The AI Healthcare Investment Inflection 2026

Per Rock Health and CB Insights, healthcare AI startups raised over $4.1B across 120+ deals in Q2 2026—a quarterly record—with at least six $100M+ rounds; the top three destinations were clinical decision support, ambient documentation, and revenue-cycle automation. July kept pace: on July 15, Daniel Ek's Neko Health closed a $700M Series C led by Lightspeed at a ~$7B valuation (up from $1.7B in early 2025), ahead of its first US clinics in New York; on July 14, molecular-design firm Chai Discovery announced a $400M Series C led by Index Ventures with OpenAI participating, at $3.8B—roughly tripling from its $1.3B December 2025 Series B in seven months. Capital is spreading from documentation and service agents toward preventive diagnostics and drug discovery—the cycle's key marginal change: mega-checks now pay for verifiable clinical and commercial data, not concepts.

OpenEvidence, the clinical evidence engine, closed a $250M Series D in January at $12B (co-led by Thrive and DST) with ~20M monthly clinical consultations. In July, multiple outlets reported the company weighed raising $200M at a $20B valuation—nearly doubling again in seven months—but is reportedly unlikely to proceed on dilution grounds; it also reportedly held acquisition talks with a large tech company (neither confirmed by the company). Reported annualized revenue is ~$300M, roughly double seven months earlier. Abridge, the ambient-documentation leader, closed a $316M Series E extension in April at $5.3B (a16z-led, with NVentures), serving 300+ health systems and 100M+ clinical conversations a year, and is extending into inpatient care and real-time prior authorization via Epic, Highmark, and Availity. What unites the twin stars: verifiable, auditable revenue and deployment data—valuations no longer resting on narrative alone.

Moonshot unveiled the 2.8-trillion-parameter Kimi K3 at WAIC Shanghai on July 16 (MoE, 104B active parameters per token, 1M-token context). It took #1 on the Arena frontend-code blind test with 1,679 points—above Claude Fable 5 (1,631) and GPT-5.6 Sol (1,618)—and delivered its open weights on Hugging Face on July 27 as promised: ~1.4TB across 96 shards, the largest open-weight release ever, with day-0 hosting from Together AI and Modal. Two implications for healthcare: first, hospitals can now run fully private deployments where data never leaves the premises, directly easing compliance pain; second, open weights are not cheap deployment—1.4TB of weights and a recommended 64+ accelerators favor large systems with GPU clusters and private-deployment vendors, while smaller institutions will stay on APIs or industry clouds. With an open flagship closing on the closed frontier, the model layer's scarcity narrative faces mounting repricing pressure.