Prospectus Summary
PROSPECTUS SUMMARY This summary highlights selected information contained elsewhere in this prospectus. This summary does not contain all of the information you should consider before investing in our Class A common stock. You should read this entire prospectus carefully, including the sections titled “Risk Factors,” “Management’s Discussion and Analysis of Financial Condition and Results of Operations” and our consolidated financial statements and the related notes included elsewhere in this prospectus, before making an investment decision. Unless the context otherwise requires, all references in this prospectus to “Tempus,” the “company,” “we,” “our,” “us” or similar terms refer to Tempus AI, Inc. and its subsidiaries.Overview
We endeavor to unlock the true power of precision medicine by creating Intelligent Diagnostics through the practical application of artificial intelligence, or AI, in healthcare. Intelligent Diagnostics use AI, including generative AI, to make laboratory tests more accurate, tailored, and personal. We make tests intelligent by connecting laboratory results to a patient’s own clinical data, thereby personalizing the results. Our novel insight was realizing that all laboratory test results, genomic or otherwise, could be contextualized for a specific patient based upon that patient’s unique characteristics, and technology could therefore guide therapy selection and treatment decisions to allow each patient to progress on their own unique path. The drugs recommended, the clinical trials explored, the care pathways evaluated, the adverse events considered—all have the potential to be refined and enhanced when test results are connected to a patient’s personal profile, enabling the right patient to be routed to the right therapy at the right time.
To accomplish this, we built the Tempus Platform, which comprises both a technology platform to free healthcare data from silos and an operating system to make the resulting data useful. Our proprietary technology has allowed us to amass what we consider to be one of the largest libraries of clinical and molecular oncology data in the world. Our goal is to embed AI, including generative AI, throughout every aspect of diagnostics to enable physicians and researchers to make personalized, data-driven decisions that improve patient care.
The ability to deploy AI in precision medicine at scale has only recently become possible. Advances in cloud computing, imaging technologies, large language models, and low-cost molecular profiling, along with the digitization of vast amounts of healthcare data, have created a landscape that we believe is finally ripe for AI. However, despite an increase in the availability of healthcare data, physicians and researchers are largely unable today to leverage this data to improve patient care. The vast majority of healthcare data remains disconnected and lacks harmonization and structure. Traditional diagnostic tests are typically based only on a single data modality, such as a blood- based biomarker or a genomic mutation, and do not connect and integrate other forms of relevant clinical data, such as outcomes, or adverse events, or pathology results, which are essential for many clinical decisions.
In order to bring AI to healthcare at scale, we began by rebuilding the foundation of how data flows in and out of healthcare institutions. We established data pipes, going to and from providers, to allow for the free exchange of data between physicians, who interpret data, and diagnostic and life science companies, who provide data. Without this capability, we believe that data could continue to accumulate without impacting patient care. Tempus has built this integrated Platform, and we are now deploying it at scale in the United States in oncology, and other areas, including neuropsychology, radiology, and cardiology, with aspirations to eventually be in all major disease areas globally. Our Platform connects multiple stakeholders within the larger healthcare ecosystem, often in near real time, to assemble and integrate the data we collect, thereby providing an opportunity for physicians to make data-driven decisions in the clinic and for researchers to discover and develop therapeutics.
Tempus is a technology company focused on healthcare that straddles two converging worlds. We strive to combine deep healthcare expertise, providing next-generation diagnostics across multiple disease areas, with leading technology capabilities, harnessing the power of data and analytics to help personalize medicine. Unlike traditional diagnostic labs, we can incorporate unique patient information, such as clinical, molecular, and imaging data, with the goal of making our tests more intelligent and our results more insightful. Unlike other technology companies, we are deeply rooted in clinical care delivery as one of the largest sequencers of patients in the United States. Straddling both worlds is advantageous as we believe Intelligent Diagnostics represent the future of precision medicine, informing more personalized and data-driven therapy selection and development.
Our Platform includes proprietary software and dedicated data pipelines that create a network of healthcare institutions through approximately 450 unique data connections, many of which supply us with complex multimodal data in near real time, across more than 2,000 healthcare institutions that order our products and services. Healthcare institutions supply us with this data in our capacity as a covered entity (for example, when we provide Next Generation Sequencing, or NGS, services on behalf of a patient), or as a business associate (for example, when we provide clinical trial matching services or data de-identification and structuring services). In addition to the data we receive in these capacities, we currently have a limited number of paid license agreements through which we acquire de-identified data directly from healthcare associations or institutions, and in certain circumstances we cover the actual direct costs associated with the technical integrations needed to create a data connection. We then integrate this data into a unified multimodal database through which we offer numerous analytical and decision support capabilities to our customers. We establish dedicated and integrated data connections with healthcare institutions to enhance the information we provide in our clinical reports, to increase the effectiveness of our clinical trial matching services, and to enable our AI Applications product line, which we believe has the ability to transform healthcare.
We have launched a suite of different products derived from our Platform, which have gained significant traction over the past five years. To date, our offerings have been used by approximately 95% of the largest public pharmaceutical companies based on 2023 revenue, and our clinical NGS volume in oncology rose from approximately 31,000 samples in 2018 to approximately 288,000 samples in 2023. Through March 31, 2024, our offerings have been used by more than 7,000 physicians across hundreds of provider networks, including more than 65% of all academic medical centers in the United States. Our database of multimodal, de-identified records has grown to be more than 50 times the size of The Cancer Genome Atlas, the largest public genomic dataset that we know of in oncology. We also now have more than 200 petabytes of data in our cloud environment. Between our sequencing and data collection efforts, we are connected in some way to more than 50% of all oncologists practicing in the United States. Our access to broad and diverse data serves as the basis for our ability to train generative AI models, and we believe our relationships with healthcare institutions provide us with proprietary data to deliver on the promise of AI in healthcare.
We originally set out to build a sustainable business model in oncology as our first proof of concept. To date, we have focused primarily on establishing and growing our oncology business, which represents the majority of both the data we have amassed and our revenue. Even though our cancer business was at an early stage, we next expanded into neuropsychology, as we believed our model was extensible across disease areas. Having gained early traction in depression, we then expanded into the radiology and cardiology categories. Each time we enter a new disease area we look to expand upon the model we deployed in oncology by developing Intelligent Diagnostics connected to clinical data, and by leveraging large amounts of de-identified data to advance patient care and accelerate drug discovery and development. Once we obtain sufficient data, which we can leverage as a proprietary training data set for generative AI applications, we expect to deploy our AI and machine learning capabilities to build algorithmic diagnostics at scale across diseases.
To manage our growth, we have assembled a world-class team of approximately 2,300 employees, including hundreds with diverse expertise across genetics, molecular and computational biology, bioinformatics, regulatory affairs, medical, product and engineering, and data science. Roughly one-third of our team is technical, with approximately 250 PhDs and MDs on staff.
We generated total revenue of $188.0 million, $257.9 million, $320.7 million and $531.8 million in the years ended December 31, 2020, 2021, 2022 and 2023, respectively, and $115.6 million and $145.8 million in the three months ended March 31, 2023 and 2024, respectively. Revenue generated from COVID-19 testing was $89.5 million, or 47.6% of our total revenue, $94.7 million, or 36.7% of our total revenue, $22.2 million, or 6.9% of our total revenue, and $2.7 million, or 0.5% of our total revenue for the years ended December 31, 2020, 2021, 2022 and 2023, respectively. Revenue generated from COVID-19 testing was $2.6 million, or 2.3% of our total revenue, and $0 for the three months ended March 31, 2023 and 2024, respectively. We stopped offering COVID-19 PCR diagnostic tests in the first quarter of 2023. We incurred net losses of $289.8 million and $214.1 million in the years ended December 31, 2022 and 2023, respectively, and $54.4 million and $64.7 million in the three months ended March 31, 2023 and 2024, respectively. We generated adjusted EBITDA of $(238.8) million and $(154.2) million in the years ended December 31, 2022 and 2023, respectively, and $(45.9) million and $(43.9) million in the three months ended March 31, 2023 and 2024, respectively. Adjusted EBITDA is a non-GAAP financial measure. For a reconciliation of adjusted EBITDA to net loss, the most directly comparable financial measure stated in accordance with generally accepted accounting principles in the United States of America, or GAAP, and for additional information about adjusted EBITDA, see the section titled “Management’s Discussion and Analysis of Financial Condition and Results of Operations—Non-GAAP Financial Measure” in this prospectus.
The Tempus Platform
The Tempus Platform combines multiple elements into a vertically integrated infrastructure that enables us to ingest data from providers, structure and harmonize the data into a common database, provide laboratory diagnostic testing, and deliver personalized results that provide clinical context by leveraging our data. We offer closed-loop, full-stack, bi-directional integrations between a clinician’s desktop and our laboratory diagnostic capabilities, analytics platform, and vast repository of multimodal data. The key elements of our Platform, represented in the diagram below, collectively help power a variety of healthcare applications for providers and life sciences researchers. We believe each of these elements is difficult for competitors to replicate, and together represent a significant competitive advantage.
Data Ingestion and Acquisition
We ingest healthcare data in near real time and at scale, including molecular, clinical, and imaging data. We developed the software infrastructure and dedicated data pipelines to aggregate large amounts of multimodal data directly from healthcare institutions. Our software connects to a provider’s electronic health record, or EHR, system, data warehouse, or third-party data provider to pull relevant structured and unstructured data that the provider has agreed to make available to us, or which is necessary to provide our diagnostic tests, including longitudinal follow-up data in certain circumstances. We have established relationships with hundreds of provider networks in the United States, including more than 65% of all academic medical centers. In addition to healthcare providers, we work with numerous industry associations, including the American Society of Clinical Oncology, or ASCO, to structure and distribute the cancer data that they collect as part of CancerLinq, which is their oncology data effort, as well as ONCare Alliance, LLC (surviving entity after merger of the National Cancer Care Alliance and the Quality Cancer Care Alliance), or ONCare Alliance, and others. We also generate data in our three high-throughput diagnostic testing labs in Chicago, Atlanta, and Raleigh.
Proprietary Data Processing
Once we ingest data, we deploy proprietary clinical data abstraction tools, including natural language processing, optical character recognition, and our abstraction software, to structure, harmonize, and de-identify the data we collect. We have developed various software tools, including algorithmic agents that leverage large language models, to streamline and help secure this process. Once appropriately de-identified, we store the data in our multimodal database.
Our Proprietary Multimodal Database
We believe most healthcare databases lack real-time functionality, depth among data types, and the scale of matched clinical and molecular records needed to meaningfully improve therapeutic research and development. Tempus is attempting to solve this problem by democratizing the use of near real-time molecular, clinical and imaging data by embedding our solution into the clinical care of patients. As our testing volume has grown, and as our dedicated data pipelines have expanded, the size of our database has increased exponentially. Since we launched our Platform in 2016, Tempus has amassed over 900 million documents, across more than 5.6 million de-identified patient records, including approximately 1.3 billion pages of rich clinical text that we use to train our large language models. The database also includes over 1,000,000 records with imaging data, more than 900,000 with matched clinical records linked with genomic information, and more than 220,000 with full transcriptomic profiles. Within oncology specifically, we believe this represents one of the largest and most comprehensive molecular libraries of cancer patients in the world. The breadth of our database, the quality and diversity of our data, as well as its regularly updating nature, allow us to offer a variety of AI-enabled solutions to the market. We believe our unique data set enables us to bring the benefits of generative AI and large language models to healthcare, as our curated, multimodal database can be used as a proprietary training set to build a variety of AI-based applications, which we intend to deploy through our existing network and distribution platform. We also retain the rights to broadly commercialize de-identified data. As the amount of data in our cloud environment continues to grow from its current size of more than 200 petabytes, we believe new AI applications and opportunities will emerge that are only possible with scale, driving further innovations in patient treatment. The graph below shows our historical database growth and its composition as of December 31, 2023:
Proprietary Software Tools and Solutions
We have developed numerous software tools that power our Platform, making our services accessible to multiple constituencies within the healthcare ecosystem and creating a back-end infrastructure that supports our various product lines. We employ AI techniques including neural networks, deep learning, large language models, and other statistical learning techniques to generate patient-specific insights. We are able to not only train and validate some of these AI models for research use, but we can also develop them into clinical-grade algorithmic tests, or Algos, and deploy them clinically as part of routine care. As our data advantage and system architecture continue to improve, we believe our existing Intelligent Diagnostics will gain further adoption thereby accelerating our ability to deploy technologies, including AI Applications, in the clinical setting.
Our Three Product Lines
We have developed multiple products derived from our Platform that are designed to create an economic model that allows us to invest in structuring and harmonizing data, which is a necessary precursor for deploying AI at scale.
Our products are organized under three product lines, each of which is designed to enable and enhance the others, thereby creating network effects in each of the markets in which we operate. We started in Genomics by generating large amounts of molecular data, which in turn gave rise to our Data and Services business through which we licensed de-identified data, which at scale has allowed us to provide a series of data-related services to our life sciences customers, such as clinical trial matching, or Trials. Over time, we expect these products (Genomics + Data and Services) to facilitate deployment of our AI Applications, or Algos, product line through which we will leverage AI models to help route patients to the optimal therapy and advance research more broadly. Our business model allows both Tempus and our customers to unlock value from the data we make available in different ways across our different product lines. We believe these network effects provide a unique advantage to our business, as the compounding value of each data record in our database serves to enhance our competitive advantage. The more data we collect, the smarter our tests become, the more applications we can launch, the more physicians join our network, further growing our database, making our tests smarter for clinicians and our database more valuable for researchers.
We describe below our three product lines—Genomics, Data and Services, and AI Applications. Genomics revenue was $151.9 million, $195.0 million, $198.0 million and $363.0 million for the years ended December 31, 2020, 2021, 2022 and 2023, respectively, and $82.1 million and $102.6 million for the three months ended March 31, 2023 and 2024, respectively. Data revenue was $36.1 million, $62.8 million, $122.7 million and $168.8 million for the years ended December 31, 2020, 2021, 2022 and 2023, respectively, and $33.6 million and $43.3 million for the three months ended March 31, 2023 and 2024, respectively. For the years ended December 31, 2020, 2021, 2022 and 2023, Genomics represented 81%, 76%, 62% and 68%, respectively, and Data represented 19%, 24%, 38% and 32%, respectively, of our total revenue. Revenue generated from AI Applications is reported within Data and services in our Consolidated Statement of Operations and was $1.4 million and $5.5 million for the years ended December 31, 2022 and 2023, respectively. Revenue generated from AI Applications was less than $1.0 million for each of the three months ended March 31, 2023 and 2024.
Genomics
Our Genomics product line leverages our laboratories to provide NGS diagnostics, molecular genotyping, and other anatomic and molecular pathology testing to healthcare providers, life sciences companies, researchers, and other third parties. We operate robotic sequencing labs in Chicago, Atlanta, and Raleigh, with automated bioinformatics and variant classification and reporting that allow us to operate as a high-quality, low-cost NGS provider that broadly serves the market. Our labs are certified by the Clinical Laboratory Improvement Amendments, or CLIA, and accredited by the College of American Pathologists, or CAP. However, unlike other laboratory diagnostic testing providers, many of our tests are connected to clinical data in some manner, which allows our suite of tests to be self-learning, becoming more accurate and precise with each new test that we run. Our current primary assays include:
We are continuing to validate xM, a high coverage methylation sequencing assay for monitoring for cancer recurrence and minimal residual disease, which we intend to launch in 2024, initially in colorectal cancer with the potential to expand into additional indications.
As we have continued to expand our laboratory testing offerings and scale, in addition to increasing reimbursement for our tests, we have achieved continued improved margins. Genomics margin, adjusted for the impact of COVID-19 testing, was (21%), 4%, 25%, 48%, 45% and 48% for the years ended December 31, 2020, 2021, 2022 and 2023 and the three months ended March 31, 2023 and 2024, respectively.
Our cancer tests have gained wide market adoption and allowed us to amass what we consider to be one of the largest de-identified clinical and molecular oncology datasets in the world, which we make available to physicians and life sciences companies. Because our Platform is designed to be extensible across disease areas, we hope, over time, to have similar success in neuropsychology, cardiology, and the other disease categories in which we choose to expand.
Data and Services
Our Data and Services product line facilitates drug discovery and development for life sciences companies through two primary products, Insights and Trials. Through our Insights product, we license libraries of linked clinical, molecular, and imaging de-identified data and provide a suite of analytic and cloud-and-compute tools to pharmaceutical and biotechnology companies. Historically, datasets in healthcare have been siloed, often lacking important contextual information such as outcome and treatment response data. Our Insights offering is designed to address this void across multiple diseases, enabling pharmaceutical and biotechnology companies to improve decision-making across the drug lifecycle—from discovery, research and development, and, ultimately, commercialization.
Customers either pay us on a per file basis or through multi-year data licensing agreements to access our de-identified database of clinical records. We work with 19 of the 20 largest public pharmaceutical companies based on 2023 revenue, and as of December 31, 2023, we have signed contracts with a remaining total contract value of more than $920.0 million, which includes approximately $300.0 million in additional potential future contractual opt-ins. See “Business” for additional discussion regarding our remaining total contract value.
We retain broad rights to commercialize most of the de-identified data we collect, and we are able to license the same de-identified records to multiple customers. Additionally, because many of our data profiles regularly update with clinical outcome and response data, the value of a single de-identified record can increase over time.