Billion-Dollar Unpopular Startup Ideas

· startups · Source ↗

Summary based on the YouTube transcript and episode description.

YC partners Garry Tan, Harj Taggar, Jared Friedman, and Diana Hu argue that AI verticals are now crowded and founders must make genuinely contrarian bets to win.

  • Flock Safety now valued at $7.5B and solves 10% of all reported crime in the US; original estimated TAM was ~$50M, making it nearly unfundable.
  • AI vertical competition has returned to pre-boom levels; no major model step-change since o1 preview (~1 year ago) means green-field is gone.
  • Brian Armstrong’s contrarian bet at Coinbase was working WITH regulators and banks when the entire crypto community viewed that as a betrayal.
  • Lyft founders launched knowing the service was likely illegal and feared going to jail; Ridejoy passed on the same idea over legal risk.
  • The forward-deployed engineer model is now the consensus enterprise AI playbook; GigaML is flipping it by replacing human FDEs with AI codegen that completes migrations in minutes.
  • OpenAI launched to mostly negative press; the academic AI establishment said scaling-law spending on GPUs without peer-reviewed papers was worthless.
  • Door Dash’s contrarian bet was refusing to be a full-stack startup (no ghost kitchens) when full-stack was the dominant 2014 startup meme.

2025-10-17 · Watch on YouTube

Related coverage