[paulmd]

AMD and Intel also already "own it all" in this sense. NVIDIA is merely a third competitor in this space, with a much more undesirable CPU IP.

You should welcome a third competitor to the duopoly that has strangled CPU development. We surely could have made much more progress if x86 was not limited to only two (really three) competitors, you can already see how much change that AMD getting back in the game has made.

And I'm not sure Huang is going to burn it all down anyway. That seems like it would be a shortsighted move that would negatively affect the long-term value of ARM.

But I mean - I don't think anyone can deny that Huang would do great things with ARM. Terrible, perhaps, but also great.

(and it says a lot that a lot of people are probably nodding along with a comparison of one of the greatest tech CEOs of all time to literal Voldemort, the public opinions on NVIDIA and Huang are just ridiculously hyperbolic)

[tarlinian]

AMD and Intel don't actually own it all, certainly not from a software perspective. That's the difference here. The upshot of this article is that Nvidia basically wants to "CUDAify" the entire datacenter software stack. Intel and AMD absolutely do not have that kind of lock-in. You don't need to use their proprietary language to write programs that run on their systems.

[klelatti]

This is such an important point.

Plus CUDA shows what can happen even when alternatives are available (OpenCL for example) you just have to use the hardware / software integration to be sufficiently ahead and establish a virtuous circle.

[Sulik]

CUDA is basically just C/C++ with parallel-programming concepts. The "alternatives" like OpenCL are still tied to their graphics API origins.