Blockchain, in essence, is just a voter protocol that can deliver trusted consensus in adversarial (Byzantine) environment. This is its main value proposal for international finance.
I'd also like to see a blockchain-based public voting and election protocol. But there seems to be not enough governmental interest in that ;)
$850 million was blocked before they went through, and about $40 million of the remaining has since been recovered.
How would a blockchain have helped in preventing or recovering from this attack? If the hackers got access to their blockchain credentials, they would have been able to issue similar transactions. It would just have been much harder to block the transactions or reverse and recover them after the fact. It seems to me that running this on a blockchain would have made things worse, not better.
Also, consider rouleth.com, which is a provably "fair" online casino. (The odds are still against you, but you can verify that the bank did not cheat when you lose ten times in a row.)
Or imagine companies with shares on the blockchain, combined with smart contracts. That would allow you to automatically enforce all kinds of shareholder agreements (for example vesting or a tag-along right).
Maybe that's not revolutionary, but it could be quite useful and save a lot in legal anf auditing fees.
But I still fail to understand the attraction to corporates. By putting your blockchain in the Azure cloud, then you have a central point of failure? I agree it's an immutable record of everything that has happened (provided there's not a "hard fork" or over 50% of the resources owned by one participant, which is hugely possible with a small network) but immutable records are possible even today.
So, what does it give you? An immutable, distributed, public database that can be trusted despite not having trust in any particular participant.
Now, if you're a bank, why have you got offices with marble floors in the nicest neighborhoods? Why do your employees wear suits and ties? Why do you get a credit rating?
If you want security, that already exists. You can use encryption to identify people, and you can use it to hide sensitive information. You can use it over public channels, no problem. For instance this forum is on an encrypted https connection, and people can't sit on the route between me and YC to find out what I'm up to.
If you want a ledger, that's also been done. In fact you can use encryption to make sure there's never an unauthorized change to the ledger. And there's a whole load of ways that ledger can be scaled so that many people can use it at once.
Someone mentioned settlement of trades as a potential use case. But what is really to be gained against the way it is currently done? At the moment, you have a bunch of ops guys settling trades in the way it's been done for many years. If there's an error, you can call someone one the other side and fix it. Mostly, it's automatic, done by various cron jobs. What do we get from putting that in something similar to bitcoin?