That’s a proposal, not a proven solution, and it assumes that the chosen algorithm, implementation, and operation are all perfect or it fails open. That’s why you need to look at this as a system rather than a chance to use the current fad: this class of proposal replaces something which is cheap, robust, easy to understand, and fails closes with something complex and unproven which most people would cannot make accurate threat assessments about. There’d have to be a huge win to make up for that and “it costs a lot more and my startup could cash in” doesn’t count.