Agreed, the only thing I'd add would be AMD's self-sabotage. Pretending to have compute capability (see also "right around the corner") for over a decade when i

Agreed, the only thing I'd add would be AMD's self-sabotage. Pretending to have compute capability (see also "right around the corner") for over a decade when it was actually so broken as to be sub-viable was an enormous own-goal that they doubled down on when they exited desperation mode without rehydrating the teams and tripled down on with the RDNA/CDNA split. By rights they should have been fast followers, but they stacked the critical mistakes so high that frankly I think NVIDIA's subsidized GPGPU classes are the smaller part of this story. If the people struggling to get OpenCL to work had been able to get it to work (on other than NVIDIA GPUs lol) the situation would look very different today.