Intriguing spiel. As I am generally very impressed by the predictive power of the scientific theory, I am inclined to agree with the speaker. What an awesome achievement it would be to have such predictive power of material properties! However, in my brief experience running DFT calculations, there seems to be a lot more adjusting of the algorithm schemes to match experiment than vice versa: either I misunderstand this aspect of DFT (very possible) or these materials computation methods will occur in a very different manner from present-day DFT. So I am optimistic but skeptical about this vision.