PhD programs are far more valuable than a professor replication mechanism. As AI rapidly commoditizes more menial technical work, I think broad expertise that informs system design will become increasingly valuable. In my field of optimization, what used to require a team of engineers will be able to be accomplished by an individual.
During my PhD, I had the opportunity to work with MIT's Graduate Student Council on ways to support PhD students who aimed to commercialize their research. We've presented this work to US DoD leaders. Two artifacts of this work are below:
Facilitating Graduate Student Entrepreneurship: Barriers and Proposed Solutions
In Technology and Innovation
Support PhDs building deep-tech ventures
In Science