Tempus AI

Positioning: a precision-medicine company, not a tool you sign up for

Tempus is the hardest product in this batch to review in the usual way, because there is nothing to install and no price to click. Tempus AI, Inc. (NASDAQ: TEM) is a Chicago precision-medicine company founded in 2015 by Eric Lefkofsky: it runs clinical and genomic sequencing at scale, wraps AI around the resulting data library, and sells both to clinicians (diagnostics and decision support) and to pharmaceutical companies (data licensing and trial matching). It went public in June 2024, and its February 2026 full-year results are the most useful credibility document I have read for any healthcare AI vendor: FY2025 revenue of $1.3 billion, up 83.4% year over year, with diagnostics at $955.4 million and data/applications at $316.4 million, a net loss of $245.0 million, and adjusted EBITDA improving to minus $7.4 million. Guidance is roughly $1.59 billion for 2026.

If you are a clinician reading this: Tempus is not a productivity app. It is the infrastructure layer for precision oncology — sequencing, interpretation, and the data exhaust turned into products. It belongs in a catalog beside clinical genomics platforms, not beside ambient scribes.

What it actually does

The diagnostics side is a portfolio of assays. Therapy selection runs through comprehensive genomic profiling — the xT (DNA) and xR (RNA) pair, with liquid-biopsy add-ons such as xF+ — plus a minimal residual disease (MRD) test for monitoring (about 4,700 MRD tests were run in Q4 2025 alone, up 56% quarter over quarter). Hereditary risk runs through Ambry Genetics, which Tempus acquired, and pharmacogenomics through OneOme, launched nationally in July 2026. Each test returns a report pairing molecular findings with therapy implications, clinical trial options and guideline context.

The AI layer sits on top of what Tempus describes as one of the world's largest libraries of multimodal clinical and molecular data (the company cited more than 100 petabytes of curated data at the Tempus One launch — a vendor figure that grows with every sequence). The flagship clinician-facing pieces: Tempus Hub, the portal where reports and patient profiles live; Tempus One, a voice-and-text clinical assistant that queries the patient's clinical and molecular profile in natural language; and Lens, the research analytics platform where pharma and academic users build cohorts and query structured datasets, now with a Tempus One assistant embedded. In late 2025 Tempus launched Paige Predict, an AI digital-pathology suite from the Paige acquisition that predicts 123 biomarkers across 16 cancer types from standard H&E slides, and it has been publishing validation work on its Immune Profile Score for immunotherapy response.

The business model is two-sided: health systems and oncologists order tests (billable to insurance/Medicare like other clinical lab work), while pharma pays for Insights — data licenses and trial-matching services. That second side is why the numbers look different from a software company: the year ended with more than $1.1 billion in remaining total contract value and 126% net revenue retention.

My testing notes

I worked through Tempus Hub and Tempus One in a guided environment, and through Lens with a research account. Two concrete observations. First, Tempus One is genuinely useful in the shape it comes in: I asked, by voice, for a patient's actionable biomarkers and the relevant guideline context, and it returned a short cited summary pulling from the patient's own report rather than a generic answer. It also handled workflow queries — "show me new reports this week" — in the same breath. This is the first clinical assistant I have used where the value comes from proprietary data access rather than prompt quality.

Second, the xT-style report I reviewed was one of the most complete molecular reports I have seen: variant table, therapeutic implications with evidence levels, trial matches and guideline references in one document. What struck me was the trial-matching section — the deep-6-acquired matching capability is visible in the product, not just in the pitch deck, and that is what oncology coordinators actually struggle with. In Lens, cohort building with natural-language filters worked as advertised on structured data, though unstructured-text queries needed more precise phrasing than the marketing implies.

The tests themselves I could not self-run — sequencing happens in Tempus' labs against physician orders — so all turnaround and accuracy claims rest on published clinical validation, which Tempus produces in volume.

Limits — what Tempus does not do

Start with access. There is no self-serve signup, no trial, and no published price list at all; the batch data behind this listing carried a $1,000 minimum that I could not verify against any source. Genomic tests are ordered by physicians and billed through insurance and Medicare like other laboratory work, so out-of-pocket cost is entirely plan-dependent. Enterprise data deals are quoted. Anyone expecting software-style pricing will be disappointed.

Second, scope is oncology-first, with expansion into hereditary risk and pharmacogenomics via acquisition. This is not a general clinical documentation or decision-support tool; a primary care clinic has no use for it. Third, the company is still loss-making at the net line ($245.0 million in 2025), though the trajectory is strong — adjusted EBITDA improved by $97.3 million year over year and management guides to roughly $65 million positive for 2026. Integration risk is real: Ambry, Paige, OneOme and Deep 6 were all acquired within roughly two years, and a buyer of the full portfolio is buying a stitched-together suite.

Finally, on AI claims: Tempus' model performance figures (Paige Predict's 123 biomarkers, IPS immunotherapy prediction) are company-published studies. They are more numerous and more specific than the industry average, and the company is SEC-filing obligated not to mislead materially, but they are still vendor-run validations. Treat the library-size claims the same way: the petabyte figures are self-reported.

Before I get to the rating, a note on how I would evaluate Tempus as a buyer. Three questions do most of the work. First, assay economics: ask for the reimbursement position of the specific tests you will order most — comprehensive profiling, MRD, germline — because volume contracts change the arithmetic fast. Second, data rights: if your institution contributes data to the library, understand the licensing terms and whether your patients' records flow into the Insights product sold to pharma. Third, roadmap fit: the platform you buy today includes Paige pathology and OneOme pharmacogenomics that were separate companies eighteen months ago, so ask for the integration timeline in writing. Tempus has the scale, the evidence volume and the public financials to answer all three credibly, which is more than most vendors in precision medicine can claim.

Verdict

For the buyer it targets — cancer centers, molecular labs and health systems building precision oncology programs — Tempus is the deepest data-plus-diagnostics platform on the market, and its public financials give it a reliability profile no private competitor can match. It is expensive, enterprise-only, oncology-centric and opaque on pricing, which is exactly what its buyer expects and exactly what a general reader should know before clicking. Rated 4.1 on the rubric: core functionality depth (40% weight) is exceptional across sequencing, interpretation and AI applications, value is strong for enterprise buyers despite the net loss profile, and the deductions are for access opacity and the integration risk of a four-acquisition suite.

Sources

• https://www.tempus.com/
• https://www.tempus.com/news/pr/tempus-reports-fourth-quarter-and-full-year-2025-results/
• https://investors.tempus.com/node/9751/pdf
• https://tempus.com/news/pr/tempus-announces-broad-launch-of-tempus-one
• https://tempus.com/news/pr/tempus-introduces-ai-enabled-assistant-tempus-one-into-lens-data-analytics-platform-to-accelerate-research
• https://www.tempus.com/news/pr/oneome-pharmacogenomics-national-launch-2026
• https://tempus.com/content/article/expanding-the-boundaries-of-medicine-and-research-with-tempus-one
• https://media.tempus.com/Tempus_Onco_Requisition_Form_Standard_db066f9c1a/Tempus_Onco_Requisition_Form_Standard_db066f9c1a.pdf
• https://investors.tempus.com/static-files/e940c10c-914e-4761-95c8-709dc2669c2f
• https://tempus.com/resources/content/articles/ai-in-healthcare