I think Intel failed to realize that they had made the right call with iPhone: their very culture isn't about being innovative, but providing microcode flexibility at high instructions/watt. They had a chance to define servers, and still do. They should have been all over the whole spectre/meltdown/timing security issues and owned creating a secure server chip. Instead, they've fretted away so many options that they never had a chance to win.
My two cents.
CPU security is actually one of the few topic Intel is extremely good at (don't trust the RISC/CISC flamewar) and could define the way cloud providers are built upon.