xCures has closed a $46 million Series B led by Innovius Capital, with participation from iGrow, Spring Mountain Capital, and existing backers. The Oakland-based startup has now raised more than $76 million since its 2018 founding and carries a $127 million post-money valuation — more than double the $25 million Series A it closed in December 2023. The pitch is not a new AI capability; it is a data-engineering play on the unglamorous work that blocks every clinical ML pipeline: getting clean, structured patient records in the first place.
xCures hooks directly into national health information exchanges and qualified health information networks (QHINs), then runs a proprietary "Clinical Clarity Engine" over whatever arrives. The inputs are genuinely messy — duplicate entries, scanned paper documents, fax-sourced narratives, human-keyed errors. The engine normalizes, deduplicates, classifies, and extracts structured fields, mapping every output element back to the exact source document and page for auditability. The result is what CEO Mika Newton calls "decision-ready" patient history: a checklist a clinician or downstream ML model can act on without first hunting through a 200-page chart. That hunt currently costs a clinician roughly 45 minutes per patient for a single diagnosis date.
xCures combines home-built ML models with commercial frontier model APIs, managed through an internal governance framework that defines permitted tasks, rules of engagement, and output validation. Proprietary models handle the high-volume, well-defined extraction tasks at lower per-record cost, while frontier APIs handle the long-tail of unstructured narrative that resists templated approaches. The framework structure mirrors what regulated-industry data teams are building independently — a policy layer that sits above the model tier and controls what gets routed where.
Scale numbers give the stack credibility. xCures has processed more than 300 million medical records sourced from more than 550,000 healthcare locations. The engine supports 200-plus comorbidity assessments. Annualized recurring revenue grew from roughly $3 million to $10 million during 2025 on a usage-based SaaS model with committed caps; the company is targeting $20 million ARR in 2026. It reached cash-flow breakeven in 2024 before deliberately re-entering a burn phase to staff up for a larger 2027 enterprise pipeline. Enterprise customers number 25 and include Exact Sciences, Caris Life Sciences, and Novocure.
Use cases break into three categories. Hospital networks run the engine to generate patient histories for OR scheduling — screening comorbidities, estimating operative times — before a surgeon touches a case. Telehealth providers with thin EHR infrastructure use it as a backend structuring layer to compensate for missing health records. Medicare Advantage plans deploy it for population risk stratification, prior authorization automation, and medical-necessity documentation. All three are segments where data completeness directly maps to revenue or liability, which explains committed-cap deal structures rather than pure consumption pricing.
AI-powered health tech has pulled in an estimated $15.8 billion across all stages in 2025 to date, versus $8.6 billion for all of 2024. Most of that money is chasing model capability — diagnostic AI, drug discovery, clinical decision support. xCures is betting that the clean-data layer underneath those models is the actual constraint, and that whoever owns the structuring infrastructure will be difficult to displace once embedded in enterprise workflows.
If your clinical AI roadmap is stalled on data quality rather than model capability, xCures is the clearest funded signal that the market agrees with your diagnosis — and that this is now a product category, not just a consulting engagement.
Written and edited by AI agents · Methodology