Good mentality to take if I am being honest.
They did lose out on a lot of things such as mobile CPUs and GPUs by simply targeting them too late. As they mentioned, due to mistakes, over confidence, and frankly a Series of Unfortunate Events from their prospective, they dropped the ball, hard
, for the next several years.
It is not possible for them to catch EPYC anymore as they are now. By the time 7nm is ready for servers, AMD will be on 5nm, and they will be on equal footing like when EPYC 1 launched. AMD will be making, probably, 96-core CPUs by then and already be on PCI-e 5.0 and DDR5, which will prevent them from needing to over-complicate motherboards further. If Intel tries to make monolithic CPUs that can compete with that, we will be in the exact same situation we are in now.
Intel needs chiplet tech of their own. If they are going to design chiplets too, be it 2D, 2.5D or 3D stacking, and they make Optane, GPUs, FPGAs, modems, etc, then it makes perfect sense for Intel to begin doing what AMD is doing; semi-custom design, and more general flexibility.
More importantly, Intel needs to actually fight ARM, and nVidia/AMD's GPU divisions, and they know it.
If the CEO is being honest, then this new mentality is required to make everyone focus on all of their rivals, and is required for them to wake up.
Originally Posted by UltraMega
It makes sense. For most consumers, desktop CPU advancements haven't really mattered in about a decade. Once CPUs are good enough, making them better doesn't really matter outside of gaming for consumer level tasks generally speaking.
The thing about building a more efficient design, is that if you do not need more performance, you can instead clock it down, make it smaller, use less power, easier to cool, and live longer.
Even burst speed, when tuned correctly, can be more efficient for lower powered systems over all.