Continuous improvement is the bedrock of pretty much everything modern. It’s a lot easier to make frequent minor revs than it is to do rare, gigantic revs.

If you work in software, think of it as normal SDLC versus once-a-decade waterfall style total respec and rebuild. Which is more efficient?

It also spreads demand more consistently, rather than having 90% of demand landing in the first year and 10% spread over the next four. What would that mean for production capacity?

Continuous processes beat batch processes every time. See also: Tesla’s ideas about model years.