🏭Industry Overview
Updated September 27, 2026Health care is the largest sector of the US economy at over $4.5 trillion in annual spending — nearly 18% of GDP. It encompasses hospitals, physician practices, pharmaceutical manufacturers, medical device companies, insurance, and social services. The US healthcare system faces persistent challenges: rising costs, physician burnout driven by administrative burden, diagnostic errors causing preventable harm, drug development costs exceeding $2 billion per approved drug, and stubborn health disparities across demographic groups.
💡The AI Opportunity
AI represents the most significant opportunity to improve health outcomes while reducing costs in the history of modern medicine. FDA-cleared AI now screens for diabetic retinopathy, certain cancers, and stroke with performance matching specialist physicians. Clinical documentation AI reduces physician note-writing time by over 50%, directly addressing the burnout epidemic. AI drug discovery companies are compressing the 12-year drug development timeline to 4–5 years. Genomics AI matches cancer patients with targeted therapies based on their tumor's molecular profile. The combination of richer health data, powerful AI models, and evolving regulatory clarity is accelerating adoption across the entire healthcare ecosystem.
🤖AI in Action
Medical imaging AI (Radiology: Aidoc, Viz.ai) detects anomalies in X-rays, CT scans, and MRIs with accuracy matching or exceeding radiologists. AlphaFold 3 (DeepMind) has transformed drug discovery by predicting protein-ligand interactions. Clinical documentation AI (Nuance DAX, Suki) transcribes physician-patient conversations into structured EHR notes, reducing physician burnout. Predictive AI identifies patients at high risk of sepsis, readmission, or deterioration. AI-powered virtual nursing assistants monitor patients between visits. In April 2026, Google DeepMind unveiled an AI co-clinician research initiative built on Gemini and Project Astra — a triadic care model where the AI works alongside a supervising physician with a separate "Planner" module monitoring safety boundaries. Academic collaborators include Harvard Medical School and Stanford Medicine, with phased trusted-tester evaluations planned across the US, India, Australia, New Zealand, Singapore, and UAE. Research-only at this stage, not approved for clinical diagnosis or treatment. On April 30, 2026, researchers at Harvard Medical School and Beth Israel Deaconess Medical Center published a peer-reviewed Science paper in which OpenAI's o1 reasoning model outdiagnosed two emergency-room physicians on text-only EHR inputs (67% versus 55% and 50% across 76 triage cases), with authors calling for prospective trials before any clinical use.
Anthropic's Beneficial Deployments team has committed $200 million across four years to the Gates Foundation, structured as a mix of grants, Claude usage credits, and technical support targeting global health (with polio, HPV, and preeclampsia and eclampsia among named disease targets), life sciences, education, and economic mobility — with regional focus on sub-Saharan Africa, India, and other low- and middle-income countries. Launch partners include the Institute for Disease Modeling and the Global AI for Learning Alliance. The commitment is the largest single philanthropic deployment from a frontier lab to date and represents a meaningful structural channel for frontier AI capability to land in global health workflows alongside the commercial-cloud distribution paths, signaling that frontier labs intend to be measured by social-impact deployments rather than only by commercial books.
In May 2026, Pennsylvania filed suit against Character.AI after a chatbot named "Emilie" claimed to be a licensed psychiatrist and produced a fabricated state medical license number while discussing depression treatment with state investigators. The complaint cites Pennsylvania's Medical Practice Act and is the first state action specifically targeting AI chatbots that impersonate licensed medical professionals — distinct from Kentucky's January 2026 Character.AI suit, which focused on harm to minors. For the broader healthcare and social-assistance category, the Pennsylvania filing sets a credentialing-fraud precedent that will likely shape how consumer-facing AI products represent themselves to users on any clinical or quasi-clinical topic, and expect platform-level disclosures and identity-verification flows to harden across the wider healthcare AI surface in response.
On the prevention side, AI is beginning to design vaccines directly. University of Cambridge researchers moved into human trials the first vaccine whose central component was AI-designed — a universal coronavirus candidate aimed at protecting against entire families of related viruses, including threats that have not yet emerged. Still early-stage, with a small safety trial followed by a larger immune-response study, it points toward AI-assisted pandemic preparedness as a complement to AI's established roles in diagnosis, imaging, and clinical documentation.
Enterprise deployment inside hospitals is also moving from pilots to whole-system rollouts. In July 2026 Sheba Medical Center in Ramat Gan became the first hospital outside the United States to deploy OpenAI's ChatGPT for Healthcare platform across its network, giving physicians, nurses, researchers and administrative staff a clinical-reasoning tool that synthesizes peer-reviewed studies and clinical guidelines and attaches citations with publication dates to each response so clinicians can verify the evidence. Sheba can load its own protocols so answers align with approved local standards, OpenAI does not train on the hospital's data, and clinicians retain the final decision. The governance details matter as much as the capability: the pattern emerging across large deployments is evidence with provenance, institution-specific grounding, data isolation, and an explicit human decision-maker — which is the difference between clinical AI and a general-purpose chatbot answering medical questions.
📊Impact on Jobs
Healthcare AI is primarily augmenting clinical roles rather than replacing them — physicians, nurses, and therapists remain essential. Medical transcription and coding roles face significant automation. Insurers argue that AI-assisted coding is also raising bills: in September 2026 the Blue Cross Blue Shield Association said the share of inpatient stays billed as medically complex rose from 37 percent in early 2023 to 40 percent by the end of 2025 with no sign of a change in care, costing its member plans an estimated 942 million dollars over two years. Hospitals dispute that framing. Radiology and pathology workflows are changing as AI pre-reads studies. New roles in clinical AI validation, health AI product management, and medical data engineering are growing. The FDA has cleared over 900 AI-enabled medical devices.
Keep track of AI in your sector
- Save the sectors you follow
- Get ⚡ alerts when their tools and companies change
- The AI tools that actually ship here, from 1,000+ AI tool profiles
- Today’s top AI Stories — the day’s most important AI news, free
Swipe for Recommended for you and My AI Tools
Your AI Hub — sample data. See desktop view example
🛠️Top AI Tools in This Industry
Microsoft clinical AI assistant (formerly Nuance DAX) that listens to patient-physician encounters and auto-generates SOAP notes, visit summaries, referral letters, and after-visit instructions. Agentic capabilities handle prior authorizations and medication refills. Deployed at 600+ health systems.
AbridgeAmbient clinical-documentation AI for physicians — listens to patient encounters and generates structured notes that integrate directly with the EHR.
Suki AIVoice-driven AI assistant for clinical documentation — used by physicians across primary care and specialty practices for hands-free note-taking.
Ambience HealthcareEnterprise ambient AI platform — drafts clinical notes from the visit conversation and supports coding and revenue-integrity workflows across health systems.
AugmedixAmbient clinical-documentation AI used across acute care, ambulatory, and skilled-nursing facilities — combines AI scribes with human review for high-accuracy notes.
NablaAmbient AI medical scribe (a European leader) that generates structured clinical notes from patient encounters and is expanding into agentic clinical workflows.
AidocAI radiology triage platform with 30 FDA-cleared tools. First comprehensive foundation model AI triage detecting 14 critical findings from a single abdominal CT with 97% sensitivity.
Viz.aiMedical AI platform detecting time-sensitive conditions from imaging. Nearly 2,000 hospitals, 230 million lives covered. FDA-cleared tools for stroke, hemorrhage, pulmonary embolism, and more.
RapidAIStroke and neurovascular imaging AI — automated large-vessel-occlusion detection, CT perfusion, and triage that speeds treatment across stroke networks. The principal Viz.ai rival.
Qure.aiGlobally deployed radiology AI for chest X-ray, head-CT, and lung-nodule detection and triage — 26 FDA indications across 9 products, used in 100-plus countries.
HeartFlowFDA-cleared AI analysis of coronary CT — non-invasive FFR-CT blood-flow assessment and plaque analysis to diagnose heart disease, reimbursed and widely clinically validated.
CleerlyAI coronary-CT analysis that quantifies and characterizes arterial plaque (AI-QCT) to assess heart-disease risk; FDA-cleared with a Category I CPT code from January 2026.
Rad AIGenerative-AI radiology workflow that drafts report impressions and automates reporting and worklist tasks, tuned to each radiologist's style.
PaigePaige, part of Tempus AI, sells AI that reads digitized tissue slides, including the FDA-authorized Paige Prostate Detect, and licenses the Virchow and PRISM pathology foundation models, mostly for non-commercial research.
PathAIDigital pathology AI with FDA-qualified tools for clinical trials (AIM-MASH) and diagnostics (AISight Dx). Breakthrough Device Designation for dermatopathology (March 2026). Deployed at Labcorp.
LunitPublicly traded cancer-detection AI — FDA-cleared chest X-ray and mammography screening plus oncology pathology biomarkers, distributed globally and via imaging-device OEMs.
DermaSensorThe first FDA-cleared AI skin-cancer detection device for primary care — a handheld spectroscopy tool that assesses lesions for all three common skin cancers at the point of care.
LumineticsCoreThe first-ever FDA-authorized autonomous AI diagnostic — detects diabetic retinopathy from retinal images in primary-care settings without a specialist read. By Digital Diagnostics.
FDA-cleared dental AI for radiograph interpretation and patient communication — used by dental service organizations and insurers.
FDA-cleared AI for dental radiograph analysis — detects caries, periapical lesions, and bone loss to support dentist diagnoses.
OpenEvidenceAI clinical decision support — evidence-based answers to medical questions at the point of care, with citations to the primary literature. Used by a large share of US physicians.
Glass HealthClinician-facing decision support that generates ranked differential diagnoses and evidence-based care plans, plus ambient documentation — used by 100,000-plus US clinicians and trainees.
K HealthClinical-AI virtual care embedding AI clinical agents into health-system EHR workflows — powering in-EHR virtual urgent and primary care with Penn Medicine and Northwell.
Epic CosmosEpic's embedded AI: the Art clinician assistant (In Basket drafting and Chart with Art ambient notes), the Penny revenue-cycle agent, the patient-facing Emmie, and the Curiosity predictive models built on the Cosmos research database of 310 million patients.
InnovaccerHealthcare data and AI platform unifying clinical, claims, and operational data for hospitals — supports population health, care management, and physician workflows.
Hippocratic AISafety-focused, patient-facing generative-AI agents that handle non-diagnostic clinical outreach — post-discharge calls, chronic-care check-ins, and care management — at scale, escalating to human staff when needed.
CommureAI-native healthcare platform unifying ambient documentation, revenue-cycle management, and hospital operations across hundreds of provider organizations; parent of ambient-scribe Augmedix.
QventusAI operational assistants for hospitals that automate patient flow, discharge planning, and OR scheduling — targeting capacity, length-of-stay, and throughput.
CodaMetrixAutonomous AI medical coding for large health systems — translates clinical documentation into billing codes across specialties, with audit trails and human oversight.
Cohere HealthAI prior-authorization and utilization-management automation for health plans and providers — speeding approvals for appropriate care and cutting denials and delays.
Voice-AI agents that automate healthcare phone calls to payers — benefit verification, prior-auth status, and member services — built on a knowledge graph from millions of calls.
RegardAI that reads EHR chart data to surface missed or under-documented diagnoses, draft care plans, and automate notes — improving diagnosis capture and documentation integrity.
Bayesian HealthContinuous, real-time AI monitoring of inpatients for sepsis and clinical deterioration — with a landmark FDA clearance for continuous AI sepsis monitoring.
TheatorSurgical-intelligence AI that analyzes operating-room video to support surgeon performance, training, and automated operative documentation across hundreds of thousands of procedures.
PointClickCare AIEHR and care-coordination platform for skilled nursing and senior living with AI features for resident risk prediction, regulatory compliance, and revenue cycle.
Long-term and post-acute care platform with AI features for clinical documentation, billing automation, and operational analytics.
xCuresHealth-AI platform whose Clinical Clarity Engine uses AI to turn scattered, unstructured medical records into structured, decision-ready clinical data for clinicians, researchers, and life-sciences teams.
SWORD HealthAI virtual physical therapy with FDA-cleared motion-sensor technology and licensed PT supervision. Delivers in-clinic-quality care across MSK, women's health (Bloom), and post-surgical rehab to thousands of employer and health plan partners covering millions of members.
Kaia HealthAI-driven digital therapeutics for back pain, knee pain, and COPD pulmonary rehab. Smartphone motion tracking provides exercise feedback comparable to a physical therapist. FDA-cleared for several indications and deployed across major US and European employer health benefits.
WysaConversational AI for mental-health support — clinically validated CBT-based chat assistant used by employers, health plans, and therapy services.
Ash by Slingshot AIConsumer AI therapy chatbot from Slingshot AI, grounded in CBT/DBT/ACT techniques — support, not a replacement for licensed care, in a category whose safety and efficacy are still being validated.
CarePredictWearable AI for activity, fall, and behavior-pattern monitoring in skilled nursing, assisted living, and home settings. ML models on continuous accelerometer data detect routine changes that predict urinary tract infection, falls, and depression days to weeks ahead of human observation.
InspirenAI vision platform (AUGi) for senior-living facilities — tracks falls, exit-seeking, and clinical events from in-room cameras without requiring residents to wear devices. Used by Brookdale, Atria, and other large senior-living operators across thousands of resident rooms.
Neuralink N1Brain-computer interface implant with ~20 patients. 1,024 electrode threads, 9+ bits/sec cursor control. Blindsight visual cortex device has FDA Breakthrough designation.
Synchron StentrodeEndovascular brain-computer interface — no open brain surgery. ~10 patients, COMMAND trial met safety endpoint. First BCI to control Apple Vision Pro.
Precision Layer 7Thin-film cortical interface with 1,024 electrodes placed on brain surface without penetrating tissue. FDA 510(k) cleared. Medtronic partnership.
ParadromicsHigh-data-rate implantable brain-computer interface (Connexus) with an FDA-approved speech-restoration clinical trial — a leading Neuralink rival on a clear regulatory path.
🏢More companies in this sector
Beyond the tool makers above — companies applying AI in Health Care & Social Assistance, including large adopters whose AI work is internal or bought in rather than sold as a product.
AstraZenecaAZNCambridge, UK-based global drug maker (Tagrisso, Imfinzi, Farxiga) and an AI acquirer: it bought pathology-AI startup Modella AI in January 2026 after a 2025 test partnership, alongside AI deals with Tempus, Algen, Immunai, Owkin and CSPC, plus its own QCS computational pathology and Evinova trial software.
BioNTechBNTXMainz-based mRNA drugmaker behind Comirnaty, co-developed with Pfizer, now focused on cancer. Bought London AI company InstaDeep in 2023 for about £362 million upfront plus up to £200 million in milestones, gaining an in-house AI supercomputer and genomics and protein-design models.
BioptimusParis-based builder of foundation models for biology: the H-optimus pathology models (H-optimus-0 open under Apache 2.0) and the multimodal M-Optimus. Incubated by Owkin; $76 million raised; models sold by usage on AWS Marketplace.
BoltzCambridge, Massachusetts public benefit corporation from the MIT team behind the open-source Boltz-1, Boltz-2 and BoltzGen biomolecular models. Sells the hosted Boltz Lab platform and Boltz API, with proprietary BoltzMol-1 and BoltzProt-1 pipelines. $28M seed (January 2026); deals with Pfizer, Takeda and GSK.
Bristol Myers SquibbBMYPrinceton-based drugmaker behind Opdivo, Eliquis and Cobenfy. A full-stack AI adopter: it owns an NVIDIA DGX SuperPOD, made Anthropic's Claude Enterprise its shared platform for 30,000+ employees in May 2026, and buys specialist tools from Schrödinger, Chai Discovery and Evinova.
Butterfly NetworkBFLYPublicly traded maker of handheld whole-body ultrasound on a chip (Butterfly iQ) with AI imaging guidance and interpretation — democratizing point-of-care ultrasound; 150,000-plus devices deployed.Digital DiagnosticsMaker of LumineticsCore (formerly IDx-DR), the first-ever FDA-authorized autonomous AI diagnostic — detecting diabetic retinopathy from retinal images without a specialist read.
DoximityDOCSThe professional network used by most US physicians, now shipping its own clinical AI: Doximity Ask (formerly Doximity GPT) and a free ambient scribe for verified US clinicians. Bought AI clinical-reference startup Pathway in July 2025, and is in litigation with OpenEvidence.
Eli LillyLLYPublic pharmaceutical company (NYSE: LLY), among the world's most valuable, and a major AI adopter that has also productized its internal AI via TuneLab — a platform giving biotechs access to Lilly's AI models trained on proprietary data.Epic SystemsPrivately-held US EHR market leader, founded 1979 by Judy Faulkner. Dominant in large hospital systems (~40% market share among >500-bed hospitals). Epic Cosmos research database covers 310 million patients across 22.6 billion encounters. AI features rolled out 2023-2024 include Inbox-message drafting (in partnership with Nuance/Microsoft), summarization, and predictive insights. Headquartered on a famously elaborate Verona, WI campus. Estimated annual revenue $4-5B.
Gilead SciencesGILDPublic biopharmaceutical company (NASDAQ: GILD) known for HIV and hepatitis C antivirals and Kite cell therapy, and a partnership-first AI adopter: Genesis, Terray, Tempus, Nucleai and Deepcell supply its AI capability. Sells no AI product.
GSKGSKLondon-based drugmaker behind Shingrix, Trelegy and ViiV's HIV medicines. A data-first AI adopter: an in-house AI/ML team fed by bought data and licensed models — Tempus ($70M, 2022), Relation (up to $110M, 2026) and Noetik's virtual cell models ($50M, 2026).
IlluminaILMNSan Diego maker of DNA sequencers whose instruments generate much of the genomic data biology AI trains on. Ships AI interpretation models (PrimateAI-3D, PromoterAI, SpliceAI) and, via its BioInsight unit, the Billion Cell Atlas, a CRISPR perturbation dataset for AstraZeneca, Merck & Co. and Lilly.
insitroAI-native drug developer founded by Daphne Koller that pairs machine learning with high-throughput biology to find disease targets and design medicines. Partners include Bristol Myers Squibb (ALS), Eli Lilly and Gilead; its lead liver-disease siRNA is in IND-enabling studies, with no drug yet in human trials.
InstaDeepLondon-based AI company founded in Tunis in 2014 and owned by BioNTech since 2023. It builds drug-discovery models for its parent and still sells to outside customers, including DeepPCB, a self-serve AI circuit-board router, and DeepChain, a biology AI platform.
IQVIAIQVClinical-research and health-data company (NYSE: IQV) that sells agentic AI to drugmakers through its IQVIA.ai platform, built with NVIDIA. Says 19 of the top 20 pharma companies have begun using its agents. Enterprise contracts only, no self-serve access.
Johnson & JohnsonJNJPharmaceutical and medical-device company (Innovative Medicine + MedTech). A two-track AI adopter: partners with Isomorphic Labs on AI drug design, and builds surgical AI on NVIDIA computing around its Polyphonic ecosystem and Monarch and Ottava robots.
Mayo ClinicNonprofit academic health system in Rochester, Minnesota, that builds AI on its own de-identified patient archive: a frontier clinical model with Microsoft, the Atlas pathology model with Aignostics, a genomic model with Cerebras, and a data collaboration with Merck & Co.
MedtronicMDTOne of the world's largest medical-device companies (NYSE: MDT); maker of GI Genius, the first FDA-cleared AI for real-time polyp detection during colonoscopy, among a growing portfolio of AI-enabled devices.
Merck & Co.MRKPublic pharmaceutical company (NYSE: MRK), known as MSD outside the US and Canada, behind Keytruda and Gardasil. An enterprise-platform AI adopter: a Google Cloud agentic AI deal worth up to $1 billion, plus Mayo Clinic and Protillion discovery partnerships. Sells no AI product. Not Germany's Merck KGaA.
Merck KGaAMKKGYGerman science and technology group, unaffiliated with Merck & Co., that operates in the US as EMD Serono, MilliporeSigma and EMD Electronics. Sells the AIDDISON AI drug-design software, partners with Recursion and Palantir, and supplies materials for AI chips from a €500 million Taiwan megasite.
ModernaMRNAPublic biotechnology company (NASDAQ: MRNA) pioneering mRNA medicines, and a prominent enterprise AI adopter — 750-plus internal AI models, an org-wide ChatGPT Enterprise deployment, and an OpenAI partnership — using AI across mRNA design, manufacturing, and operations. Does not sell an external AI product.
NoetikAI-native cancer biotech in South San Francisco building OCTO foundation models and virtual cell simulations from its own spatial tumor data. GSK licensed the models in January 2026 for $50 million in upfront payments and near-term milestones, plus annual fees.
NovartisNVSSwiss pharmaceutical company (Entresto, Cosentyx, Kisqali) that licenses AI-native discovery models rather than building its own: Isomorphic Labs ($37.5M upfront, up to $1.2B milestones), Generate Biomedicines and Chai Discovery's Chai-3, on top of its in-house data42 data platform.
Novo NordiskNVODanish maker of Ozempic and Wegovy (NYSE ADR: NVO), rebranded as "Novo" in September 2026. A multi-vendor AI adopter with strategic partnerships with OpenAI, AWS and Anthropic, plus NVIDIA research on the Gefion supercomputer. Sells no AI product.
PfizerPFENew York-based drugmaker behind Comirnaty, Paxlovid and Eliquis. An AI adopter that licenses frontier biomolecular models from Boltz (January 2026) and Chai Discovery (June 2026) and has them retrained on its own historical data into private, Pfizer-only versions.
PhilipsPHGDutch health-technology company behind MRI, CT and ultrasound scanners, the Azurion image-guided therapy platform and patient monitors. Its AI ships inside those devices, such as FDA-cleared SmartSpeed Precise MRI reconstruction, with an NVIDIA-backed MRI foundation model still in development.Precision NeuroscienceBrain-computer interface company developing the Layer 7 Cortical Interface — a thin-film microelectrode array with 1,024 electrodes placed on the brain surface without penetrating tissue. FDA 510(k) clearance (April 2025). Partnered with Medtronic (January 2026) to co-develop integrated surgical navigation system. Founded by a Neuralink co-founder.
RocheRHHBYSwiss pharma and diagnostics group (including Genentech) running one of the drug industry's largest AI compute fleets, 3,500+ NVIDIA GPUs, plus Genentech's Prescient Design antibody-ML team, Claude-based research agents, and a pending $750M acquisition of AI pathology firm PathAI.
SanofiSNYParis-based drugmaker behind Dupixent and major vaccine lines, and one of pharma's most vocal AI adopters: staff-wide tools like plai and a Claude-powered assistant, plus AI discovery deals with Exscientia (now Recursion), Owkin and Earendil Labs, the last worth up to $2.56 billion.
Siemens HealthineersSMMNYGerman medical-imaging and diagnostics maker that builds AI into its CT, MRI and interventional systems, with roughly 100 FDA AI-enabled device authorizations, second only to GE HealthCare. Owns Varian (bought for $16.4 billion in 2021); Siemens AG holds about 67%.
Takeda PharmaceuticalTAKTokyo-based biopharma behind Entyvio, Takhzyro and Qdenga, and a large AI adopter. It runs parallel AI discovery deals with Nabla Bio, Iambic (up to $1.7B+ in milestones) and Insilico Medicine (~$600M), and reports 63% of its workforce using generative AI tools.
UnitedHealth GroupUNHParent of UnitedHealthcare and Optum (NYSE: UNH). An operations-scale AI adopter putting nearly $1.5B into AI in 2026 across claims, prior authorization, its Avery member assistant and ambient clinical notes; Optum Insight also sells AI tools. Faces the Lokken lawsuit over its nH Predict care tool.
Veeva SystemsVEEVPleasanton, CA vendor of cloud software and data for life sciences (Vault, Vault CRM, Veeva Data Cloud), not a drugmaker. Adds AI agents built on Anthropic and Amazon models, plus the Falcon agentic-labor line; acquired Ostro (~$100M) and Copli in 2026.34 further companies we profile in Health Care & Social Assistance.