Nvidia acquires Hugging Face in $12.93 billion open AI bet

Nvidia buying Hugging Face is the closest thing this cycle has to Microsoft buying GitHub for the AI era. Hugging Face is the default hub for open-weight models, datasets and evaluation tools; Nvidia is the default provider of compute. Unifying the two gives Nvidia much deeper visibility into which models developers are actually using, how workloads behave and where demand is headed, at exactly the moment open models from DeepSeek, Zhipu and others are starting to rival closed labs on cost and quality.

For the race to AGI, this is less about another big check and more about control of the ecosystem’s “town square.” If Nvidia can keep Hugging Face credibly neutral while weaving it tightly into its toolchain, it strengthens the hardware moat and makes it harder for rival accelerators to gain share. On the other hand, concentration risk goes up: the leading chip vendor will now own the dominant open-model platform. Regulators in the US and EU will study whether this gives Nvidia too much leverage over open AI infrastructure, especially as more national AI strategies lean on open-weight models for sovereignty. In practice, open-source and open-weight communities now have to decide whether “open, but owned by Nvidia” is good enough or whether alternative hubs need to grow.