No evidence of any useful non-Bitcoin blockchain applications so far, so no reason to expect any Amazon/google type successes. If someone figures out something useful to do with “blockchain” it will probably blow up quickly.
1. for the CEO/manager in a big company - a very high security mechanism, verified by the billions of $ in bitcoin. Great quality assurance, and politically, that's a good choice to make for something as risky as computer security.
Nobody got fired for choosing blockchain.
2. Trust in needed when you want many companies to collaborate.
Nobody like centralized trust(a-la banks). They charge a lot of money, they usually end up as a monopoly, and they have power over you. nobody likes to give up power.
IBM's blockchain, i.e. Hyperledger, is open-source, governed by many companies, so i assume the money/power situation is far better. You could probably leave IBM's hosting if you really wanted.
It's ridiculously expensive per transaction, your account can be wiped out if someone gets your token, it's not "untraceable", etc.
These are all problems that are close to being solved. Sharding will solve high transaction fees, multisigs and account recovery mechanisms already exist for private key safety, zkSnarks solve privacy.
Blockchain trading is simply a dark pool with less protections, and dark pools have also been a thing in finance for decades.
Keep in mind that much of finance is not obviously useful from an outside perspective, so "here's a trade you could do" is not enough to show usefulness.
"Dai is an asset-backed, hard currency for the 21st century."