Any application intended to resist modern government surveillance is going to be extremely difficult to write, because it has to be resistant to bogus secret "court orders". The only way I know to do that is to have many independent developers engage in multi-party signatures of reproducible builds based on audited and reviewed open source code. If they're just going to run a company that develops it in a proprietary manner how will they achieve that?
I am more interested in Pond. It's being written by an actual cryptographer and he already has real, working code (though it's nowhere near releasable). It's up front about its security model and which threats can break it. It's built on top of Tor and even supports using the TPM chip so that when you press delete, the data is really really gone beyond the ability of any forensics tools to recover. It's even designed to resist traffic analysis. Anyone can run a server.
The main differences are that, obviously, Pond is not developed by a company, and it is focussed on asynchronous email style messaging rather than instant messaging. It's also got a very strong threat model that means it compromises on usability - for instance, there are no addresses in Pond, instead you are expected to hand out small files (perhaps on NFC tags?) to people who you want to be able to receive messages from (this is an anti-spam measure).
Despite all that it's a very interesting piece of research.