I understand little about the entertainment industry, but to a gross approximation, I think it is largely an iterated game where ~2,000 people come together in clumps of ~10 to build a defined project on a tight timeline, and then they disperse again, and many of the clumps for a given project have worked together before, and if you stay in the game long enough you become known for being good at whatever hyperspecialized thing you do.

This does not describe the traditional practice of building startups, but I think as the field gets more understood and as velocity increases that there is a chance that this model gets adopted more frequently.

I could certainly envision a future where a lot of SaaS is produced by semi-persistent flocks of teams in a studio-esque system, with the goal of "Derisk the business, build the product, install the marketing/sales engine, turn crank three times, sell to PE, cycle rinse repeat." In this model, the people stay the same (out of some broader cast of characters) while the products rotate relatively quickly. We see the opposite in startups historically; a startup is, if successful, a 10+ year commitment, with different strata of employees staffing it at different times for 3, 4, 5 year stints.

If you predict we move to the entertainment model, you should also predict that there will be some people with "star power" who have played the game multiple times, won it multiple times, and are able to make marginal projects successful largely because other people say "Oh, X is signed onto this? Eff yes I want in on that adventure then." Serial founders are a bit of a thing in startups. Engineering managers (and sales managers, etc) quite commonly bring along people who worked with them previously. You would expect to see the impacts of these behaviors get much more acute.