortization of acquired intangibles (primarily from Ambry), and ongoing investment in R&D and go-to-market infrastructure. The trailing twelve-month GAAP EPS was negative $1.19 as of late 2025. The reconciling items — primarily SBC and acquisition amortization — are real economic costs, not accounting artifacts. An investor who values the business on adjusted EBITDA is accepting a multiple that excludes these costs; an investor who values it on GAAP is valuing a business that, despite $1.265 billion in revenue, has not yet demonstrated positive earnings power in full accounting terms.

Eric Lefkofsky is the founder and CEO of Tempus. He is a serial entrepreneur whose prior ventures include Groupon, where he took a marketplace from concept to IPO, and Uptake, an industrial AI company. His technical and commercial credibility is high, and his 24% beneficial ownership aligns his incentives with long-term shareholder value in the conventional sense. The complicating fact is the insider selling: between December 2024 and February 2025, Lefkofsky sold approximately 4.6 million shares — roughly 11% of his holdings — for proceeds exceeding $250 million, executed under a Rule 10b5-1 trading plan. For a company whose CEO is publicly characterizing the business as early-stage and still investing heavily in growth, the scale of this liquidity realization warrants noting. The counterargument is that a founder who still holds tens of millions of shares and 24% beneficial ownership retains substantial skin in the game. Both are true simultaneously.

The growth trajectory shows a business scaling toward profitability while the two core engines develop at different rates.

Period	Diagnostics Revenue	Data & Applications Revenue	Total Revenue	Adj. EBITDA
FY2023 (est.)	~$280M	~$170M	~$450M	~($130M)
FY2024	$451.7M	$241.6M	$693.3M	~($105M)
FY2025	~$900M	~$365M	$1,265M	$5M
FY2026E	~$1,200M	~$390M	$1,590M	~$65M

Two observations from this table: First, the Diagnostics business is the growth driver in absolute terms, but its growth rate is partly acquisition-driven and partly dependent on reimbursement dynamics that Tempus does not fully control. The FY2025 Diagnostics number of approximately $900 million includes Ambry's hereditary testing volume, which grew 23% year-over-year organically in Q4 2025. Second, the Data and Applications segment — the one with 69.7% gross margins and 126% net revenue retention — is growing at 25% to 37% annually, which is slower than Diagnostics but on a revenue mix that produces substantially more value per dollar. The thesis requires this segment to become the dominant revenue source over time. The 2026 guidance implies D&A growing to approximately $390 million — still only 25% of total revenue, not yet the majority.

The penetration argument runs across three dimensions. In oncology testing specifically, Tempus has reached 50% of U.S. oncologists — a remarkable saturation figure for a company that has been selling for roughly eight years. The remaining 50% of oncologists, plus expansion into non-oncology specialties (cardiology, psychiatry, neurology — where Tempus has begun deploying its TIME platform for cardiac risk and mental health applications), represent the near-term growth surface. MRD (minimal residual disease) testing — blood-based tests that detect cancer recurrence with greater sensitivity than standard imaging — is the most important new product in the pipeline: volume grew 56% quarter-over-quarter in Q4 2025 to approximately 4,700 tests, from a small base but on a trajectory that suggests a new revenue layer building alongside the established tumor profiling business. International expansion is not yet a material contributor; it represents years of untapped opportunity given that the precision medicine infrastructure build-out in Europe, Asia, and Latin America is earlier than in the U.S.

The valuation requires a framework that can accommodate a company with $1.265 billion in revenue, 80% growth, and $5 million in adjusted EBITDA. Price-to-sales multiples are the appropriate lens: at approximately $50 per share, Tempus trades at roughly 8 times trailing revenue and approximately 6.5 times 2026 guided revenue. These are growth multiples for a growth company — not a value entry point, but not an obviously absurd one given the growth rate and the quality of the data asset. The traditional earnings-multiple framework cannot yet produce a meaningful number: the company is at adjusted EBITDA breakeven, guided to $65 million in adjusted EBITDA for 2026, and multiple years from generating the kind of GAAP earnings that would support a standard price-to-earnings analysis.

The specific catalyst that would make this stock actionable for a patient investor is the FDA designation of the xT CDx as an Advanced Diagnostic Laboratory Test, which would allow Medicare reimbursement at rates reflecting the clinical value of the test rather than the legacy fee schedule. Current average selling price is approximately $1,640 per oncology test. The ADLT pathway, which Guardant Health successfully navigated for its Guardant360 CDx product, allows reimbursement at $3,000 to $5,000 per test. Even a partial uplift toward $2,100 to $2,500 per test would add hundreds of millions of dollars in revenue and dramatically accelerate the path to GAAP profitability. The timeline for this approval is uncertain — and that uncertainty is precisely what makes the stock a catalyst story rather than a compelling-at-current-price story.

The most intelligent bear argument on Tempus is that the diagnostics laboratory business — which generates two-thirds of revenue — faces competitive and reimbursement pressures that the premium valuation does not adequately price. Foundation Medicine has been sequencing tumors longer and has Roche's global commercialization apparatus. Guardant Health is winning the liquid biopsy market, which has clinical advantages for monitoring and recurrence detection over tissue-based sequencing. Caris Life Sciences matches Tempus test-for-test in many oncology protocols and is building its own data licensing business. If the pricing uplift from ADLT approval is delayed or partial, and if the data licensing business does not accelerate its growth rate meaningfully above 25%, the current revenue multiple requires a profitability ramp that the business cannot deliver on the guided timeline. The counter to this bear is that 126% NRR, 95% pharma penetration, and 50% oncologist reach are numbers that only a truly differentiated platform produces — and that the ASP gap between current reimbursement and the clinical value of comprehensive tumor profiling represents a known catalyst with historical precedent.

For this analysis to shift from requiring a catalyst to compelling, one of two things must happen: the xT CDx receives ADLT designation and the resulting pricing uplift materializes in reported results, or the Data and Applications segment accelerates above 40% annual growth and establishes itself as the primary revenue engine, making the Diagnostics reimbursement risk secondary to the platform's economics. If the ADLT approval arrives in 2026 and the ASP moves toward $2,200 to $2,500 per test, the FY2026 adjusted EBITDA will substantially exceed the $65 million guidance, the path to GAAP profitability becomes visible, and the current valuation will look conservative in retrospect. If the approval is delayed another 12 to 18 months, the stock will trade on the current $65 million EBITDA guidance against a multiple that assumes substantially better results.

The data is real. The moat is real. The path to profitability is visible but not yet walked. At 6.5 times forward revenue, the stock is pricing in an optimistic-but-not-heroic execution scenario. The ADLT decision is the variable that separates optimistic from heroic.