Fell wuck….
F
I got a pretty cool AI backed discord bot I made. It went rouge(not poorly programmed) and performed over 100 searches in an hour yesterday and got black listed by many mega corporations. I’ll start the bidding at 10 billion.
Rogue*
I doubt it started using makeup.
For another billion I can get a robot arm, print out a face, and let it go ham with a can of spray paint.
The beginning of the end of free access open models. Their end goals, all AI companies mixed, is to regulate AI so hard that we cant self host no more, so that we can sell our souls forever to the AI overlords which is planned to replace us.
Once chaos starts, they will retreat in their self sufficient bunkers, let AI clankers and drones to clean up chaotic citizens, until population has shrinked enough for the pedophilic elite to start living their dream life in the world they will have stolen from us.
Nah Nvidia’s all for local models - people need GPUs to run them! OpenAI is trying to develop their own chips, so it isn’t in Nvidia’s interest for OpenAI to win. Probably ditto for Anthropic. Nvidia wins in a world where there is no AI provider big enough they can invest enough to bypass Nvidia. This is probably why Nvidia has been releasing their own open models, training sets, etc.
Nvidia doesn’t consider consumer markets to be worth paying attention to anymore.
For now.
Nvidia has two massive problems.
- They know that AI is a bubble. If the big AI companies go belly up, Nvidia loses all their revenue.
- The big AI companies are working on DIYing their own AI chips. If they manage to, Nvidia loses all their revenue.
That’s why they have been at work with local AI for a while. They publish their own open-weights models (Nemotron) and they build comparatively cheap local AI hardware (DGX Spark, RTX Spark).
That way, when the whole AI bubble pops, they still have quite a bit of a foot in the door with customers.
Yeah I think the growth is going to be in these little models that think big. Like the Qwen3.8-flash idea. Using little tricks to make a small model perform better than it should. Localized AI is going to scale much better than being dependent on giant resource hogging DCs. And god forbid someone figure out how to chain these things together in some sort of peer to peer network utilizing all that processor time that goes unused throughout the day, like some sort of AI@home setup. Pretty sure that’s how Skynet starts.
While true, this same infrastructure is used by corporations that either feel they need the privacy of on premise model, or decided paying by the token is too expensive/unpredictable.
NVidia would love every business to decide they need to have dedicated private gpu heavy infrastructure.
Yeah they are definitely pulling a VMWare. Only big customers matter.
Nah I bet they get buzz killed by some loser in his mom’s basement.
13 billions for something that if disappeared overnight, it would be just a mildly inconvenience and replaced by a 1:1 competitor in less than a day?
Mild inconvenience? HF hosts millions of models and datasets, that would be a huge loss and break all software that depends on it
Competitors (like ModelScope) already exist but are rarely used outside of China
GH is down every second day, yet we’re still alive
Didnt hugging face take its inspiration from another website that hosted scientific big models to begin with? I forget.
Most people use hugging face as a glorified model download storage. I can see a competitor pop up almost instantly.
Where the “value” is in the name recognition and ability to take models they dont like down.
How the hell can it be worth billions?? Isn’t it a fancy repo for models? What’s the actual value?? I swear billion is the new million for tech buyout
Microsoft spent $7.5 billion for Github, and managed to find use for the repo.
I can see some sort of subscription service where Nvidia allows you to run on model on their hardware. And also something has to boost their valuation to drive the stock up, and releasing new cards right now is a waste with RAM prices going ever upwards.
HuggingFace already has something for that called HuggingFace Spaces, though I don’t know how it is monetised. Even runs on different tiers of Nvidia GPUs already
Assumedly will be able to beat everyone on price
This has been true since Facebook bought Instagram for 1 billion.
They also sell access to use the models on hardware.
Dear Mr. Huang,
I have a proposal that will make your company a lot of money. In fact, I guarantee this deal will will increase NVIDIA’s revenue by 19 billion USD.
All you have to do is invest 20 billion in my company, Al Is Awesome, LLC, and I will purchase 19 billion worth of NVIDIA cards to get my project off the ground.
Thank you for your consideration.
Sincerely,
Schipelborp
CEO, Founder, Al Is Awesome, LLCI would never invest in someone who can’t spell their own name. Lucky for you, ai bros don’t give a shit about that.
I hope they buy it. it will only speed up the inevitable destruction of LLMs/AI when they lock selfhosting access behind subscriptions.
please, by all means, shoot yourselves in the foot.
Here’s the thing, AI isn’t going away. Even when the bubble bursts, which it will, it’s a useful tool that can do useful things. I also don’t think in its current form it’s as useful or “necessary“ as the major tech bros want us to believe. I’d much prefer smaller models running on my hardware than just being controlled by a handful of those tech bros. But NVidia buying one of the major players in the local AI scene worries the shit out of me.
But AI is just a tool. It’s not good or evil. I can use a hammer to build a house or bash someone’s skull in. It’s not the tools fault either way.
it’s funny. all your talking points are the exact same as crypto and NFTs. how’d those work out again?
oh? a grift you say? a vehicle to siphon wealth away from the masses you say? building a trust that only fascist régimes and criminals use?
say it isn’t so!
Sounds like there’s a market for a hugging face comeptitor.
Nvidia is seeing the writing on the wall that local models are the only viable future for AI. They have to try and squash that now before it’s too late. I mean, it is already is too late, as they’ve hooked there wagons to these AI companies, and that shit’s going to end and nividia is going to be a bag holder. I can’t wait.
We need to get models into a distributed distribution. Huggingface is way too central.
Yeah, would love to see more of a torrent type website that simply hosts torrent files to models.
some sort of arr stack for models tbh
It feels like one of those technologies that can’t truly be contained. It’s like how the government keeps trying to end piracy. It’s like a
hydrationhydra. Cut off one head and two more show up. Local models will be the dagger to premiere models.drink water!
Damn hydration. Always growing too many heads.
Hail hydration
yeah fully agree. They may go more underground, thinking if govs try to block them, but they’ll still be there.
There will be a new generation that learns about VPNs and torrents. Life is cyclical.
Nvidia is seeing the writing on the wall that local models are the only viable future for AI
Not necessarily. HuggingFace already lets you run models in their servers. Nvidea might be buying HF so that they can sell cloud-compute to even more people.
or to lock the models behind a paywall… look what’s happened with 3d printing, so many sites now charge for STL files.
From what I’ve seen, the sitse that charge for STLs are priced based on the sculptor, unless you mean something else
Some do, others just charge for everything or charge for something that’s released for free elsewhere… Cults3d and Yeggi are bad for that from recent memory.
NVidia wins if people run local models on their hardware too. Hugging face is not a competitor, it is an enabler.
What about graphics card not made by Nvidia? I run local models with AMD chipset and GPU. Works great.
I doubt I’m the only one.
Maybe, but CPUs are coming out with LLM tuned chips, and I can run a basic model on an i5, 8gb ram, and no dedicated card. It’s not super powerful, but for most users, it’s more than enough for what they use the big models for. Also, if it does take a discrete card to get that needed boost in performance, then at least consumers would be able to get GPUs again.
I think as hardware improve and is further designed around LLM efficiency, and local models are tuned for specific uses and being able to run on lesser hardware, it will make Nvidia obsolete for large swaths of the population. A good GPU will still be necessary for high performance, graphic/physics intense gaming, but that’s a really small subset of all users.
Hopefully Nvidia just shits and has to grovel back to the consumer to get there marketshare back when all of the DCs go tits up.
I can run models on my 3yo midrange smartphone. Gemma-4-E2B totally runs on there. Bonsai-8B too. But neither is really good for most tasks.
Heck, “a basic model” even runs locally on the 8MB RAM of an ESP32-S3, but it’s utterly worthless at anything.
If you get a bit more into self-hosting AI it quickly becomes obvious that for any actually useful real world tasks you need at least 24GB you can dedicate to the LLM alone, and if this is fully GPU vRAM, the performance is way, way higher than on CPU, even with an NPU.
any actually useful real world tasks you need at least 24GB you can dedicate to the LLM alone, and if this is fully GPU vRAM
Agreed. For anyone who doesn’t have a 24GB card, the AMD 7900 XTX is basically the only sanely priced 24GB GPU left in 2026 and can be had for under a grand new, whereas even a used 3090 is $1200 and $2000 new.
Qwen 3.8 27B runs nicely on my XTX. I have to use Claude at work, and Qwen 3.8 in llama.cpp with the OpenCode desktop app leaves nothing to be desired in comparison. 3.8 27B is about on par with Sonnet 5 according to benchmarks, though I even prefer it over Opus because Qwen writes in clear, understandable language. We are at the point now where anyone with a 24GB GPU has little reason to use data center LLMs.
Bitcoin went through a similar trajectory. Graphic cards to specualty chips.
except the chips are bought out by AI companies years ahead of time. it will take a long time for nvidia to switch back.
Doesn’t matter to Nvidia. It’s just about having a plan B if all those future orders start to collapse.
Nvidia releases open models too
What is logistics involved if models r just available on torrents, instead of central repo. Maybe validity of model is not possible?
I think you could just store the signatures centrally for validity, no need to store the whole models.
So if models r released for free / open weights. This should be achievable, is that correct?
Is it simply: someone need to implement it?
Kind of. There’s also the ‘someone needs to find and start using it’ problem, which is doubly true for torrent sharing
They going to buy it with an IOU?
Nvidia is the only dog in this race that’s actually turning profits hand-over-fist
Only if the line keeps going up. Ai companies have been “selling” chunks of themselves to NVIDIA for “cash” then these companies turn around and buy hardware from NVIDIA. When the crash eventually comes, NVIDIA will have some percentage of zero (or close enough to count) value company. Basically acting like a bank, just cutting the banks out of the process. NVIDIA shows revenue and the AI company churns their super-inflated value to be even more inflation.
nah, they’re basically the best shovel sellers around. practically all their deals take payment up front for the hardware (+lease obligation) on top of some form of constant monetization for the compute used a/o various types of financing schemes (that involve getting outside parties to foot the bill).
jensen knows this game very well and is playing everyone else superbly.
it’s obviously a bubble and they will correct along with everything else eventually but compared to the rest NVDA will likely come out far far ahead of everyone else because they bring in actual revenue.
OK, sure, theyre selling shovels but theyre lending a bajillion dollars in shovels for equity in unprofitable companies and when the bubble bursts they’ll have no cash and pieces of paper in defunct companies. Unable to meet their operating costs they’ll die or be pieced off. You can’t run a company off of IOUs.
And their local AI push (Nemotron, DGI Spark, RTX Spark, Huggingface, …) is all done to diversify before the crash.
Local AI is constantly getting better, both on server-grade hardware and on hardware regular users can actually own.
In fact, for Nvidia it’s better to have a solid standing with consumer-grade local AI. If Anthropic, Google and OpenAI manage to DIY their own AI chips, then Nvidia will lose a large portion of their revenue within a day.
Most of that money is their own money though, just being paid back to them by the companies they “invested” in. This makes the line go up and adds to Jensen Huangs personal fortune, but it is not actually a profit for the company.
Their main manufacturing business is now making components for datacenters, and those are also starting to lose ground.
Jensen Huang is personally adding to his fortune though, no doubt about that. But so are all the other tech oligarchs, even if their companies are huge bubbles on the verge of explosion.
nvidia found suckers for thier chips.
Is that like an NFT?
RIP.
I had a battery coin charged up to 5.3e288 in Raccoin today, so I know how it feels, Jensen.
Damn. That’s like bad…
Fuck off
