Nvidia for Hugging Face will be even worse than Microsoft for GitHub and we know now how low the bar is here.
Nvidia has a strong interest in open source AI being a strong contender to proprietary models. They don't want a single AI company to win, as that company would then have enormous leverage over Nvidia, and could perhaps even erode the Cuda moat with their own chip spending.
They want the AI market to be hypercompetitive, with AI companies focusing on competing amongst themselves instead of with Nvidia on one side, and many self-hosters / small inference providers buying Nvidia GPUs for large open models on the other.
It's in Nvidia's business interest to be a good steward of Huggingface, while being a good steward of Github is at best incidental to the interests of Microsoft.
Linus would heavily disagree!
source: https://diginomica.com/kubecon-china-33-and-third-linux-long...
And I'm not so sure that Mac is out of the equation
And Nvidia hardware was running on Linux before AI as well...
> And I'm not so sure that Mac is out of the equation
Mac servers exist but anyone serious about running at scale isn't using them for a number of reasons, not least of which is that Apple simply doesn't make real servers anymore...
That could change.
Who says they don’t switch to hardware manufacturing for AI servers
nVidia driver sets quite low number for VkPhysicalDeviceLimits::maxTexelBufferElements. I don’t think that’s a hardware limit because I have D3D12 backend doing the same thing on the same hardware.
Vulkan performance is not great on nVidia. On all AMD cards I am testing, Vulkan 1.3 is the fastest backend. On nVidia however, D3D12 is faster despite tensor cores (WMMA / wave matrix multiply accumulate / cooperative matrices) are only available through Vulkan, D3D12 is using shader cores exclusively.
They will be as free as the Washington Post after the aquisition by Bezos.
Wait and see is the only thing we can do.
This is just one company. If other hardware is better and they don’t provide access to it, won’t people just leave HF??
Is this somehow more of a monopoly than I am imagining?
The acquisition sets up corporate interests to be able to increasingly drive it.
I hope better for Nvidia and hopefully it can be cleared up, and maintained.
But more importantly for business, if HF fails to use the upcoming ASIC inference chips for their ZeroGPU after the acquisition, their competitors using them might have an advantage.
Why is it alarming for a shovel salesman to also sell plants?
Jack Donaghy: The only thing I will be discussing with the House Subcommittee on Baseball, Quiz Shows, Terrorism, and Media is vertical integration.
Liz Lemon: What's vertical integration?
Jack Donaghy: Imagine that your favorite corn chip manufacturer also owned the number one diarrhea medication.
Liz Lemon: That'd be great, 'cause then they could put a little sample of the medicine in each bag.
Jack Donaghy: Keep thinking.
Liz Lemon: [beat] Except then they might be tempted to make the corn chips give you...
Jack Donaghy: Vertical integration.
That's still vertical integration, you're just looking down the tube from the wrong end.
Suppose a company that sells PCs also makes CPUs. That's obviously vertical integration, right? CPUs are part of the supply chain for PCs. Likewise if they make SSDs.
But CPUs and SSDs are also complementary goods. They're both sold to PC OEMs in order to produce the downstream product, or end customers when they build their own PC or want to upgrade a component or two.
Entering both of the complementary markets is vertical integration because the obvious reason for a company to do it is to integrate the separate goods into a single offering:
> then they could put a little sample of the medicine in each bag.
But then they have a perverse incentive. You get a bag with five servings of chips and one serving of medication -- the first hit is free. And you shouldn't need the drug to begin with.
https://en.wikipedia.org/wiki/Horizontal_integration
Anything else you care to throw at the wall to see if it sticks, just go ahead and assume I disagree with you given how our interaction has gone so far.
Always like that with Microsoft apologists
That page shows 99.5%+ uptime across all 11 of the services provided.
To do that at a small company world cost significant money.
To have it integrated and cheap for all those small shops is very valuable to them.
and again don't know how this is acceptable in 2026, for a paid CI service
and I'm not talking about 6 9s, I mean, 99.5% is terrible
it's 2026, not 2016
and that 40 hr may be that over 40 days you just have to randomly stop working if you depend on GHA for something
and why are other services not struggling like this?
The argument is that AI models trained on open source should include the copyright info for all projects blended into the weights. And since they don’t, it violates the terms of the license.
The copilot service and the training on open source would be the same if Microsoft hadn't bought GitHub, it would just be branded differently.
The load increase from AI is definitely significant, but I think with proper investment they would not be having this many issues.