I can’t say for sure, but having read opinions on the RISC (reduced instruct set computer, e.g. ARM, MIPS, RISC-V) vs CISC (complex instruct set computer, e.g. x86) debate over many years: Generally RISC was and still is viewed as good idea, leading to simpler CPUs, which in turn would reduce the size and power consumption as well. RISC architectures had quite good success in embedded applications thanks to that, but for laptops/desktops, they didn’t get there until I guess with Apple recently.

However, Intel seems to have managed to mitigate the inherent disadvantage by basically subdividing the complex instructions into micro instructions, which (by my understanding) means that a modern x86 CPU is kinda similar to a RISC one under the hood, but with some more complex instruction decoder logic on top. So last I heard this still gives a slight advantage to RISC in terms of die size and power consumption, but not as big as was once thought.

One maybe more relevant difference is the path they took to arrive at modern laptop CPUs: Apple’s M chips are descendants of the iPhone SoCs, which were designed for low power first, and were then improved for performance as a secondary objective, whereas Intel’s are descendants of powerful desktop/server chips, that got power-saving features added in over the years. Maybe that approach is just inherently harder.

More likely though, Intel’s economic incentives just weren’t as strong to actually improve power consumption, what with them having had a monopoly on laptop chips. AMD not being able to do it either could probably be explained by the fact that it doesn’t and never had anywhere near the budget of Intel or Apple. They were just barely hanging on by making (consumer/gaming) desktop CPUs and console chips, which aren’t that big of a market compared with servers and laptops, and (the latter especially) have shit margins, and both are more about performance and less about saving watts.