If AI makes us so much more productive, why is everyone scared instead of making bigger plans?

I work significantly faster with AI. Not “we can lay off half the team” faster, but the difference is there. At the same time, even colleagues who long dismissed the possibility are now starting to worry about their jobs.

There are two things I don’t understand:

1. Why lay off developers instead of building more with the same team?

If a team gets more done with AI, why not develop more products, ship features faster, and take market share from competitors? Why produce the same output at a lower cost instead of aiming for more revenue with the existing team?

2. How much additional pressure can better models actually create?
For many of my tasks, AI already produces nearly perfect code and integrates it well. If that already works, the additional productivity gains from further improvements should eventually get smaller.

So why do we expect continued job losses as models improve? Am I underestimating how much can still be automated beyond code generation itself?

Something else that makes me skeptical: some companies also automate reviews and testing once those become the next bottleneck. I worry about technical debt, while we developers barely get a say in these decisions.

I see the productivity gains. What I’m missing is an explanation of why they should lead to less demand for developers. Where is the flaw in my thinking on these two questions?