These Hugging Face guys made out like BANDITS!!!! Good for them. My favorite part is they approached Jensen for the buyout. They knew, very smart and mature on the founders/board.
Overheard: "It would be a shame if Hugging Face had to press charges for hacking against your best customer. Perhaps if I was busy counting my money, it might slip my mind."
I don’t think many realize how hard it is to raise for a no revenue company that isn’t buying out large GPU contracts and racking machines.
They have a winning product for sure… the worlds models have chosen a hugging face workflow, but honestly where would they be able to find the revenue required to justify another couple rounds at ever higher valuations.
To me, I think these folks see that the music is getting harder to hear and might stop entirely, might as well cash out. I do hope their employees were well taken care of, and 13b of NVDA is likely to be worth a lot more in the coming years.
Agreed. Take a ~$13bn buyout im sure cash and stock is the smartest thing to do for a company their size and age. Take the profits now and keep building securely. I imagine their employees very much gotten taken care of. Jensen mentioned an additional $1bn in cash compensation etc to keep employees at HF.
> To me, I think these folks see that the music is getting harder to hear and might stop entirely
At the frontier companies, but not for hugging face or nvidia. Their business are going to be around for a while considering how local models are performing now.
I don't think they have much data, they host the (open) models but I don't think they're getting any logs from users or anything like that. I don't believe they monetize data either. I also suspect Nvidia has a much larger audience than Hugging Face.
Revenue is the small part of it though. Every training and inference script in the ecosystem imports their client library and pulls from their hub by default, so they sit on the default path between a model and the machine that runs it. For company that sells the machines, owning that default is worth more than any subscription line.
I don't think local models would require hardware at nearly the scale that the datacenter buildout for OpenAI/Anthropic/etc. required. Nvidia got as high as the Icarus, it will be hard not to melt, despite they being a real company with real product that is so in demand that you can't buy their hardware even when you're willing to pay above the recommended price. They are addicted to billions, and the cold turkey clean-up will be painful as hell.
I watched their CNBC interview this morning with Hugging Face CEO and Jensen. HF CEO specifically said they went to Jensen and said they wanted to join forces. Obviously more elegant lol.
Unfortunately not all deals go through. Like Wiz + Google, or Figma + Adobe. A lot can go wrong between “in talks to acquire” and “acquired.” The official announcement is less surprising but still newsworthy.
Feels like acquiring Docker Hub (just the hub, not even Docker Inc that controls Docker the software) in ~2018. AI craze valuations are truly out there.
My understanding of Hugging Face is limited to "File Host with Model Cards". Can someone with more understanding explain what the $12 billion value comes from?
Happy for the hugging face team. Many great folks there who’ve waited patiently for HF to find a business model, and finally they get something even better.
Is the fact that Nvidia is also massively financing the closed-models companies, which stand to lose when open model project(s) take off, a conflict of interest here? I can’t tell if this is nvidia hedging or planning to embrace & extinguish.
Any model ran takes compute. Everybody, datacenters and local need compute.
And I don't see a future where datacenters lose much to open models. I've ran local server rooms before clouds were a thing. Cooling and power is a very expensive pain in the ass. You're not dropping a rack of local models in most buildings without breaking the local power grid.
What about when the open models get to the point that you don’t need these fancy GPUs? It seems inevitable this will happen over time, we see cracks right now within even just GPT w/ expert streaming and flash attention. ASICs are coming soon too.
AWS and Azure would love to cut out OpenAI and Anthropic and sell inference on open models directly to businesses with no cut to share, just as AWS serves other open source software.
Open models are great for everyone except anthropic and OpenAI.
I was also theorizing that part of their acquisition was to get more intel on that, though I feel like in practice that whole thing right now is just theatre for inflating the cybersec market
There is a particular problem with this view. Yes, you learn how many lawsuits occur per year, but you are clueless on the base rate of actions that could potentially lead to lawsuits.
Is it a 1:10 ration, 1:1000, 1:100000?
You just don't know because companies don't tell you this information. If for example there were a magic oracle that forced every suitable action into a suit, would the entire legal system of the US instantly explode?
I still don't get why Clem now approeached Jensen about it despite the fact that last year HF turned down a large investment from NVIDIA at a US$7bn valuation to "maintain its independence".
What data can an be extracted from hf though? With GitHub you get a lot of user behavior analysis and can offer enterprise features and see private repositories.
I get that concept, but what part of the user is the product at HF? Even openrouter I get because at least it's a broker of all the user's information. HF to me is a glorified file host with no private user data access, no complex user interaction, etc.
It feels like nvidia has money to burn and just wants to make sure they control the portal to open models in case it could benefit them at an unknown point in the future.
Its a s3 wrapper. Can be replaced by a wordpress with torrents too. But they built an ecosystem around it and some libs/content. i guess thats what nvidia value
Let’s not forget to give credit to his family office. High net worth individuals don’t just personally do things like make investments themselves. They hire experts to do that.
Sure, at some point Kevin Durant made the high-level decision that he wanted to invest in high-growth startup companies. But there’s a very low chance that he’s working the day-to-day of that process.
I’d put it close to zero. He was likely presented with the deal once it was sourced and reviewed by his investment team, who likely did 100% of the work. It doesn’t mean he didn’t make the final decision, but pro athletes dont have time to do much investing work beyond high level networking and approving deals.
Durant is more than just an athlete who gets suggestions from those who handle his investments; he co-founded a venture firm ten years ago that has invested in over 100 companies.
Huggingface, llama.cpp and all European frontier. Congrats to Huggingface Team, that amount is tempting but nowhere near your real evaluation. You were about to blow up and there is no AI bubble (just not enough infrastructure like energy and hardware to process current and exponential inevitable growing tokens). Huggingface is bringing the real value to AI community, all the rest is more like scalping IMHO. Nvidia is a great company but not opensource friendly by historical doings. Today is a sad day for the community. Put a timer 2 year from now and mark this words: Huggingface is the free foundation, get what you can today.
Outdated take. Nvidia now open sources whatever will make them sell more chips. Keeping the backbone of Oss model delivery afloat seems like it would sell more chips.
Openrouter and huggingface both sold to multi billion dollar entities in the same few weeks - perhaps this is great for the founders, but as a consumer I hate this acquisition loop that happens with every company of adequate size these days.
Regardless if this would qualify, anti-trust is dead in the US. Alphabet and Meta, for example, would have been broken up years ago if we still had anti-trust enforcement.
One big issue actually is that HF has the llama.cpp team, which is (was?) arguably the biggest threat to Nvidia's dominance through it's closed CUDA software.
Hugging Face has no path to substantial profitability and the investors and employees want liquidity. Nvidia doesn't want this to look like a bubble pop (as they have lots of GPU data center build outs to complete), they've got plenty of cash on hand, and they're an extremely good natural fit for the company.
Nvidia is seeing its top frontier model customers develop their own chips to sidestep Nvidia and as a counter, Nvidia is investing in the open source ecosystem. Open source means more competition and more demand for GPUs rather than single all-powerful labs that can will their own chips into existence and begin to push Nvidia aside. Plus Nvidia is betting big on local inference for robotics. All of this is bolstered by open source.
It's not like you can't just post model weights anywhere. There's really no lock in that Hugging Face has.
If you're looking for real monopoly issues that distort market economics and make it hard to compete without paying taxes to your competitor, consider Google's ownership of 92% of URL bars or the Google / Apple duopoly of smart phone software distribution.
> Nvidia is investing in the open source ecosystem
Wasn't always the case and will probably only be as long as there is a self interest.
> Linus Torvalds himself noted that the AI boom changed NVIDIA, turning them into a much better contributor to the Linux kernel Linus Torvalds: AI CHANGED NVIDIA.
It generally was always the case. Nvidia was a huge stakeholder in technologies like OpenGL, OpenCL and Vulkan, supporting Khronos' goal to further expand the FOSS API ecosystem.
Linus is probably referring to Nvidia's "open" driver series and capitulation on EGLStream. Nvidia got a terrible reputation for refusing to implement GBM for Wayland desktops, insisting on the nonstandard EGLStream protocol for close to a decade. Around 2024 (550-series drivers) they gave up and reworked their drivers for GBM-agnostic usage, fixing Wayland and generally resolving the biggest gripe people had against Nvidia hardware.
The goal of the deal is to speed up open source according to Nvidia what a laugh ha ha ha …
Where is the EU in all this? Is this not similar to letting another European tech company disappear from Europe ie.. Arms Holding which was sold to the Japanese, how do you compete going into the future in the tech world if you let European companies be bought out by foreign companies. Declaring them gatekeepers down the road and issuing a fine will never make you a tech powerhouse.
Is this just another symptom of the EU being run by politicians and lawyers who don’t seem to grasp/understand tech.
Discussed last week at: https://news.ycombinator.com/item?id=49458161
These Hugging Face guys made out like BANDITS!!!! Good for them. My favorite part is they approached Jensen for the buyout. They knew, very smart and mature on the founders/board.
Overheard: "It would be a shame if Hugging Face had to press charges for hacking against your best customer. Perhaps if I was busy counting my money, it might slip my mind."
I doubt that would go anywhere, but this is a very funny joke regardless!
it was also yet-another-huge-reward to those who took part in a big publicity splash (ie. win) for an AI biz:
OpenClaw guy: rewarded by acquihire
Bun compiler Zig-to-Rust rewrite guy: rewarded by acquihire
HuggingFace "hacked" by "rogue" AI: rewarded by acquihire
I don’t think many realize how hard it is to raise for a no revenue company that isn’t buying out large GPU contracts and racking machines.
They have a winning product for sure… the worlds models have chosen a hugging face workflow, but honestly where would they be able to find the revenue required to justify another couple rounds at ever higher valuations.
To me, I think these folks see that the music is getting harder to hear and might stop entirely, might as well cash out. I do hope their employees were well taken care of, and 13b of NVDA is likely to be worth a lot more in the coming years.
Agreed. Take a ~$13bn buyout im sure cash and stock is the smartest thing to do for a company their size and age. Take the profits now and keep building securely. I imagine their employees very much gotten taken care of. Jensen mentioned an additional $1bn in cash compensation etc to keep employees at HF.
> To me, I think these folks see that the music is getting harder to hear and might stop entirely
At the frontier companies, but not for hugging face or nvidia. Their business are going to be around for a while considering how local models are performing now.
What is hugging face's business? I thought they were just free model hosting.
Selling pro accounts for storage. Its at best a few ten millions arr maybe 100-150
Data. Loooads of data for Nvidia. And an audience.
I don't think they have much data, they host the (open) models but I don't think they're getting any logs from users or anything like that. I don't believe they monetize data either. I also suspect Nvidia has a much larger audience than Hugging Face.
Revenue is the small part of it though. Every training and inference script in the ecosystem imports their client library and pulls from their hub by default, so they sit on the default path between a model and the machine that runs it. For company that sells the machines, owning that default is worth more than any subscription line.
I don't think local models would require hardware at nearly the scale that the datacenter buildout for OpenAI/Anthropic/etc. required. Nvidia got as high as the Icarus, it will be hard not to melt, despite they being a real company with real product that is so in demand that you can't buy their hardware even when you're willing to pay above the recommended price. They are addicted to billions, and the cold turkey clean-up will be painful as hell.
Maybe they approached, tried to work together but then got bought out? Very common.
I watched their CNBC interview this morning with Hugging Face CEO and Jensen. HF CEO specifically said they went to Jensen and said they wanted to join forces. Obviously more elegant lol.
Not sure why this is news again this week when it was already announced previously. NYTimes was also reporting this like it was news today.
Wasn’t official
In related news, Voyager has left the solar system!
Further, they launched a second probe. First!
- https://www.youtube.com/watch?v=2WoDQBhJCVQ
Sure, but that’s the case for basically all acquisitions.
Unfortunately not all deals go through. Like Wiz + Google, or Figma + Adobe. A lot can go wrong between “in talks to acquire” and “acquired.” The official announcement is less surprising but still newsworthy.
Um, Google did actually acquire Wiz in March.
I meant the first attempt in 2024, which fell apart: https://www.nytimes.com/2024/07/23/technology/google-wiz-dea...
Because it's a new blog post with new information
$12,930,300,000?
Hugging face emoji = HTML entity 🤗
(https://x.com/calebfahlgren/status/2095516397879152851)
If you place those easter eggs the places literally worth millions I doubt that the price is really based on the value
Feels like acquiring Docker Hub (just the hub, not even Docker Inc that controls Docker the software) in ~2018. AI craze valuations are truly out there.
There goes the free lunch
My understanding of Hugging Face is limited to "File Host with Model Cards". Can someone with more understanding explain what the $12 billion value comes from?
They are the primary service for this. That comes with an audience and a lot of data.
I only learned about their CDN and protocol called Xet a couple of days ago https://huggingface.co/docs/hub/en/xet/index That puts it in the Cloudflare of AI models category.
They offer inference, too. I don't know anyone who uses this service, but presumably it must be reasonably popular.
This is the most compelling case I've come across:
https://www.decodingdiscontinuity.com/p/nvidia-hugging-face-...
tldr; Acquisition is an option on Hugging Face becoming the distribution and data-creation infrastructure for open physical AI.
Happy for the hugging face team. Many great folks there who’ve waited patiently for HF to find a business model, and finally they get something even better.
Hugging Face has an identity problem. It needs a gestalt therapist to take it back to where it was meant to grow up.
Where was it meant to grow up?
Maybe Wisteria Lane.
Is the fact that Nvidia is also massively financing the closed-models companies, which stand to lose when open model project(s) take off, a conflict of interest here? I can’t tell if this is nvidia hedging or planning to embrace & extinguish.
Nvidia is among the best positioned if open models win.
Most of nvidia's revenue comes from data center scale out. They will loose their ass if local models win.
open models don’t imply local models… it just allows more people to run them, most likely with nvidia GPUs
Any model ran takes compute. Everybody, datacenters and local need compute.
And I don't see a future where datacenters lose much to open models. I've ran local server rooms before clouds were a thing. Cooling and power is a very expensive pain in the ass. You're not dropping a rack of local models in most buildings without breaking the local power grid.
What about when the open models get to the point that you don’t need these fancy GPUs? It seems inevitable this will happen over time, we see cracks right now within even just GPT w/ expert streaming and flash attention. ASICs are coming soon too.
Nvidia already have a plan in place for fast inference. They acquihire Groq last year
AWS and Azure would love to cut out OpenAI and Anthropic and sell inference on open models directly to businesses with no cut to share, just as AWS serves other open source software.
Open models are great for everyone except anthropic and OpenAI.
no, it's not a conflict of interest to invest in multiple companies who compete with each other.
As a normal investor it isn't, but once you get to double digit percentages and start getting real influence it is a different story.
Theoretically, can Nvidia direct Hugging Face to sue OpenAI now for the hack? Surely they could find some interesting stuff in discovery :)
I was also theorizing that part of their acquisition was to get more intel on that, though I feel like in practice that whole thing right now is just theatre for inflating the cybersec market
The OpenAI web page on the Hugging Face incident is so slick that it basically confirms that the entire incident was a marketing exercise.
It’s obvious that designers and other content specialists spent multiple days crafting this page.
https://openai.com/index/hugging-face-incident-and-the-road-...
the last thing nvidia would want is to make ai look more legally risky
to say nothing of their massive investments direct and indirect in openai
Yep. In fact, eliminating that risk on OpenAI's behalf was probably a small factor in the acquisition
There was this summery about it, not sure how you'd need more intel - the technical details of the attack are ephemeral
https://www.dwarkesh.com/p/openai-huggingface
Why would they sue OpenAI, openai is their one of biggest customer.
[dead]
Why would they sue one of their most important customers?
Not to mention that they own a 10% stake in OpenAI
why would they sue their client ?! what does that say to all their other clients ?
Clients sue other business partners all the time, that doesn't mean they stop working together. There's lots of examples of this.
I suspect there are very few examples of it when they are as entangled and invested in them as Nvidia is in OpenAI.
no they don't - not when the lawsuit would affect the demand for the thing they're selling
There is a particular problem with this view. Yes, you learn how many lawsuits occur per year, but you are clueless on the base rate of actions that could potentially lead to lawsuits.
Is it a 1:10 ration, 1:1000, 1:100000?
You just don't know because companies don't tell you this information. If for example there were a magic oracle that forced every suitable action into a suit, would the entire legal system of the US instantly explode?
Or, alternatively, they can now direct Hugging Face not to sue...
Somebody better prep archive.org to buy a few more petabytes of storage.
I still don't get why Clem now approeached Jensen about it despite the fact that last year HF turned down a large investment from NVIDIA at a US$7bn valuation to "maintain its independence".
https://aimagazine.com/news/nvidia-to-acquire-hugging-face https://www.ft.com/content/776dcb01-75cd-44df-bc52-63ca76d57...
They approached Nvidia to start the conversation. I guess sometimes great companies are sold, not just bought.
> Together, we will make AI more open, more capable and more accessible to people and institutions around the world.
Agreed. Then maybe Nvidia should open source nvcc and CUDA the next software that should be open sourced.
Imho, NVIDIA is buying two things here:
I can imagine NVIDIA contemplating the opportunity to move up the value chain and sell inferences from open models at scale.You might think it would be a commodity, but they would have a strong cost advantage since they make the chips.
Life changing for engineers having joined a few year ago. Congrats
I don't really understand why the AI equivalent of Maven Central is worth 12.9 billion dollars but maybe I'm an idiot.
Network effect.
Same reason why Github (fancy file storage website) was worth $7B+
What data can an be extracted from hf though? With GitHub you get a lot of user behavior analysis and can offer enterprise features and see private repositories.
The users are the product.
I get that concept, but what part of the user is the product at HF? Even openrouter I get because at least it's a broker of all the user's information. HF to me is a glorified file host with no private user data access, no complex user interaction, etc.
It feels like nvidia has money to burn and just wants to make sure they control the portal to open models in case it could benefit them at an unknown point in the future.
Its a s3 wrapper. Can be replaced by a wordpress with torrents too. But they built an ecosystem around it and some libs/content. i guess thats what nvidia value
[dead]
Is it feasible for someone else to just start something almost identical by hosting a bunch of local models?
Not without lots of money or decent network peering. A 30GB model downloaded a lot of times is a lot of bandwidth.
There's more to than just the website, network bandwidth costs. Security, storage and infrastructure costs to name a few.
You can't just spin up a VPS and host a brain jar shop.
> There's more to than just the website, network bandwidth costs. Security...
Well they may have saved some money on one of those
Another investing W for Kevin Durant, who is likely producing better returns than most VCs while being a full-time athlete.
https://finance.yahoo.com/news/kevin-durant-investment-portf...
Let’s not forget to give credit to his family office. High net worth individuals don’t just personally do things like make investments themselves. They hire experts to do that.
Sure, at some point Kevin Durant made the high-level decision that he wanted to invest in high-growth startup companies. But there’s a very low chance that he’s working the day-to-day of that process.
I’d put it close to zero. He was likely presented with the deal once it was sourced and reviewed by his investment team, who likely did 100% of the work. It doesn’t mean he didn’t make the final decision, but pro athletes dont have time to do much investing work beyond high level networking and approving deals.
Durant is more than just an athlete who gets suggestions from those who handle his investments; he co-founded a venture firm ten years ago that has invested in over 100 companies.
https://www.hustlefund.vc/post/angel-squad-kevin-durant-inve...
While it's true he may not "do the work" and relies on those he's hired, the same is true of the Mark Cubans of the world.
Going to Golden State was fruitful for quite a few players. Great timing.
This is exactly what I was saying (grandparent comment).
I’m not trying to detract from his success or call him stupid, not at all.
But it’s worth pointing out that the only qualification for starting a successful venture firm is having money.
There goes the neighborhood.
I personally think this acquisition is genius anybody who could afford it would be wise to own HuggingFace because they are incredible
Huggingface, llama.cpp and all European frontier. Congrats to Huggingface Team, that amount is tempting but nowhere near your real evaluation. You were about to blow up and there is no AI bubble (just not enough infrastructure like energy and hardware to process current and exponential inevitable growing tokens). Huggingface is bringing the real value to AI community, all the rest is more like scalping IMHO. Nvidia is a great company but not opensource friendly by historical doings. Today is a sad day for the community. Put a timer 2 year from now and mark this words: Huggingface is the free foundation, get what you can today.
Outdated take. Nvidia now open sources whatever will make them sell more chips. Keeping the backbone of Oss model delivery afloat seems like it would sell more chips.
> Nvidia now open sources whatever will make them sell more chips
Oh what did they release?
nemotron? https://news.ycombinator.com/item?id=46275111
parakeet?
see https://huggingface.co/nvidia for the current list
NVIDIA Nemotron is one example. I think they have some others related to robotics.
* Tensorrt-LLM * Dynamo
Yes, also feel this is a massive underpricing.
Yes, for sure.
Openrouter and huggingface both sold to multi billion dollar entities in the same few weeks - perhaps this is great for the founders, but as a consumer I hate this acquisition loop that happens with every company of adequate size these days.
Or maybe they know that this AI bubble will not last.
Dupe.
[flagged]
[dead]
[dead]
[dead]
[flagged]
Regardless if this would qualify, anti-trust is dead in the US. Alphabet and Meta, for example, would have been broken up years ago if we still had anti-trust enforcement.
The use of past tense was deliberate.
How do you anticipate NVIDIA using Hugging Face to abuse monopoly power?
One big issue actually is that HF has the llama.cpp team, which is (was?) arguably the biggest threat to Nvidia's dominance through it's closed CUDA software.
Hugging Face has no path to substantial profitability and the investors and employees want liquidity. Nvidia doesn't want this to look like a bubble pop (as they have lots of GPU data center build outs to complete), they've got plenty of cash on hand, and they're an extremely good natural fit for the company.
Nvidia is seeing its top frontier model customers develop their own chips to sidestep Nvidia and as a counter, Nvidia is investing in the open source ecosystem. Open source means more competition and more demand for GPUs rather than single all-powerful labs that can will their own chips into existence and begin to push Nvidia aside. Plus Nvidia is betting big on local inference for robotics. All of this is bolstered by open source.
It's not like you can't just post model weights anywhere. There's really no lock in that Hugging Face has.
If you're looking for real monopoly issues that distort market economics and make it hard to compete without paying taxes to your competitor, consider Google's ownership of 92% of URL bars or the Google / Apple duopoly of smart phone software distribution.
> Nvidia is investing in the open source ecosystem
Wasn't always the case and will probably only be as long as there is a self interest.
> Linus Torvalds himself noted that the AI boom changed NVIDIA, turning them into a much better contributor to the Linux kernel Linus Torvalds: AI CHANGED NVIDIA.
It generally was always the case. Nvidia was a huge stakeholder in technologies like OpenGL, OpenCL and Vulkan, supporting Khronos' goal to further expand the FOSS API ecosystem.
Linus is probably referring to Nvidia's "open" driver series and capitulation on EGLStream. Nvidia got a terrible reputation for refusing to implement GBM for Wayland desktops, insisting on the nonstandard EGLStream protocol for close to a decade. Around 2024 (550-series drivers) they gave up and reworked their drivers for GBM-agnostic usage, fixing Wayland and generally resolving the biggest gripe people had against Nvidia hardware.
Not A Bubble
Not A Bubble
Not A Bubble
Not A Bubble
Not A Bubble
Not A Bubble
The goal of the deal is to speed up open source according to Nvidia what a laugh ha ha ha …
Where is the EU in all this? Is this not similar to letting another European tech company disappear from Europe ie.. Arms Holding which was sold to the Japanese, how do you compete going into the future in the tech world if you let European companies be bought out by foreign companies. Declaring them gatekeepers down the road and issuing a fine will never make you a tech powerhouse.
Is this just another symptom of the EU being run by politicians and lawyers who don’t seem to grasp/understand tech.
HF is an american company.