That statement however has nothing to do with whether or not the form of education at colleges is better than bootcamps at training individuals to become good at some task set X.
It does say something about the barriers to entry in those other fields. For instance in robotics, money for hardware, access to cutting-edge research and journals are the minimum requirements for anyone interested in jumping into that field, and this jump is made easier for students who have access to all of this via their university.
I would hypothesize that the form of intense applied study that bootcamps offer is better suited than the traditional 4 year CS degree for training individuals in any field, whether that be research in theoretical computer science or building a distributed database system, mainly because of the higher degree of specialization, applied project based approach, and dedicated mentor based guidance in the bootcamp model.
This topic is fascinating. I think we have so much more space for vastly improving the output of the education process at every level of the system. But I think a lot of improvement domains will depend on the degree to which we can alter and manipulate the current status quo in industry and the system itself.