20VC: Lessons from Jensen Huang on "Founder Mode" | How to Know if OpenAI or Anthropic Will Kill your Company | How USV Liking Music Made Them $1BN on an Investment | The Five Year Desert to Product Market Fit & a $5.3BN Valuation with Shiv Rao @ Abridge

What It Covers

Shiv Rao, founder and CEO of Abridge, details building a $5.3B healthcare AI company from a 2018 founding through a five-year wilderness period to dominance in clinical documentation. He covers vertical AI strategy, competing against foundation models, GTM lessons in healthcare's fragmented $5.3T market, in-house model development, and founder decision-making under pressure.

Key Questions Answered

• •Vertical AI Moat Building: Speed to scale in regulated industries creates compounding advantages that foundation models cannot easily replicate. Abridge generates 40% of product outputs from in-house models, built through daily learning from user edits across specialties and languages. This post-training layer — not the base model — represents the majority of differentiated value. Vertical AI companies that reach deeper into the stack control their P&L and own their destiny against frontier model competition.
• •Healthcare GTM Segmentation: The US healthcare market is not one $5.3T opportunity — it is dozens of distinct segments requiring precise targeting. Roughly 800,000 practicing physicians are concentrated inside large integrated delivery networks and academic medical centers like Emory and UCSF. Founders who stay down-market too long miss the window to capture enterprise scale. Timing the move upmarket correctly, as Abridge did post-pandemic when physician burnout hit 40-50%, determines whether a company wins or stagnates.
• •Counter-Positioning Against Incumbents: When competing against entrenched players like Microsoft's Nuance acquisition, identify workflows where the incumbent cannot compete without damaging their existing business. Abridge built around the spoken clinical conversation as a signal layer — not just note generation — enabling expansion into billing, orders, and revenue cycle workflows that Epic and Microsoft cannot pursue without cannibalizing their core products. This structural asymmetry forces large competitors to watch rather than respond.
• •Founder Mode as Tours of Duty: Founder mode does not mean micromanaging every function. Jensen Huang's lesson, delivered to Rao in a midnight cold call, reframes it as falling in love with whatever the current job demands — including living on airplanes five days a week when enterprise sales requires it. Founders should identify the highest-leverage challenge at each company stage and fully commit to that specific tour of duty rather than maintaining uniform involvement across all functions.
• •Thesis Resilience vs. Tactical Flexibility: Surviving a five-to-six-year pre-PMF period requires distinguishing between the core thesis and the execution path. Rao held firm on healthcare conversations as the foundational signal while pivoting repeatedly on product sequencing, go-to-market approach, and business model — including abandoning a direct-to-consumer app that created ethical conflicts around sensitive data monetization. Founders should define the hill they will die on before fundraising, then treat everything else as adjustable variables.
• •Investor Pattern Matching as Signal: Before pitching Union Square Ventures for the $5M seed on a $15M pre-money valuation, Rao spent years studying USV's public behavior — specifically their music commentary on Twitter, which showed consistent pattern abstraction across country, indie hip-hop, and Swedish death metal. This cross-genre taste signaled the cognitive flexibility needed to evaluate healthcare through a technology lens. Founder-investor fit assessment should go beyond track record to evaluate how a partner thinks across unrelated domains.