Well no. A Global latency of 2.8 ms is just bullshit. As others pointed out, they did not beat the speed of light. I hate stuff like this. If you really have a worthy product just point out what it really can do, not what in a lab environment worked once.
I agree with the buzzword bingo complaint, but they really do have a claim that is defensible here.
Their claim is that you can run an application distributed in a cloud around the world with data in their database, and read queries to their database will get results in an average of 2.8 ms.
This puts a lot of caveats on that 2.8 ms claim, and makes it something that is both good and believable. Which makes the claim very much not bullshit.
