Lessons from Building Open Source Libraries

By Y Combinator

Categories: VC, Startup

Summary

Hugging Face co-founder Thomas Wolf shares surprising insights on the power of open-source AI: open models unlock 100x more possibilities, while 'open science' can catalyze a win-win ecosystem instead of a race to the bottom.

Key Takeaways

  1. Open source enables rapid exploration and creativity by providing powerful pre-trained models as a starting point, unlocking 100 million GPU hours of training.
  2. Leverage open models to quickly build interactive prototypes and explore alternative use cases the original creators didn't envision, rather than being limited to their intended uses.
  3. Open science and a 'win-win' ecosystem mindset is more powerful than a closed, competitive race to the bottom when building AI companies and products.
  4. Transitioning between fields like physics, law, and entrepreneurship can provide diverse perspectives and skillsets that shape impactful open-source projects.
  5. Founders should be cautious of demos that look good but don't translate into truly valuable user experiences, even when built on top of open-source models.
  6. Treat your time as a valuable resource, in contrast to the 'falling down rabbit holes' mindset of pure research, when transitioning from academia to entrepreneurship.

Topics

Transcript Excerpt

I'm excited to welcome today Thomas Wolf, the co-founder and chief science officer of Hugging Face. We're here today at beautiful San Diego for the Europe's 2025 conference. So, let's get started. Thomas, you had a very unusual career path before you became the founder of one of the best open source AI companies. You studied originally physics, then even did law and then started uh hugging face. Tell us about how that journey shaped eventually hugging face. >> Yeah, I mean even in physics. So I ...