I don't remember either, but I do remember Linus Torwalds not being entirely satisfied with NVIDIA. That is actually something very hard forget.

https://www.youtube.com/watch?v=iYWzMvlj2RQ

Every time I have to reinstall CUDA, I fondly remember this video.

Apple used nVidia GPUs in most of its computers. Around 2008-2009, nVidia produced a line of faulty GPUs that were used in the Macbook Pro. This led to widespread hardware failures, and was known as "Bumpgate" because the problem was caused by faulty solder bumps inside the GPU chip. nVidia refused to fully reimburse Apple for repairs because it wasn't contractually obligated to do so. In response, Apple switched away entirely from nVidia GPUs.

I think it was well into 2013 because both the 2011 and 2013 MBPs I had suffered from it. but I thought that was an apple issue (ie not soldering the dicrete gpu well enough to save money on it)

after the 2 faulty products in a row, I haven't bought an apple product since.

No, it was Nvidia because of the way the BGA solder was done. The problem was across manufacturers.

The infamous early 2011 MBPs had AMD GPUs that basically crapped out the entire logic board. That was what got me to switch back to Windows.

You are missing out, the M series laptops are insanely good. Now wondering how the field of ML would have developed if Apple had had Nvidia chips all the way...

And that's kidna funny because now Apple Sillicon has pretty great graphics on their own.

Maybe that's the lesson he learned and why he _supposedly_ was making sure his GPUs were still getting to China. Because he knows that the alternative would lead to good Nvidia Chinese competitor later on