Hugging Face has retained a bank to gauge acquisition interest at a valuation of $13-billion or more, roughly R211-billion, Business Insider reported on Sunday. Talks are early, no bidder has been named, and no deal has been reached.

The platform is where developers and researchers share, find, test and deploy AI models, and that price would nearly triple the $4.5-billion, about R73-billion, it was worth after its 2023 Series D.

The scale explains the number. Hugging Face hosts more than three million models and a million datasets for around 13 million users, and its library is pulled roughly 13.7 million times a week. It added about 1.18 million models in 2025 alone, more than every previous year combined, and has grown beyond hosting into enterprise subscriptions, dedicated compute and private model deployment. The awkward part is what a buyer would be acquiring.

Earlier this year, Hugging Face turned down $ 500 million from Nvidia at a $7-billion valuation, saying it did not want a single dominant investor swaying its decisions. Chief executive Clem Delangue has said the company is close to profitability, only recently began spending what it raised in 2023, and is optimising for long-term sustainability rather than short-term profit.

The interest follows Stripe's agreement to buy model router OpenRouter, reported at around $ 7 billion.

TechCrunch questions whether Hugging Face is genuinely considering a sale or simply fielding offers, given how consistently Delangue frames the company's responsibility to the community that trusts it with their models and data. TechTimes calls it a neutrality paradox, arguing the value any acquirer pays for exists only for as long as they leave the platform alone. BigGo tracked the platform metrics behind the valuation, including the growth in hosted Spaces this year.

This sits closer to home than it looks. If you are building AI here, the economics rarely work on American API pricing against rand revenue. The alternative is open-weight models pulled from Hugging Face and run on your own or rented hardware, which is the path most local AI startups take.

That path exists because one neutral platform distributes almost everything, free, in one place. A hyperscaler or model lab owning that layer would not need to close it to change it, since rate limits, tiering and preferential placement do the same work quietly.

Nothing has happened yet, and Delangue may well not sell. But if you have a production dependency on Hugging Face, this is a good week to work out what your fallback looks like.

You might also like our piece on the Cognition valuation and what buyers pay for AI infrastructure, Anvaya's seed funding from 3CV, and why SA enterprise AI adoption keeps outrunning the strategy behind it.

Get more SA tech and business news and subscribe to The Open Letter.