Half the Fortune 500 Now Builds on Open AI Models. Here Is What That Means for You
Hugging Face CEO Clem Delangue says open-source AI is growing fast, and the companies using it are not just the tech giants you might expect.
Key points
- Hugging Face, a platform where developers share free AI models and data, now serves roughly half of the Fortune 500 companies.
- CEO Clem Delangue says open-source AI, meaning AI whose underlying code and blueprints are published freely for anyone to use, is more important than at any previous point.
- The company has grown into the rough equivalent of GitHub (the popular code-sharing site) but specifically for AI tools and datasets.
- Businesses typically start with a big commercial AI provider, then migrate parts of their work to open models once they understand their needs.
Hugging Face started as a quirky chatbot app. Today it is the place where AI builders go to share, borrow, and download the raw ingredients of artificial intelligence: pre-trained models (AI systems that have already been taught on large amounts of data and are ready to be customised), labelled datasets (organised collections of examples the AI learns from), and the tools to stitch them together.
About half of the Fortune 500, the list of America's 500 biggest companies by revenue, now use something from Hugging Face. That is a striking number for a platform that most consumers have never heard of.
CEO Clem Delangue, speaking to TechCrunch AI, described a pattern he sees constantly. A company signs up with one of the large commercial AI providers, pays for access to a closed model, then gradually figures out exactly what it needs. At that point, many switch some or all of their work to open-source models, meaning free AI systems whose code and training details are public. The cost savings are real. So is the control.
Open-source does not mean amateur. Some of the best-performing models available today are fully open. Meta's Llama series, Google's Gemma family, and Mistral's models all sit on Hugging Face and are free to download and run.
Does this affect ordinary workers and customers?
Yes, more than most people realise. When a mid-sized insurer or a retail chain builds an internal AI tool using a free open model instead of paying for a big commercial one, that decision shapes which jobs change, which tasks get automated, and how much the company invests in AI elsewhere. Cheaper AI infrastructure means more companies can afford to experiment, not just the ones with enormous technology budgets.
One honest caveat worth stating plainly: Delangue runs a platform that benefits directly when open-source AI looks attractive. The story he tells, where open always wins eventually, is also the story that is best for his business. Many companies do stay with closed commercial models and do just fine.
Still, the underlying trend is real. More usable, capable, free AI tools exist today than at any point before.
One doable takeaway: If you run a small business or manage a team, spend one hour on the Hugging Face website browsing what free models exist in your field. You do not need to build anything yet. Knowing what is available, and roughly what it costs to run, is the kind of information that will matter in the next budget conversation you have about AI.
