Open Source Will Win But Not How You Think: Power, Scale & the New AI Battleground

Description

At TNW Conference 2025, Hugging Face cofounder and Chief Science Officer Thomas Wolf sits down with Christina Criddle for a lively conversation about how open source is shaking up the world of AI. Thomas shares the story of how Hugging Face started out trying to build a modern version of the Tamagotchi, only to pivot into open source almost by accident. What began as a side project quickly turned into one of the most widely used libraries in AI, with researchers and hobbyists alike downloading models to experiment and build new tools. They talk about the huge difference between the massive budgets at companies like OpenAI and Meta and the smaller but very active open source community. Thomas explains why open source models can still compete with the best, even though they cost far less to train. He also talks about how big salaries and fierce competition for AI talent are making it harder for startups to hire, and why he believes some of that will eventually settle down. The session also looks at where founders and developers should focus if they want to build something successful. Thomas shares his view that the real opportunity is in creating applications that make AI easy to use in everyday workflows, rather than trying to train giant models from scratch. Finally, Thomas explains why Hugging Face is now putting so much energy into open source robotics. He describes their vision to make robots more affordable and customizable, so more people can build, improve, and trust them. From cheap robotic arms to a fast-growing community of developers around the world, he believes this could be the next big step in making advanced technology available to everyone. If you’re curious about the future of AI, open source, and how technology can stay accessible beyond the biggest tech companies, this is a conversation you won’t want to miss.