Just to play devils' advocate here... I switched from a PC to a Mac years ago. It was an expensive undertaking as I had to replace all my Windows software with Mac software. In addition, some hardware wouldn't work on the Mac due to lack of software written for it, so that had to be replaced as well.
Yes - it is vital to do the homework on what software and hardware you need. Many areas are well-supported on Mac, some (e.g. audio production) arguably
better supported - others (CAD has been mentioned) mean you should probably stick with PC. Also, although M1 is an improvement in most respects, that's at the expense of easily being able to run X86 Windows in bootcamp or VM (
yes you can
run Win11 for ARM in a VM - but with a
long list of caveats, such as being stuck with a 'developer preview' version and highly ambiguous licensing).
SoCs are new to Macs. By design, you can't NOT have better performance over a non-SoC design. Once everything is SoC based, that initial leap in performance from a non-SoC will simmer down a bit. PCs are already looking to add SoC designs, so Apple's speed boost won't be that huge for very long.
PCs are still going to be hampered by the need for backwards compatibility, which means that every x86 processor has to carry around a big chunk of circuitry to translate x86 code to RISC micro-ops that ARM ISA-based systems just don't need. It's a pretty good bet that ARM will
always be able to fit more cores & accelerators into the same space & thermal/power envelope as x86.
Even if Windows-on-ARM takes off, it still has to satisfy a huge, conservative corporate market that will demand high-levels of backward compatibility with old code, old technologies, protocols etc. and a developer culture that has always presumed that code will be running on x86. Apple, have always had
much more freedom to drop old standards - and a
lot of the advantages of the M1 come from things like having a lot of software optimised for modern frameworks like Metal etc. Last I looked, MS were still supporting 32 bit Windows which could run binaries from the 1990s - Apple are quite happy to kill compatibility with 10 year-old code, and have been (indirectly) leading their developers to develop CPU-agnostic code for years (switching to a radically different CPU every decade or so can do that). Also, Apple now have the luxury of being able to design processors
just for the computers and operating systems that
they want to make - probably even prioritising their own
software like FCPx and Logic. So, yeah, the gap will narrow, and maybe eventually PC will take the lead, but if Apple play their cards right they should have the lead for a few years yet. Rumours so far have a successor to the M1 coming out next year.
Another thing to consider there is literally nothing you can do to upgrade a Mac (these days), on your own. They've soldered the SSD to the board. The RAM, et al. What you order is pretty much it. It's not like you can buy a Mac with less RAM and buy some RAM later to beef it up.
As for RAM - you need to compare like for like:
no computer with LPDDR RAM will have user-upgradeable RAM, because LPDDR relies on surface-mounting and ultra-short tracks to the CPU for the "low power" bit, and simply isn't available in plug-in form.
I'm not going to defend the soldered-in SSDs on Intel Macs - they're a perishable component and should be on a blade, even if it isn't standard M.2/PCIe. Reality is, you'd probably have to try hard to wear it out.
With the M1, though - there's now a user-payoff in having unified RAM built in to the SoC package and a proprietary SSD interface in the form of hyper-fast RAM and SSD access.
And don't get all caught up in the M1 hype. I
Again - it's a case of comparing like-with-like. What M1 is really giving you is the performance of a middling desktop i7 PC with a basic
discrete PCIe GPU burning 300W of power - but
in a ~15W ultra-thin laptop. It
thrashes Intel's ultra-mobile processors with integrated GPUs - and probably gets an even bigger boost on tasks that can take advantage of the built-in hardware video codecs and neural engine. What it means is that a thin'n'crispy laptop with decent battery life is now good enough for pretty much all "general" tasks and occasional "heavy lifting" like media production.
Once you've got desktop-levels of performance at such low power there's a good chance that you can scale to workstation-level performance by adding more cores and specialist acceleration hardware (most x86 CPU and GPUs from the mid-desktop level up rely mainly on increasing the core count) - which we've seen with the M1 Pro and Max. No, the new MBPs are not the fastest things to be called "laptops", but the competition tends to be from brick-thick "mobile workstation/mobile gaming" space heaters with 1 hour battery life.
