Hacker News Reader

Comment by SillyUsername | Hacker News Reader

This isn't a good thing for open source.

Nvidia don't share CUDA, don't open source their drivers, don't support capable but older hardware (forget Pascal etc), and generally charge a premium over competitors for hardware.

This acquisition will likely solidify this general stance, reduce free compute allowance for a subscription, cap downloads, push advertising, generally favour models that are Nvidia prescribed or have been sponsored, and potentially ban models and datasets that are deemed risky legally (abliterated etc).

I see no other reason why Nvidia would want this kind of vertical.

View on HN

To play devil's advocate: Hugging Face was probably not profitable, and someone has to pay the bills long-term. Nvidia benefits from Hugging Face and the operating costs are pennies to them. For them it makes sense to acquire HF just to keep it running. HF shutting down or self-destructing in the search of profits would be costlier for Nvidia than just buying it now

That said, I share all your concerns. The days of a permissive hands-off HF may be numbered

Why do you think it ‘probably was not profitable’ it has a dominant position and there are many ways to monetise that.

They had incredible revenue growth the last few years and just broke 100M in revenue. I don't know what their internal spend was , but that was almost half of their recent round in ARR. Mostly likely they were profitable or on a clear trajectory to revenue growth. Huggingface hosts a lot of data and models, but mostly static cold storage is pretty cheap tbh.

Their revenue mostly comes from enterprise deals which are very profitable, and they are in a hot industry with loads of enterprise customers paying them. Storage is also extremely profitable with very fat margins at a large scale like what HF is operating at.

While nvidia may not want open source drivers, they want open weight models and this acquisition allows them to keep pushing for that.

Commoditize your complement.

Nvidia don’t fundamentally care about open weight models

They care about their high margin GPUs being the dominant platform, otherwise they would have to reduce the price of their gpu/increase vram amounts.

Nvidia are making truck loads of money and want it to continue

They actually do when all their proprietary model customers are also building their own chips.

CUDA is partially open and partially closed.

Closed Core: Driver, compiler (nvcc), runtime, assembler (PTX/SASS). Closed Libraries: cuDNN, cuBLAS, OptiX.

Open Libraries: CCCL (Thrust, CUB, libcudacxx), CV-CUDA (Vision), CUDA-Q (Quantum). Open Tooling: CUDA Python bindings, NVIDIA Kernel Module (Linux driver outer layer).

Nvidia is working with Red Hat to develop the NOVA driver and with Valve to develop NVK, which will be the future open-source drivers.

Nvidia supported the Pascal architecture for 10 years. No other graphics company did that. AMD recently made the RDNA 2 architecture legacy after only 5 years.

Nvidia is that most promotes open models and contributes significantly through Nemotron and various hardware-level optimizers. I think the value has more to do with the potential dominance of the open model ecosystem.

https://x.com/ctnzr/status/2072715518373777812

+1 on @SillyUsername. Combined with recent acquisition of OpenRouter we see exactly what we were increasingly concerned about: tightening (chocking?) of Open Weights locally-run models. The sequence and timing of these events make the intent very clear.

Now the Open-Models crowd will try to move to some other place. But fragmentation will weaken the position. Yes, there is Civic, there are purely Chinese websites ... for those who speaks Mandarin. Which only reinforces the point.

Nvidia does frequently release new OSS Nemotron models where they implement new mechanisms.

Dummytron doesn't count.

NVIDIA is open source software first for almost everything. You cherry pick a couple of things, but it's not the general sentiment.

Reply on news.ycombinator.com