Tempus AI earnings have reached the point where the company can no longer be described only as a fast-growing diagnostics platform with an artificial-intelligence narrative attached. Third-quarter revenue rose 23% to $411.5 million, gross profit increased 27% to $266.7 million, and adjusted EBITDA expanded to $25.6 million from $1.5 million a year earlier. Those figures show operating leverage beginning to appear. They also reveal why the most important question is not whether Tempus can grow. It is whether the company can convert a growing clinical network into durable, cash-generating economics without asking investors to ignore stock-based compensation, acquisition risk and the volatility of non-operating gains.
The quarter supports a stronger interpretation of the business model than the headline growth rate alone. Tempus generated $303.5 million of diagnostics revenue and $108.1 million from Data and Applications. The first segment creates clinical access, testing volume and new patient information. The second sells higher-margin access to the information, analytical software and research services that the clinical system helps produce. If those activities reinforce one another, every incremental diagnostic relationship can improve the value of the data platform, while every pharmaceutical collaboration can make the diagnostic network more useful. That is the flywheel investors are underwriting.
Yet a flywheel is not the same as a moat, and adjusted profitability is not the same as economic profit. Tempus reported a $14.9 million GAAP net loss. That loss included $50.8 million of stock-based compensation and related payroll taxes, while it was reduced by $42.8 million of unrealized gains on marketable securities. The first item transfers value from shareholders to employees even when it does not consume cash in the current period. The second can reverse without saying anything about the performance of diagnostics or data licensing. A careful reading therefore has to hold two conclusions at once: the core business improved materially, and the path from adjusted earnings to repeatable owner earnings remains unfinished.
The Quarter Was Stronger Than the Revenue Headline
Tempus reported total revenue of $411.5 million for the three months ended September 30, up from $334.2 million a year earlier. The 23% increase was broad enough to matter because both sides of the model contributed. Diagnostics revenue grew 20% to $303.5 million, while Data and Applications revenue rose 33% to $108.1 million. The faster growth of the data segment is strategically important. Diagnostics still supplies almost three quarters of revenue, but Data and Applications carries structurally higher gross margins and can convert the clinical network into recurring enterprise relationships.
The composition of diagnostics also improved. Oncology testing volume increased 32%, and minimal residual disease volume reached approximately 12,000 tests, four times the prior-year level. Minimal residual disease testing is designed to detect very small quantities of cancer-associated material that may remain after treatment or reappear before ordinary imaging and symptoms provide a clear signal. The commercial value is not simply another laboratory test. Repeated testing can create a longitudinal relationship with physicians and patients, generate clinically useful data across time and connect the laboratory to treatment monitoring.
Growth in hereditary testing was only 3%, which prevents the diagnostics result from being read as uniform acceleration. The portfolio is being pulled by oncology and emerging molecular-monitoring products rather than every category moving at the same speed. That concentration is not automatically negative: oncology has large unmet needs, high information intensity and strong pharmaceutical-research demand. It does mean investors should distinguish between a platform that is broadly accelerating and one whose strongest franchises are carrying slower areas.
Gross profit reached $266.7 million, up 27%, and reported gross margin increased to 64.8% from 62.8%. The segment detail explains the improvement. Diagnostics gross margin rose to 62.4% from 61.0%, while Data and Applications margin increased to 71.5% from 68.5%. The data business contributed only about 26% of revenue but almost 29% of gross profit. That gap is exactly what a favorable mix shift should look like: the faster-growing segment contributes a larger share of gross profit than revenue.
This resembles the analytical distinction in Block2Learn’s review of Thermo Fisher’s life-sciences recovery. Reported growth matters, but the composition of growth and the margin attached to each revenue stream determine how much value is actually being created. Tempus has the added complexity of operating a laboratory network partly to generate the information and relationships that make its data products more valuable. The diagnostic cost base is therefore not only a cost center. It is part of the acquisition engine for the higher-margin business.
Diagnostics Is the Data-Acquisition Layer
The phrase “data flywheel” is often used so loosely that it loses analytical value. A company does not obtain a durable advantage merely because it stores many records or mentions machine learning. The relevant mechanism has several steps. Tempus must persuade physicians and health systems to order tests, produce accurate and useful results, connect molecular findings with longitudinal clinical information, structure the data so that it can be analyzed, preserve patient privacy and regulatory compliance, and then deliver products that help researchers or drug developers make better decisions.
Each step can fail independently. If the test is not clinically useful, physicians will not order it. If data are incomplete, biased or poorly standardized, model performance and research value can deteriorate. If access rules are unclear, pharmaceutical customers may hesitate. If the software does not improve research productivity, the database becomes an expensive archive rather than a product. The flywheel exists only when the clinical and commercial loops reinforce each other in practice.
The quarter provides evidence that they are beginning to do so. Oncology volume growth creates more opportunities to add molecular observations. Repeated minimal residual disease testing adds a time dimension. Data and Applications growth suggests that external customers are willing to pay for access, software and analytical services. The company’s $550 million-plus of year-to-date data-licensing bookings adds visibility to future demand, although bookings must not be confused with recognized revenue.
That distinction is essential. A booking records a contractual commitment or expected value under agreed terms. Revenue is recognized when the company satisfies performance obligations under accounting rules. Cash may arrive on yet another schedule. A five-year agreement can support backlog and customer confidence without producing five years of revenue immediately. The correct chain is therefore bookings, delivery, revenue recognition, cash collection and renewal. Every link matters.
Tempus disclosed a particularly useful example in its October agreements with Bristol Myers Squibb and Pathos. An SEC filing describing the arrangements states that the oncology and neurology agreements provide for aggregate cash payments of $192.5 million over approximately five years. The oncology statement of work includes $137.5 million of fixed payments for multimodal data, Lens software access and related services. The neurology order form includes up to $55 million, including payments tied to delivering as many as 500,000 de-identified records and maintaining updates.
This is more informative than a headline bookings number. It shows what customers are buying, how long delivery may take and where execution risk remains. The oncology contract has defined fixed payments, but Tempus must continue to provide access and services. The neurology contract depends partly on delivering records at an agreed pace. A large total contract value is attractive; the conversion schedule determines near-term economics.
Higher-Margin Data Revenue Is Not Pure Software
The Data and Applications segment produced a 71.5% reported gross margin. That is well above the 62.4% margin in diagnostics, but it should not be treated as effortless software revenue. Clinical data require curation, governance, infrastructure, medical expertise and continuous updating. Enterprise customers may also require custom analysis and research services. The business can scale, but it does not scale exactly like a standardized software subscription.
This matters for expectations. A conventional software company may add customers to an already-built product with relatively low incremental delivery cost. Tempus may need to keep investing in laboratory operations, clinical integrations, dataset expansion, regulatory processes and research staff to maintain the quality of what it sells. Those investments can deepen the moat while delaying the margin profile that investors associate with mature software.
The favorable interpretation is that Tempus has two complementary forms of operating leverage. Laboratory scale can spread fixed infrastructure across more tests, which helps diagnostics margin. Data scale can allow a richer dataset and software platform to support more enterprise relationships, which helps Data and Applications margin. The skeptical interpretation is that the company is combining two capital-intensive activities and presenting the higher-margin output as evidence that the whole system will eventually resemble software. The latest quarter moves the evidence toward the favorable case, but one quarter cannot settle the issue.
Investors should watch whether Data and Applications continues to grow faster than diagnostics while preserving margin above 70%. They should also watch whether diagnostics margin improves as oncology and minimal residual disease volumes expand. If both happen together, the business is demonstrating genuine dual-engine leverage. If data growth requires rising custom-service expense or if diagnostics expansion compresses laboratory margin, the flywheel may be less efficient than the narrative suggests.
Adjusted EBITDA Improved, but the GAAP Bridge Matters
Adjusted EBITDA rose to $25.6 million from $1.5 million. Non-GAAP operating income reached $14.4 million compared with an $8.7 million loss a year earlier. These changes are too large to dismiss as cosmetic. Revenue increased faster than many operating costs, gross margin expanded and the company converted a larger share of growth into adjusted profit. Management now expects approximately $65 million of adjusted EBITDA for the full year, while raising revenue guidance to $1.60–$1.605 billion.
The full-year guidance implies roughly 25% annual revenue growth. It also provides a useful internal consistency check. Tempus generated $1.142 billion of revenue over the first nine months, leaving approximately $458–$463 million required in the fourth quarter to reach the range. That would represent another sequential step up from $411.5 million, so the guidance assumes continued momentum rather than a flat exit rate. The company has stronger visibility from testing volume and enterprise commitments, but the target still requires execution.
The GAAP numbers prevent the adjusted result from becoming the whole story. Tempus reported an operating loss of $56.9 million and a net loss of $14.9 million. The difference between adjusted and reported performance includes stock-based compensation, employer payroll taxes related to equity awards, acquisition-related items and other exclusions. Stock-based compensation and related payroll taxes alone totaled $50.8 million in the quarter, equivalent to more than 12% of revenue and almost twice adjusted EBITDA.
Stock-based compensation is not identical to cash wages. It can align employees with shareholders and preserve cash for investment. But it is not free. Existing investors own a smaller percentage of the company when new shares or options are issued, unless repurchases offset the dilution. A business that reports positive adjusted EBITDA while regularly issuing a large amount of equity compensation has not yet proved that its growth can support employees, investment and shareholders from internally generated economics.
The $42.8 million unrealized gain on marketable securities is the mirror image. It improved net income without reflecting better laboratory utilization, stronger data contracts or more efficient operations. Unrealized gains can become realized cash later, but they are exposed to market prices and may reverse. The quarter’s $14.9 million net loss therefore understates the operating gap if the securities gain is treated as non-recurring, while adjusted EBITDA overstates owner economics if equity compensation is treated as costless. The most useful answer sits between the two.
This same discipline appears in Block2Learn’s analysis of Novartis and its acquired-pipeline impairment. Adjusted measures can illuminate current operating momentum, while GAAP results capture real consequences of capital allocation and compensation. Neither should be used alone. Tempus’s improvement is genuine because the operating bridge moved sharply in the right direction. The remaining gap is genuine because employees, acquisitions and infrastructure still have to be financed.
Partnerships Extend the Platform Into Drug Delivery
The most strategically important announcement may not be a data-license contract. On October 7, Tempus announced a multi-year collaboration with Merck and Moderna related to the potential commercialization of intismeran autogene, an individualized neoantigen therapy being evaluated with Keytruda in melanoma and other cancers. Tempus would manage tissue and blood-sample collection and transfer, and, subject to regulatory approvals, provide next-generation sequencing used in treatment design and manufacturing.
That workflow shows how the diagnostics network could become infrastructure rather than merely a source of data. An individualized therapy requires a chain that begins with a specific patient’s tumor, moves through sequencing and computational design, continues into manufacturing and returns a therapy to the patient. Reliability, turnaround time and integration matter as much as the sophistication of the model. Tempus’s clinical footprint could place it inside that chain.
The opportunity is significant because infrastructure can be more durable than a one-time data sale. If pharmaceutical partners build development and commercialization processes around Tempus, switching costs may increase. The risks are equally specific. The therapy must succeed clinically and obtain regulatory approval. The sample-to-manufacturing chain must work at commercial scale. Reimbursement and physician adoption must support demand. Tempus may invest before the revenue opportunity becomes visible.
The comparison with AbbVie’s portfolio transition is useful. Drug developers create value through therapeutic intellectual property and market access. Tempus is trying to create value around the information and workflow layer that helps many therapies reach the right patients. That diversification can reduce dependence on any single drug, but it also means Tempus’s economics depend on partners, regulatory timelines and the quality of the broader biopharma pipeline.
The company’s two FDA 510(k) clearances for ECG-PH and ECG-MR add another form of optionality by extending its AI-enabled cardiology footprint. The FDA record for ECG-PH lists an August 21 decision and identifies the product as pulmonary-hypertension machine-learning notification software. These products suggest that the platform may move beyond oncology into non-invasive cardiovascular analysis. Expansion across disease areas can enrich the dataset and widen the customer base, but it also creates execution complexity. A platform that pursues oncology, cardiology, hereditary testing, neurology and whole-genome research simultaneously must allocate capital and medical expertise carefully.
Personalis Adds Scale and Integration Risk
Tempus also said regulatory review of its pending acquisition of Personalis had cleared, with a shareholder vote planned for November 30. Personalis brings capabilities in advanced cancer genomics and minimal residual disease, directly overlapping with one of Tempus’s fastest-growing areas. The strategic logic is easy to see: more technology, more clinical relationships and more oncology data can strengthen the platform.
The financial logic requires more work. Acquisitions can accelerate product breadth, but they can also import costs, integration risk, duplicate infrastructure and future impairment. The value depends on whether Tempus can combine laboratories, commercial teams, data systems and product road maps without disrupting customers or weakening turnaround times. Cost savings matter, but revenue retention and scientific quality matter more.
This is why acquisition accounting should remain part of the Tempus AI earnings debate. A platform can improve adjusted EBITDA while excluding acquisition-related expenses, then later record impairments if expected revenue does not arrive. Investors should track the purchase consideration, financing method, incremental dilution, integration costs, retained talent and the combined minimal residual disease roadmap. The acquisition should be judged on cash returns and strategic acceleration, not merely a larger dataset.
Tempus ended the quarter with $854.4 million of cash and marketable securities, giving it capacity to fund integration and continued expansion. Liquidity reduces near-term financing risk. It does not eliminate the cost of capital. Cash spent on an acquisition cannot also be used to extend runway, repurchase diluted shares or fund organic product development. The correct question is whether Personalis can create more value inside Tempus than the same capital could create elsewhere.
Three Scenarios for the Next Twelve Months
Scenario One: The Flywheel Becomes Measurable
In the constructive scenario, oncology volume remains above 25%, minimal residual disease continues to scale, and Data and Applications grows faster than diagnostics. Segment gross margins remain near or above current levels. The BMS and Pathos agreements begin converting bookings into recognized revenue and cash on schedule, while the Merck-Moderna workflow advances toward commercial readiness. Full-year guidance is achieved, and adjusted EBITDA expands again in 2027.
The critical confirmation would be improving operating cash flow and a narrowing GAAP operating loss without a larger percentage of revenue going to stock-based compensation. If Tempus can produce more revenue, more gross profit and more cash with slower growth in diluted shares, the platform thesis becomes much stronger. In this scenario, the data asset is not only valuable to customers; it is valuable to owners.
Scenario Two: Growth Continues but Conversion Stalls
In the middle scenario, revenue remains above 20% growth and enterprise bookings stay healthy, but integration, hiring and product investment absorb most of the gross-profit improvement. Adjusted EBITDA remains positive while GAAP operating losses persist. Stock-based compensation stays near the current share of revenue, and free cash flow does not improve meaningfully.
This outcome would not invalidate the business. It would show that Tempus is still in an investment phase. The valuation question would become more demanding because investors would have to estimate when growth spending normalizes and how much dilution occurs before then. Strong reported growth could coexist with weak per-share value creation.
Scenario Three: Clinical and Contract Execution Diverge
In the adverse scenario, diagnostic volume remains healthy but enterprise contracts convert more slowly than expected. Custom delivery costs reduce Data and Applications margin, the Personalis integration creates disruption, or individualized-therapy workflows take longer to commercialize. Marketable securities gains reverse, revealing a larger net loss, while cash consumption increases.
The warning would not necessarily appear first in revenue. Bookings can support optimism while delivery schedules slip. The more useful signals would be deferred revenue, remaining performance obligations, customer concentration, segment margin, operating cash flow and management’s explanation of implementation timelines. A platform can sign valuable contracts and still disappoint if integration and delivery are harder than expected.
What Investors Should Monitor
The first indicator is revenue mix. Data and Applications should continue gaining share without relying on one customer or one unusually large project. A higher data mix supports margin, but concentration can make quarterly results volatile.
The second is the conversion of bookings. Tempus should explain how the $550 million-plus of year-to-date data-license bookings become revenue and cash across time. The BMS and Pathos filing provides a good model because it separates fixed payments from delivery-dependent amounts.
The third is diagnostics quality and volume. Oncology growth, minimal residual disease tests, reimbursement, average selling price and turnaround time should be read together. More tests are valuable only when clinical utility, collection and unit economics hold.
The fourth is the bridge from adjusted EBITDA to cash. Operating cash flow, capital expenditure, acquisition payments and working capital reveal whether reported operating leverage is financing the business. Adjusted EBITDA is a useful milestone, not the final destination.
The fifth is dilution. Investors should compare stock-based compensation, share count and repurchases with revenue and gross profit. A company can grow rapidly while per-share economics improve slowly.
The sixth is acquisition execution. Personalis retention, product integration, laboratory utilization and incremental margin should be tracked against the strategic promises made when the deal was announced.
The seventh is regulatory and clinical progress. FDA clearances, reimbursement coverage and the development path of partner therapies determine whether optional products become commercial businesses. Regulatory milestones should be connected to actual orders and adoption rather than treated as revenue by implication.
The eighth is data governance. De-identified clinical information can be economically valuable only if privacy, consent, cybersecurity and usage rights remain credible. A failure here would damage both diagnostics relationships and enterprise demand at the same time.
The Block2Learn Assessment
Tempus delivered a quarter that materially improves the credibility of its operating model. Revenue growth was broad, gross margin expanded, both segments increased profit, and adjusted EBITDA moved from almost break-even to a meaningful positive contribution. The faster growth and higher margin of Data and Applications support the thesis that clinical scale can be converted into a valuable enterprise data business.
The most compelling part of the story is the relationship between the two segments. Diagnostics is not merely a lower-margin business that investors must tolerate before reaching the software layer. It creates physician access, patient workflows and longitudinal information that make the higher-margin products possible. Data contracts, in turn, can finance deeper clinical integration and make the diagnostic network more useful. The architecture is stronger because each side has an independent customer and revenue base.
The financial presentation still requires discipline. The $42.8 million unrealized securities gain made the GAAP net loss look better without improving operations. The $50.8 million of stock-based compensation and related payroll taxes made adjusted performance look better without eliminating an economic cost to owners. These items do not cancel the progress. They define the remaining test.
The best evidence over the next year will not be a larger artificial-intelligence narrative or another large bookings headline. It will be a repeatable conversion sequence: diagnostic volume creates useful data, data products create contracted demand, contracts produce recognized revenue, revenue expands margin, and margin becomes cash without excessive dilution. Tempus has now demonstrated the first four links more clearly. It has not yet completed the last two.
The broader lesson extends beyond healthcare. Block2Learn’s analysis of Meta’s AI cash-conversion debate showed why a technology investment should be judged through its conversion into economic output rather than the scale of spending or sophistication of the model. Tempus presents the same principle in a different form. Clinical data can become a formidable asset, but only when it improves decisions, supports products, earns customer payments and creates value per share.
Continue Through the Block2Learn Learning Path
Understanding Tempus requires more than comparing one earnings release with consensus expectations. The business sits at the intersection of healthcare economics, data governance, laboratory operations, pharmaceutical development, acquisition accounting and equity compensation. Each layer changes the interpretation of the next. A growing database is not automatically a moat. Positive adjusted EBITDA is not automatically free cash flow. A regulatory clearance is not automatically commercial adoption. A large booking is not automatically current revenue.
The Block2Learn Learning Path develops the financial structure needed to separate those concepts. Free Start introduces the language of markets and company analysis. Foundation builds risk, accounting and capital-allocation principles. The Investor Operating System turns those principles into a repeatable framework for evaluating uncertain evidence. More advanced layers connect technology, market structure and portfolio thinking.
Tempus AI earnings now provide real evidence that the platform can scale. The next phase will show whether that scale belongs mainly to the company’s narrative, to its income statement, or ultimately to its shareholders.
Information is abundant. Structure is rare.
Sources
- Tempus AI third-quarter 2026 results, November 2, 2026.
- Tempus agreements with Bristol Myers Squibb and Pathos, Form 8-K, October 23, 2026.
- Tempus collaboration with Moderna and Merck, October 7, 2026.
- FDA 510(k) record for Tempus ECG-PH, decision dated August 21, 2026.
This article is provided solely for informational and educational purposes and does not constitute financial or investment advice, a recommendation, or an offer or solicitation to buy or sell any financial instrument or digital asset. See our Financial Disclaimer.
This article was generated with the support of AI and reviewed by the Editorial Team. For more information, see our Terms of Service.
