Abridge raises at $5.3B valuation as AI for doctors doubles in value within months
Jun 2, 2025
Key Points
- Abridge, an AI platform that transcribes and summarizes patient conversations, raises at $5.3 billion valuation, doubling its worth in months as Andreessen Horowitz and Coastal Ventures lead the round.
- The healthcare AI startup spent years in obscurity until ChatGPT's arrival shifted investor sentiment on generative AI efficacy in medicine.
- Venture capital is racing to back vertical AI applications that solve domain-specific problems like compliance and workflow integration that general-purpose models don't address.
Summary
Abridge, an AI platform that transcribes and summarizes patient conversations for doctors, raised funding at a $5.3 billion valuation. Andreessen Horowitz and Coastal Ventures led the round, which doubles the company's valuation from $2.75 billion earlier in 2025.
Abridge was founded in 2018 by Shivra, a cardiologist, to address a core problem: physicians spending time on illegible handwritten notes and cumbersome documentation instead of patient care. The product lets doctors walk into a room, have a conversation while making eye contact, and have Abridge handle transcription and summarization.
The company spent six or seven years in relative obscurity, facing skepticism from healthcare peers about AI efficacy. ChatGPT's arrival changed the trajectory. Shivra said the company's "heartbeat was getting more and more faint" before generative AI shifted sentiment. Abridge has now raised over $400 million total. Earlier backers include IVP, Elad Gil, Spark Capital, Bessemer, and Union Square Ventures.
The investment reflects a broader trend. Venture capital is racing to back vertical AI applications—tools that make large language models useful for specific professions like medicine, law, and sales. Unlike ChatGPT, which achieved 80+ percent penetration among consumer users, vertical applications address domain-specific constraints around data handling, compliance, and workflow integration that general-purpose models do not solve.