You got to quote people in the right context though. In general yes the RISC/CISC divide has not mattered much for large chips.

But Apple's aggressive use of so many decoders has been considered by many as a kind of new development. Not necessarily foreseen by Jim Keller.

Secondly he is obviously talking about CPU cores. A lot of M1 performance comes from heterogenous computing, using specialized hardware for many tasks.

What you are also not taking into account here is that he is talking about fat cores. Once you aim to have lots of smaller cores, x86 complexity starts mattering more. It is very hard to imagine something like Esperanto Technologies ET-SOC-1 with 1088 RISC-V cores doing with x86.

We also see how Ampere, Nvidia and other are making Arm chips with way more cores than you find in the x86 world.