I do enjoy the speaker’s enthusiasm and his critical delivery of modern in silico material engineering approach. His solution is certainly a vital step in modern material design.
However, I would currently be a little more inquisitive with regard to introducing experiments only as a post-hoc approach. While computational approaches are a terrific way to explore new, and possibly unknown, substances and materials, chances are that further experimental constraint(s) would be needed during the models’ development. A well-defined experimental test may help ensure that, for instance, important details of lesser likelihood are not omitted.
I would therefore be in favour of taking in silico and laboratory experiments in a feedforward, rather than the consecutive, way.