ACQ2: Building the Open Source AI Revolution (with Hugging Face CEO, Clem Delangue)

· ai · Source ↗

Summary based on the YouTube transcript and episode description.

Hugging Face CEO Clem Delangue argues thousands of companies will build their own specialized AI models rather than relying on a handful of API providers.

  • Hugging Face has 5M+ AI builders, 3M+ shared models/datasets/apps, and a new model/dataset/app built every 10 seconds.
  • Company raised $500M+, spent less than half, and is profitable — unusual among AI infrastructure firms.
  • Delangue’s contrarian thesis: all tech companies will eventually train their own specialized models, not consume APIs from 5-8 giants.
  • Hugging Face valued at $4.5B with strategic investors including Nvidia, Google, Amazon, Salesforce, Intel, AMD, IBM, Qualcomm.
  • Company pivoted in 2019 from a teen chatbot to AI infrastructure after Thomas Wolf ported Google’s BERT to PyTorch over a weekend — the tweet got 1,000 likes and revealed developer demand.
  • Delangue frames AI as ‘software 2.0,’ not AGI progress — skeptical of existential-risk arguments as a pretext for closing off open-source AI.
  • AI startups need scientist co-founders and must discard Lean Startup thinking — training cycles are months, not days, and step-function improvements matter more than incremental 5% gains.

2025-03-20 · Watch on YouTube

Related coverage