LLMs can spit out really great and impressive PoCs but that's all they are, PoC, MVPs. Any developer worth their salt can tell you that an MVP is the easy part, it's scaling that up (load/traffic as well as business logic) is the hard part.
LLMs can let you jump start the MVP process and make it faster to iterate on early ideas. In that way, it's very useful (and this is _not_ me writing off LLMs) but the foundation you are building on is shakey.
Maybe this says more about me but almost every time I've used a boilerplate/starter project (Not talking about bootstrap but themeforest-type templates you can buy) I've found it very underwhelming after the initial honeymoon period. I think LLMs are similar in this aspect. At the start you feel like you are saving a ton of time (and you are) but as the project goes on it gets more and more complicated and the chance that the starter project (or LLM-generated project) is able to to scale to your new needs is rare, at least without heavy modification.
I maintain that "vibe coders" (people who can't actually program on their own or are weak at it) cannot create sustainable projects. That doesn't mean they won't get funding, and hire people who can take their POC/MVP and turn it into something that you can sell and won't collapse under its own weight.
Again, I'll state I'm not anti-LLM and I use it daily but these "hot takes" about how teens are changing the indiehacker landscape are incredibly suspect IMHO. Are they spamming the indie landscape? Sure, that doesn't surprise me but other than a few flukes I don't expect to see "vibe coders" displace indie hackers anytime soon.
