I'm a very early Zed adopter, and have gotten much more confident in it in the last 6 months mainly because of their AI plans. Note that my strongly positive opinion on this actually has very little to due with the _utility_ of their ai functionality, but rather in that it appears to be a sustainable funding model.
My biggest worry with Zed since I started using it (again, early adopter) was that it would eventually need to be monetized, and likely enshittified. I'm not at all a fan of subscription software, but probably would've happily handed over $20-$50 for a one time purchase (or, maybe $20 for a 1 time purchase of a major version, with another $20 at least 3+ years out or something).
In the last year, Zed has become a sort of AI reseller. You can buy their 'pro' plan, get so many openai/anthropic/gemini tokens, and set a max budget.
For me, this is probably as good of a business model gets in terms of staving off enshittiffication. Zed can happily take a cut, I can preview a bunch of different services, and if you don't care about ai at all, well the core editor is still free. My only worry about this model is that I think I'd have a hard time getting my employer to pay for a Zed pro plan over copilot, so I think they may have trouble monetizing enterprise users with this plan.
In any case, seeing an obvious/relatively innocuous method of sustainable dev has been a tremendous relief to me (and I'm sure the Zed devs as well).
Its biggest remaining flaws IMO remain its small extension ecosystem which hopefully is simply a matter of momentum. I'm sure they could do something like provide richer examples of how to port an extension from vscode, so it's not an intractable problem.
I also continue to have concerns re: the security of their chat/cooperative editing system as it's currently difficult (impossible?) to self host, but perhaps that will also be improved.