The competition I see is from two separate directions, 1) AI companies training on the prompts you send them 2) users building their own a la carte solutions.
I don't think it entirely erases Saas, but it does take a huge bite out of it. Building a Saas product seems somewhat trivial at this point.
The larger hurdle now looks like having startup capital (free time to completely fail, or money to pay multiple founders to do that.) That's the risk, and the reward is income generated over time.
There also don't seem to be tons of SaaS paradigms at scale that differentiate companies other than entrenchment. It's somewhat realistic to assume that a handful of companies are going to pop up that connect simple applications to SaaS backends, which lowers even more the amount of income you can generate from a SaaS and would open up the level of competition to folks who can vibe code.