Everyone is making better software than NVIDIA today with AI, to the extent they are un-nerfing hardware and making stuff run 2x faster. Why would anyone in their sane mind pay NVIDIA anything when they can vibe engineer a faster library?
... they can force them and they will - I think it is telling that they are beginning to consider extracting software rents, which could indicate they are beginning to prepare for some lack of sustainability in hardware sales - though right now they cannot make them fast enough, 'em chips.-
> Couldn't customer #1 just have AI reverse engineer, clean room recode, and open source publish it, because its software ?
Probably.
But... Nvidia is currently close enough to a monopoly in this space, and our antitrust enforcement is so weak/nearly-non-existent, that I suspect that anyone who either did this or used the software that resulted from it would be blacklisted by Nvidia and never get another Nvidia GPU.
So (unfortunately) chances are very good anyone in a position to do this will just shut up and pay the rent.
Lmao! Sorry AI companies, even if you somehow survive to profitability, daddy Nvidia will just decide you have to pay rent in perpetuity. 99% margin on chips isn’t enough.
One thing I've noticed is a lot more HF assets require personal information collection to download. Not sure if that's new corporate compliance or an attempt to manage bandwidth (or a marketing channel).
Everyone is making better software than NVIDIA today with AI, to the extent they are un-nerfing hardware and making stuff run 2x faster. Why would anyone in their sane mind pay NVIDIA anything when they can vibe engineer a faster library?
because NVIDIA can force them
... they can force them and they will - I think it is telling that they are beginning to consider extracting software rents, which could indicate they are beginning to prepare for some lack of sustainability in hardware sales - though right now they cannot make them fast enough, 'em chips.-
What software is this? A value-add nobody uses, or a core CUDA component everyone is stuck with?
At what point do market forces take over, and give Nvidia the collective middle finger ?
Couldn't customer #1 just have AI reverse engineer, clean room recode, and open source publish it, because its software ?
It's all in how your attorneys define "clean room" now isn't it ?
> At what point do market forces take over, and give Nvidia the collective middle finger ?
I haven't bought an Nvidia gpu since the 980. /shrug
> Couldn't customer #1 just have AI reverse engineer, clean room recode, and open source publish it, because its software ?
Probably.
But... Nvidia is currently close enough to a monopoly in this space, and our antitrust enforcement is so weak/nearly-non-existent, that I suspect that anyone who either did this or used the software that resulted from it would be blacklisted by Nvidia and never get another Nvidia GPU.
So (unfortunately) chances are very good anyone in a position to do this will just shut up and pay the rent.
Folks over time do build their own, or alternative tools.
In the model space the open source efforts have been not disappointing at all.
Apple dropped them for a reason...
Surprised its not based on wattage
Time for a Linus Torvalds middle finger.
Lmao! Sorry AI companies, even if you somehow survive to profitability, daddy Nvidia will just decide you have to pay rent in perpetuity. 99% margin on chips isn’t enough.
Someone put up a ishuggingfacedead.com death watch and just post articles like this until it's obvious.
One thing I've noticed is a lot more HF assets require personal information collection to download. Not sure if that's new corporate compliance or an attempt to manage bandwidth (or a marketing channel).
First it was licensing per cpu, now gpu?
buy the hardware and pay rent, even better /s