In the end, Dave doesn't say that several people working on the same schematic can't be done or is never done - what he says is that typical workflows are nothing like what this startup thinks they are, and that their potential market, while maybe not void, is likely much smaller than what they claim.
We can agree that actively collaborating on the same schematic is likely to happen on large projects, in large teams, in large organizations. Problem is, this is the exact kind of setting where it would be very hard to make them switch to such a tool. They usually have been using one of the major tools for a long time, swear by it, engineers are skilled with it, they have large in-house libraries carefully crafted over the years, etc. Making them change tools would be very hard. (By "actively collaborating" I mean several people actually needing to *edit* the same schematic at the same time - while collaborating in general, as Dave also said, is a common thing and mainly comes from design reviews, part selection, pre-schematic design - yeah you don't need to get right away drawing stuff on a schematic actually.)
It's just that those startups all want to shove the software development model down everyone's throat and apply it to everything. It almost never works. Applying typical software dev methods to hardware has shown to consistently fail, while usually making engineers very angry.
And the bonus points of this tool apparently using AI and doing some automagic stuff just adds to the smelly odor.