Tempus’s data growth needs a closer look at repeatable work

Tempus reported on November 2, 2026 that third-quarter revenue reached $411.5 million. Data and Applications contributed $108.1 million, growing 33% year over year. Quarterly release.

The more useful business question is how much work must be repeated when another research customer arrives. A dataset can be licensed more than once, but the economics depend on what accompanies the file: definitions, access rights, delivery obligations, support and an explanation of how the result was produced. This article offers an analytical framework, not a finding about undisclosed Tempus costs or a judgment about treatment effectiveness.

What can actually be reused

Tempus's October 23 announcement says BMS will license multimodal oncology data through 2031 and gain access to Lens. Those are commercial arrangements and intended research capabilities, not demonstrated treatment outcomes. Agreement announcement.

Consider an illustrative customer asking whether a particular research cohort can support a new hypothesis. The provider may already possess relevant records. Yet someone still has to translate the question into a workable delivery, resolve ambiguous definitions and make the output understandable to the customer's scientists. If that entire process starts again for every customer, a larger archive can still support a labor-intensive service business.

A different outcome is possible when reusable definitions, documented transformations and a stable delivery interface reduce the next project's setup work. The advantage then sits partly in the operating system around the records. The customer buys a shorter path from a research question to something its team can inspect and use. That is a more precise hypothesis about operating leverage than simply assuming that every additional data license has almost no cost.

The two models can coexist. A standard delivery may be inexpensive while an unusual request requires substantial specialist work. An analyst therefore needs evidence about the mix, rather than assigning a software margin to every dollar described as data revenue. Public segment totals alone cannot reveal the incremental effort behind an individual contract.

Contracts and evidence answer different questions

FDA's July 2024 guidance emphasizes relevance and reliability when electronic health records or claims support regulatory decisions. It is nonbinding guidance, not an endorsement of a particular vendor. FDA guidance.

For business analysis, this creates a distinction between a commercially useful research resource and a result strong enough for a particular clinical claim. A buyer may reasonably pay for exploratory work that narrows a list of hypotheses, even when that work cannot establish that a treatment improves outcomes. Equally, a scientifically promising resource need not generate attractive returns if producing each usable delivery remains costly.

One hypothetical project makes the distinction concrete. A team commissions an exploratory comparison, receives a reproducible analysis, then decides that further work is unwarranted. The supplier may have delivered valuable decision support even though no drug advances. Conversely, a polished output that the customer cannot reproduce can consume more time than it saves. These are possible mechanisms, not claims about an actual Tempus project.

Renewal may reflect easier tools, faster research preparation or familiarity with delivery. It cannot by itself establish treatment effectiveness. Assessing commercial demand and clinical value separately does not diminish either; it assigns the evidence to the question it can actually answer.

What would make the business argument stronger

The commercial case would become more persuasive if repeat projects required less setup, customers renewed for usable outputs, and growth could be supported without proportionate increases in specialist delivery work. Evidence that customers merely purchased larger initial packages would support a different interpretation. It could indicate strong demand without yet demonstrating repeatable operating leverage.

Standardization also involves a trade-off. Too much may exclude valuable unusual questions; too much customization may limit scale. The useful observation is which work becomes a common foundation and which needs project-specific investment, rather than whether customization disappears entirely.

Contract value, recognized revenue and cash collection also belong on different timelines. A long agreement may improve visibility, but its headline size cannot substitute for an understanding of deliveries and obligations. This is an accounting-reading principle, not a proposed reconstruction of the company's undisclosed contract schedules.

Tempus's latest figures provide a timely reason to examine these questions. The next analytical step is to connect commercial scale to repeatable work, and repeatable work to customer use. Clinical benefit needs its own evidence. Keeping those assessments separate gives investors a clearer way to judge progress without asking a quarterly revenue number to prove more than it can.