That's considering production stays constant, which shouldn't be the case since new companies are being created all the time.

That assumes people are being more productive on net. I suspect generative AI is still a net negative in terms of productivity overall. Lot’s of time spent playing with it and tracking down bugs more than offsetting what little benefit it’s providing people.

I’ve heard people on HN suggest it’s definitely helping them, but I’ve yet to see it in person. So ehh maybe they are just working on very different things.

Never works this way. Efficiency gains will result in the expectation of doing more.

Maybe that “more” will be in other areas of the business or economy and thus lead to less headcount, but given how profitable tech is.. I’d assume they’d be redeployed to build more product

Most companies "need" to grow, so if you did see 10% more productivity, you aren't going to lay off 10% of your workers--you're just going to enjoy hopefully 10% more profit (grossly simplified).

Not to mention that I'm so skeptical right now that I highly doubt these AI tools are helping IT workers be even 5% more productive.

Historically, every time a new web framework comes out that swears it'll make web devs so much more productive, it doesn't seem like that ever causes a noticeable change in the job market.

What about https://en.wikipedia.org/wiki/Jevons_paradox ?