Few examples: (i) immutable data can clash with privacy laws, (ii) double spend is generally not a problem in centralized systems, (iii) distributed ledgers means you don't have full control of your books and records, (iv) ownerless contracts cause problems with KYC/CFT/AML/sanctions, (v) there are still courts and government force that can compel people to do things etc. ...
In order to really use these it would need a sizeable shift on how the financial structure looks like. That might happen but a lot of what I see is just a fast run through the best errors in finance over the last hundred years. For example, I see mispriced contracts because people don't understand forwards, free options being giving away etc., stuff that to me (not a lawyer) looks like ignoring securities laws, ...
I am not totally bearish on some of these ideas and technologies, but right now I think they are used and envisioned too narrowly by trying just to replace things that are often poorly understood. Example: I think the idea of making market making directly investable is great (i.e. not just being shares of an investment bank or similar). There are probably real applications outside of crypto (index components vs ETF maybe). Similarly, for DeFi to really "make it", I think it needs to be able to do unsecured lending.