The fastest desktop chips have like 64 cores with SMT. The M1 cannot compete.
Unless you mean for single-threaded taks, but this is not what these multicore CPUs are made for.
At those tasks, I think the M1 consumes about 2-3x less power than AMD Zen 3.
I am talking about the demonstrated performance of the architecture, that is, what the CPU core is capable of. The baseline performance limit, if you want. Fastest currently shipping Intel or AMD cores are about on par with M1. There is no question that large multi core arrangements deliver more aggregated performance, but that is the question of scalability. There is nothing preventing Apple from building multi-core designs of their own. And frankly, I expect Apple to do even better here, because of their very low per-core energy consumption. They won’t need to downclock their cores as much, unlike Intel or AMD, retaining a larger relative performance advantage. In a large design, Intel can only deliver half the peak performance per core (2.4 GHz instead of 4.8 GHz for example), while Apple could deliver 80-90%. An 16-core Apple chip should be able to take on 28-core Xeons or 32-core EPYC for example.