Artificial Intelligence (AI) in Cancer Market

The global artificial intelligence (AI) in cancer market reached USD 2.20 billion in 2025 and is estimated at USD 2.64 billion in 2026. Revenue is projected to reach USD 11.34 billion by 2034, a 20.0% CAGR over 2026–2034. The market covers oncology-attributable AI software, clinical workflow services, pathology analysis and genomic interpretation rather than all medical AI or the full value of diagnostic scanners.

Artificial intelligence used in oncology analyzes radiology images, pathology slides, genomic sequences and clinical records to support cancer detection, diagnosis, treatment planning and longitudinal monitoring. Commercial revenue includes oncology-specific software subscriptions, validated image-analysis services, precision oncology algorithms and attributable platform licensing. General-purpose cloud infrastructure, scanner hardware and unrelated cardiovascular or neurological AI are excluded. Human clinicians retain responsibility for interpretation and treatment decisions within the intended use of each product.

The International Agency for Research on Cancer estimated 20.6 million new cancer diagnoses in 2024, with lung, breast and colorectal cancers accounting for major portions of the burden. High imaging volumes, workforce shortages and more complex molecular testing increase the appeal of validated decision-support systems. Screening algorithms can prioritize suspicious mammograms or chest images, while pathology models identify regions requiring expert review. Commercial deployment still depends on local clinical validation and integration into established care pathways.

FDA medical-device review distinguishes marketing authorization from research-stage algorithm development. The FDA maintains a list of authorized AI-enabled medical devices and reports more than 1,600 authorizations across all medical specialties as of September 2026. Ibex announced FDA 510(k) clearance for Prostate Detect in February 2025; Paige Prostate also has U.S. marketing authorization. Lunit filed a 510(k) application for its mammography-based five-year risk prediction model in December 2025, a submission that must not be described as clearance.

Oncology AI spending also extends into molecular profiling and pharmaceutical research. Tempus announced multimodal foundation-model work using de-identified genomic, imaging and clinical records, and PathAI supplies pathology workflows for laboratories and biopharma sponsors. The distinction between an approved companion diagnostic assay and an AI algorithm is important: Tempus xT CDx is a regulated genomic test, while broader machine-learning analytics require their own intended-use and evidence assessment.

Source: IARC Global Cancer Statistics; FDA AI-enabled Medical Devices; Ibex Prostate Detect.

The market measures oncology-attributable AI software and related analysis services deployed for cancer screening, pathology, molecular interpretation, treatment support and research. Hardware revenue and non-oncology AI services are excluded.

Growing cancer incidence increases image and pathology workload

IARC reported 20.6 million new cancer cases in 2024 and 9.8 million cancer deaths worldwide. Breast, lung and colorectal cancers generate substantial imaging, pathology and molecular-testing workloads. Radiology teams review mammograms and chest examinations, while pathologists interpret increasingly complex biopsy specimens and biomarkers. AI software can flag suspicious findings or quantify slide features for expert assessment, but performance must be validated for the intended patient population and workflow. Higher screening throughput and more comprehensive cancer care create purchasing opportunities for products that improve prioritization without introducing unsafe false positives.

Regulated intended use determines clinical adoption

The FDA regulates AI-enabled device functions through existing medical device pathways, including 510(k), De Novo and premarket approval. Its September 2026 list includes more than 1,600 AI-enabled devices across specialties, not all oncology products. Ibex Prostate Detect received a 510(k) clearance in 2025, while Lunit announced a 510(k) submission for a separate breast-risk model later that year. Hospitals must distinguish a cleared clinical indication from investigational decision support. Regulatory review, quality management and post-deployment performance monitoring create costs but can improve trust in products used for cancer care.

Digital pathology converts slides into analyzable data

Philips expanded its partnership with Ibex in March 2025 to connect digital pathology infrastructure with AI tools for prostate, breast and gastric cancer. Whole-slide scanners and image-management systems create the technical basis for computer-assisted tissue review, including prioritization and biomarker quantification. PathAI also provides AISight digital pathology workflows and works with laboratories on deployment. Clinical adoption depends on scanner compatibility, stain variability, image quality and the responsibility of the reviewing pathologist. Laboratories need validated processes rather than a standalone algorithm disconnected from case management.

Molecular complexity creates demand for multimodal interpretation

Tempus reported work on multimodal oncology foundation models drawing on de-identified genomic, imaging and clinical records. Its HRD-RNA algorithm, announced in February 2026, analyzes gene-expression patterns relevant to homologous recombination deficiency. Such products address the challenge of interpreting large biomarker panels and longitudinal cancer data, but analytical performance is not equivalent to a therapy-selection approval. Clinical use must match the test label, applicable regulations and evidence for the specific treatment decision. Pharmaceutical sponsors also purchase analytical services for patient stratification and exploratory research.

Privacy and workflow integration limit procurement speed

Cancer AI systems often require pathology slides, radiology images, sequencing results and longitudinal records to operate effectively. Institutions must protect sensitive patient information, control data access and evaluate whether algorithms generalize across scanners and populations. PathAI announced a nationwide Labcorp digital pathology deployment in February 2026, showing the scale of integration needed across laboratory networks. Hospitals and vendors also need interoperable image archives, review queues and audit trails. These requirements lengthen procurement cycles, especially where local rules restrict transfer of patient data between jurisdictions.

Sources: IARC GLOBOCAN; FDA AI-enabled medical devices; Ibex Medical Analytics; Philips; PathAI; Tempus AI.

Mammography AI moves from pilots to organized screening programs

The UK government announced in September 2025 a national research platform intended to support large-scale AI screening trials involving 660,000 women. Lunit reported a five-year agreement to deploy mammography AI across the UAE SEHA network in March 2025. These projects require integration with screening schedules, radiologist reading procedures and quality assurance. Their clinical value depends on prospective evaluation of detection, recall and interval-cancer outcomes, not simply retrospective image classification. Vendors with evidence from multiple screening populations can compete for public-sector deployments.

Prostate pathology gains cleared decision-support tools

Ibex announced FDA 510(k) clearance for Prostate Detect in February 2025, describing heatmaps intended to help pathologists identify missed cancer in biopsy slides. Paige Prostate has U.S. marketing authorization for assistance with prostate cancer detection. In March 2026, Ibex reported an HNL Lab Medicine deployment covering prostate biopsy and resection workflows. These products are assistive tools within pathology practice, and their indicated uses should not be extended to unrelated tissue types. Adoption depends on digital slide quality, clinical review and measured error reduction.

Pan-cancer platforms combine genomic and clinical signals

Tempus announced expanded collaboration with Bristol Myers Squibb in May 2026 to apply AI and multimodal real-world data to clinical development programs. Its separate work on foundation models combines de-identified genomic, imaging and longitudinal clinical information. Biopharma sponsors may use these tools to refine trial cohorts, characterize biomarkers and generate research hypotheses. The outputs still need prospective confirmation before being treated as validated clinical endpoints or companion diagnostics. Data licensing and analytics services create revenue distinct from hospital imaging software subscriptions.

AI-enhanced radiotherapy planning adds a treatment workflow

GE HealthCare introduced AI-enabled radiation oncology workflow enhancements at ESTRO 2025, including integration of third-party applications and MR-guided processes. Segmentation, treatment planning and adaptive imaging may reduce repetitive contouring work when clinical teams validate results. Cancer centers must assess geometric accuracy and plan quality before applying algorithm-generated contours or recommendations. The commercial opportunity includes oncology software licenses and integration services rather than the entire value of linear accelerators or imaging scanners. Multidisciplinary radiotherapy departments must verify that imaging and treatment-planning applications exchange data without altering approved clinical workflows.

Cloud pathology networks support distributed specialist review

PathAI announced a Precision Pathology Network in July 2025 and later expanded its Labcorp collaboration to deploy AISight Dx across laboratories in February 2026. Digital slide management supports remote expert review and consistent application of analytical tools, but deployment requires secure data handling and integration with laboratory information systems. Philips and Ibex likewise expanded their interoperable pathology offering in 2025. These developments shift procurement toward integrated platforms that can support clinical workloads and research analytics under separate validation and governance requirements.

Technology categories are defined by the principal input and clinical workflow: radiology images, digital tissue slides, genomic and multi-omics records, or other structured clinical information. A multimodal system is allocated by its main commercial indication so its revenue is not counted in several categories.

Other clinical-data AI covers oncology decision support based primarily on electronic records, symptoms or treatment history. Its adoption depends on documented clinical benefit, integration with existing records and controls preventing unsupported automated treatment recommendations.

Application groups follow the principal tumor indication for which the algorithm is sold or used. Breast and lung cancer screening rely heavily on imaging; prostate and gastrointestinal applications are prominent in digital pathology. Pan-cancer analytics are allocated by actual use rather than counted across every tumor group.

Colorectal and gastrointestinal cancers, hematologic malignancies and rare tumors create additional uses for pathology, genomic interpretation and treatment support. Smaller disease-specific datasets can make external validation more difficult, particularly when algorithms are transferred across hospitals.

Oncology AI buyers include hospitals, diagnostic laboratories, biopharmaceutical companies and research institutions. The end-user classification follows the organization paying for the software or analytical service, not every clinician who sees an algorithm-assisted result.

Institutional purchasing can involve a radiology department, pathology service or centralized IT team. Cybersecurity review, interoperability testing and user training often determine deployment speed even when the underlying algorithm has regulatory authorization. A hospital may also require prospective monitoring to detect performance drift after scanner or staining changes.

Clinical Stage identifies the principal workflow supported by an AI solution: screening, diagnosis, treatment planning or follow-up. These functions require different clinical evidence and cannot be inferred from the word AI alone. A tool used for research does not automatically qualify as an approved diagnostic device.

Therapy monitoring and prognosis remain active development areas because longitudinal datasets are complex and treatment effects vary by tumor biology. Hospitals should distinguish exploratory risk scores from results supported by clinical validation and applicable device authorization.

Regional demand is shaped by cancer screening infrastructure, digital pathology adoption, regulatory pathways and availability of labeled imaging datasets. Country evidence below distinguishes actual deployments from research policy and broader digital-health initiatives.

Oncology AI competition spans specialist software developers, molecular data companies and established imaging-system manufacturers. Buyers evaluate clinical evidence, regulatory indication, integration cost and ongoing model monitoring rather than algorithm accuracy in isolation. Different vendors address breast imaging, prostate pathology, oncology molecular interpretation and radiation planning.

Breast imaging vendors compete on prospective screening evidence, radiologist workload and integration with mammography systems. iCAD describes ProFound tools for detection and breast density assessment, while Lunit has pursued screening contracts and risk prediction research. Pathology providers face different challenges: stain variation, whole-slide image quality and the need to support a pathologist’s final diagnosis. Philips and Ibex expanded interoperability in 2025, and PathAI announced a Labcorp deployment in 2026. These partnerships can lower implementation friction but do not remove the need for product-specific clinical validation.

Molecular and multimodal AI adds another competitive layer. Tempus announced collaboration with Bristol Myers Squibb on oncology clinical development and has reported foundation-model research combining clinical, imaging and genomic information. Large manufacturers such as GE HealthCare and Siemens Healthineers can distribute oncology-related AI within existing radiology infrastructure, while specialist vendors differentiate through disease-specific evidence. Commercial success depends on recurring software use, workflow reliability and the ability to document clinical benefit across institutions.

• Lunit
• PathAI
• Tempus AI
• Ibex Medical Analytics
• Paige
• iCAD / DeepHealth
• Philips
• GE HealthCare
• Siemens Healthineers
• Therapixel

Oncology AI revenue is projected to rise from USD 2.64 billion in 2026 to USD 11.34 billion in 2034, equivalent to 20.0% CAGR. Clinical uptake will depend on authorized intended uses, integration into screening and pathology workflows, and evidence that software improves decisions without unacceptable errors.

Medical imaging is expected to remain the largest technology category because mammography and chest imaging can be deployed through established screening and radiology pathways. The UK is investing in broader evaluation and deployment of AI imaging tools, while the UAE has a documented mammography software agreement. Providers must demonstrate performance on the populations and scanners where products are used. Higher case volumes alone do not establish clinical value; recall rates, missed cancers and follow-up capacity are important procurement considerations.

Digital pathology is positioned to expand as laboratories digitize slide workflows. Ibex and Paige provide specific prostate cancer decision-support products, while PathAI and Philips support the infrastructure required for image management and algorithm integration. Expansion into additional tissues requires separate validation and, where applicable, regulatory authorization. Laboratories are likely to favor systems that connect slide scanning, case allocation and pathology review rather than stand-alone algorithms that add manual steps.

Precision oncology research will generate demand for multimodal analytics and biomarker interpretation. Tempus has described models combining genomic, imaging and clinical records and agreements with pharmaceutical sponsors. The commercial boundary remains important because genomic laboratory testing revenue, AI licensing and general-purpose data services are not interchangeable. Research algorithms must be distinguished from clinically validated companion diagnostics, especially when treatment selection is involved. Companies that provide auditable evidence and controlled model versions can better serve regulated customers.

Forecast risks include data privacy restrictions, inconsistent performance across demographic groups, limited digital pathology infrastructure and reimbursement uncertainty. FDA device oversight and European clinical evaluation requirements may slow product launches but encourage investment in rigorous validation. Competitive pricing could reduce software revenue per study even as adoption grows. Sustained growth therefore depends on measured workflow benefits, secure integration and clinical outcomes rather than extrapolating research accuracy to routine patient care.

Coverage measures oncology-specific AI software and analytics across screening, tissue pathology, genomics, treatment planning and research use cases.