Most CPU manufacturers are scaling to multiple cores, not because it’s a huge benefit, but because they have nowhere else to go (not being able to run up clockspeeds like the old days allowed).
The truth is there is a sweet spot. Power consumption increases linearly with clock frequency (at least if you can raise the clock without raising voltage, otherwise it increases by voltage squared as well), so you have a choice to make. If I have the ability to double power consumption, should I double the number of cores, or should I double the clock frequency? Turns out, it depends on the workload.
Of course, doubling the frequency means all that extra power dissipation is happening in a smaller die, and heat removal is a function of die area, so if I double the frequency I may need to increase the die size and spread things out as best I can to allow thermal spreading. But if I do that, the distance between transistors increases, meaning the time-of-flight (6ps/mm) on wires increases, and also the capacitive loading on the source/drains, which tends to slow things down. So I may have to increase the voltage to compensate. But that generates more heat. Etc. etc.
So, yes, doubling the number of cores can be a more elegant solution. And it turns out that for most real workloads, somewhere around 6-8 cores is the sweet spot, with decrease return on investment as you go beyond that. Of course, some workloads are not parallelizable, but these tend to fall into two buckets. (1) stuff that runs fast enough as it is, and doesn’t benefit from increased clock speed very much. For example, interactive apps that can already process faster than their inputs can be generated. (2) stuff (like engineering or science software, encodes, etc.) where more speed is always better. This stuff tends to be comparatively rare, but for those situations you are better off with a design where high single-thread performance was the goal.
M1 is in a pretty nice sweet spot, with very high single-thread performance/watt, plus a reasonable number of cores. There is still plenty of room to (1) raise clock speed, because the power dissipation is so low and (2) add more cores (returns will diminish if they go beyond another 4 or so high performance cores, but even more cores than that would be helpful for certain workloads, e.g. Mac Pro situations).
But, by the way, none of this is because the CPU manufacturers hit a wall. This was ALWAYS the trade off. The issue was, back before, say, the mid 1990s, the number of transistors that could be squeezed on a die was too low to enable much in the way of multicore designs, so the only choice we had back then was higher frequency! So you sort of have it backwards.
