AI Healthcare

Healthcare AI has moved past hype into routine deployment. Google's DeepMind Health, IBM Watson Health, and startups like Tempus and Recursion Pharmaceuticals are processing real patient data and generating real clinical outcomes. The FDA has approved over 500 AI/ML medical devices, validating the category. The winners cluster into three buckets: diagnostic imaging AI (most mature, FDA-approved, proven ROI), drug discovery acceleration (highest capital requirements, longest payoff, but unicorn potential), and clinical operations (automating intake, coding, billing—immediate ROI). Tempus and Recursion have each raised over $500M by positioning as "AI-first pharma platforms," while imaging startups like Zebra Medical Vision ($300M+) nail reimbursement puzzles. But healthcare AI faces three non-trivial challenges: data fragmentation (patient records scattered across EHRs, labs, and imaging), regulatory slowness (FDA approval timelines), and clinical adoption friction (doctors skeptical of AI). The defensible winners combine strong clinical data networks, regulatory expertise, and distribution through health systems or pharma partnerships. Pure "AI for healthcare" plays rarely succeed without solving the distribution problem.