We can get ideas from past transitions.

When the iPhone came out with "Web apps only" many OS X developers got into jailbreaking because iOS was based on OS X and they could use their existing knowledge. A little later when the official iPhone SDK came out, those developers were the hottest in the world and got paid huge amounts to develop simple apps.

When Quark XPress decided not to support OS X back around 2001, Adobe InDesign took many of their customers. There are Mac apps that will never be updated (or will take five years like Adobe) for 64-bit or ARM and people will be looking for replacements.

I'm not sure it'll have a big effect on software. The vast majority of application software will work the same, only with fat binaries containing both x86 and ARM code. The few bits that are optimized for particular CPUs, like video codecs, already exist for both CPUs.

The most likely opportunity is applications that are currently too sluggish except on big Mac Pros, but would be more popular if people could run them on cheap ARM macs. One might be real time video manipulation for live video conferencing.

You can get your feet wet by developing for iPad, or ARM Linux.

There's going to be a gap made by people who don't port over their software, certainly worth looking at that gap and see if there's something you can provide.

Most software should be just a recompile if it's well packaged. Of course that's a huge if and not everybody wants to support Mac anyway, and I lost a ton of my steam library (not that anybody really plays games on Mac anyway) in just the 32-64 bit transition recently. I also lost textwrangler, which was the greatest Mac text editor ever created and I refuse to pay a subscription for its 64 bit successor.

Fwiw: the textwrangler-equivalent unregistered version of BBEdit (basically without the features that used to distinguish the two apps) is free.

Get good at ARM.

It isn't just Apple that will be selling ARM computers. Soon we will see good workstations based on server-type ARM chips that will support absolutely insane bandwidth and number of threads. For now we have

https://www.nvidia.com/en-us/autonomous-machines/embedded-sy...

The software ecosystem is ragged so there is work to be done. ARM has a bright future whether or not Apple switches macs over.

What's there to get good at? C is C, Swift is Swift, and CUDA is CUDA. There's little asm out there and it has already been ported to ARM because of phones.

ARM is much less developed than x86 from the viewpoint of a Linux developer. In the case of x86, for instance, you can usually download Python wheels that are already built, you can find wheels that are built for various SIMD levels, with highly optimized numerics, etc. Building software on ARM that isn't phonish is more of an adventure.

If your goal is "personal workstation that smokes current workstations" based on an ARM processor with a monstrous core count and memory bandwidth you are going to have to confront the problem of making a 64-core machine feel snappy, which means you need to take advantage of parallelism to a high extend, understand the very different memory consistency model, etc.