Healthcare Tech Stack: AI redefines hospitals & outpatient care

For two decades, the healthcare tech stack has been defined by the EHR at the center. Epic and Oracle Cerner captured roughly 60% of the US hospital market. Around this core, hospitals layered point solutions for revenue cycle, patient engagement, and clinical decision support. This architecture worked when software was a back-office function. It breaks down when AI needs real-time access to clinical data and the ability to take autonomous actions within a workflow.

At the clinical workflow layer, AI scribes from Nabla, Abridge, and Suki are replacing manual documentation. Approximately 35% of US health systems had deployed or piloted an AI scribe by mid-2025. At the CDS layer, OpenEvidence has emerged with 8.5 million monthly consultations. At the operations layer, Qventus and LeanTaaS are using AI for patient flow and staffing. At the patient engagement layer, Hippocratic AI and Hyro are automating scheduling and follow-up.

Epic is playing defense-as-offense, expanding AI features across its platform. Oracle Cerner is betting on cloud and Microsoft’s AI. Doximity launched DocsGPT for 80% of US physicians. OpenEvidence is building the AI-native clinical intelligence layer. OpenAI is entering through partnerships. Tempus is approaching from the data side with its $6.1B IPO.

Three models are competing: EHR-embedded (Epic/Oracle), AI-native overlay (OpenEvidence, Abridge), and platform-as-infrastructure (OpenAI, Google, Anthropic). The most likely outcome is a federated model where EHR vendors control structured workflow AI, AI-native startups own unstructured clinical intelligence, and foundation model providers supply the underlying compute.