Are you kidding? I wouldn't say it's impossible, but "no reason to think that techniques for performing more traditional sql-style queries on distributed data won't quickly surpass it" is quite a stretch. There's plenty of reason to believe that. Mainly because of the number of databases and the sheer volume of thought put into RDBMS', both theoretical and practical. Sure, when it comes to massive scalability there are dozens of promising new technologies that beat SQL databases, but they all make tradeoffs, there's just no way not too when you're talking about something as powerful as SQL.

I imagine there will be some great improvements that come out, but I think we're past the point of any major data store revolutions unless something fundamentally changes like quantum computing becomes a reality and we have to all basically start over.