Well said, MC and angus. I used to think ab initio is the future of chemistry because it potentially has the true predictive power, but later realized that chemistry is much too complex to be explained solely from the Schrodinger equation. But I'm still a believer of ab inito methods because it can provide information independent of experimental evidence; that is, it's more like to make predictions than to repeat or simulate experiements. MD is a complementary tool of ab inito methods. The results of MD strongly depend on the potential functions, and the best ones come from ab inito methods. I also think MD is the future of computational chemistry. There are two reasons. First, MD connects microscopic properties to macroscopic ones. That's important in materials science. Secondly, the interest in nonequilibrium systems. Biological systems are all nonequilibrium (the equilibrium is death). There are nearly no known principles regarding nonequilibium behaviors. Then the best way to find out what happens is to run real time simulations. To summarize, ab initio MD is what I'm going to bet on.