NVIDIA paid $12.93bn for Hugging Face, not the ten billion being repeated

NVIDIA agreed to buy Hugging Face for $12,930,300,000. That is not a rounding of ours. It is the figure in the first sentence of Jensen Huang's own announcement, posted on the NVIDIA blog on September 3.

On September 9, Nathan Lambert — who worked at Hugging Face, and whose read on the deal is the most informed one published so far — wrote that the platform is "worth easily the ~$10B they paid, but every year."

Nathan Lambert@natolambert

I never got to comment on Nvidia-HF because I was OOO, but something that struck me is how HuggingFace is a steal for Nvidia. HuggingFace's core ability is understanding how to influence the discussion and direction of AI, relative to their resources.

For Nvidia, who is so rich and powerful, this ability is worth easily the ~$10B they paid, but every year. There are many sources of capital other than just hard cash -- HuggingFace is rich in those sources of soft power. It'll be a challenge for Nvidia to keep that culture and to foster it, but that challenge will be valuable growth for the company.

Nvidia is a better buyer than one of the three clouds, as the clouds would feel more pressure to make money on it, to compete directly with github, etc. HuggingFace should now be given the directive of Jensens ambition, to be free from having to think of themselves as a "functioning business unit" and to be the most powerful comm's army in the AI discourse of all time.

HuggingFace should be free to win the hearts and minds of the next 100 million AI developers, and I'm excited to see it happen. Having worked at HuggingFace, and stared down the barrel of making the company work as an independent public entity, this outcome should've been seen as somewhat inevitable.

Congrats again to @Thom_Wolf @julien_c, @ClementDelangue, and all my friends at @huggingface for making something with such incredible value. Much of the culture was about creating value for the whole community, rather than capturing it for HuggingFace. This is now in line with Nvidia's "all-in" bet on open models (so intelligence isn't monopolized), and this next part of the chapter will be fun.

My first day of work at HuggingFace, fresh off a red eye in Paris, still feels like yesterday (2022). When I posted the prediction below, last year, I actually got hate for it. lol.

The argument survives the correction. The number does not. Twelve point nine three billion is 29 percent above ten, and if you are sizing what NVIDIA thinks open-model distribution is worth, that is not a rounding error.

Why does anyone have ten in their head? Probably because that is what a deal of this shape sounds like, and because nobody re-reads the first line of a blog post they have already summarised. We would gently suggest re-reading it. The price is exact to the hundred dollars, which is what a share-exchange calculation produces and what a negotiated headline number never does. Somebody divided a fixed pool of NVIDIA stock by a strike price and printed the result. (The byline on the post, as the page renders it, is 黄仁勋.)

So what does $12.93bn buy?

Eighteen million developers, three million models, five hundred thousand datasets, a million applications and two hundred thousand companies. Divide the price by the developers and NVIDIA paid roughly $718 a head (our arithmetic, not theirs). For comparison, that is in the range enterprise software buyers pay for a seat that generates revenue, and Hugging Face developers mostly do not.

Which is Lambert's point, and it is the right one. Hugging Face's asset is influence over where the open-model conversation goes, and influence has no line on a P&L. He argues NVIDIA is a better owner than any of the three big clouds precisely because it feels no pressure to make the thing profitable, and can instead point it at winning "the hearts and minds of the next 100 million AI developers."

Our read is that the neutrality clause is the whole asset and the whole risk. Huang commits in the post that developers will choose their own models, frameworks, clouds and inference providers, and that "NVIDIA compute will not be required to build on or deploy through Hugging Face." A registry everyone trusts is worth twelve billion. A registry that quietly routes to one vendor's silicon is worth a fraction of that, because the second-best registry becomes free.

So watch the deploy button on a model page, not the press release. If, a year from now, the default inference path on huggingface.co has a green tint, the thesis Lambert wrote and the promise Huang made will have come apart, and the number to remember is the one NVIDIA printed itself.