Thinking's only half the battle.
I'd like to believe this would be the next step in the rise of the machines, but it strikes me that the one thing machines don't do so well is be inventive.
If we want a machine with Einstein-like capability it won't just have to be able to think like him, it'll have to be able to imagine itself sitting on a light-ray and viewing the universe from that perspective like him (this, if you didn't know, is one of the ways he came up with the theory of relativity).
If we want a machine that's able to construct new and useful models that enable accurate and testable predictions to be made and to do that it's going to have to be able to do more than just work through existing problems - it's going to have to come up with new ideas and I suspect that's a much tougher nut to crack.