> I believe there will be a need for far fewer of us.
I disagree. Throughout the history of computing the number of SWE's has gone up and they have become, overall, more capable [0]. That hasn't cause a need for fewer of us, in fact the demand remains overall strong.
I predict instead it will drive the need for more of us rather than less. And potentially even "better" (or at least healthier as I see it) jobs. What I mean by that is, instead of outsourcing (locally or abroad) companies will be able to bring someone in-house. In the past, dropping the load on 1-2 people would have been untenable, with LLMs/AI a number of things become plausible. And this scales to every level. The company that previous outsourced their development and could have hired 3-5 developers (but that wasn't enough) can now consider hiring them.
Yes, the larger corporations might decide they can get the same done with fewer employees, or they might decide to do more with the same.
I do think a number of SaaS tools might be in more trouble if companies decide to just build in-house alternatives that are custom for their exact needs. But perhaps SaaS will adapt to "source available" or some other kind of way to allow easier extension of their core offering so that companies just build a sidecar or thin layer over the core that covers their needs better than a "one size fits all" SaaS ever could.
Perhaps this is too rosey of a view of the future but it's one that I think is at least partly backed by the history of the industry.
[0] And at least what they're able to produce, even if you disagree about the quality.