I haven’t done the math on it so I might be off, but it could provide the benefits of both dedicated smallsat (less wait, more choice over the orbit) and bigsat payload sharing (lower cost) with less downside. In addition, it could have a cascading effect on the space.
You would just be launching feedstock and some components, so, like you mentioned, you wouldn’t have as much weight dedicated to infrastructure. Also monetary risk of a launch failure would be lower since there wouldn’t be a 100m satellite on board.
I’d also assume that because satellites wouldn’t have to withstand a launch, you’d be able to use less physically robust, more intricate components and silicon. Which would decrease aspects of development and production cost.
Construction would definitiley be more expensive at first given the current level of demand, which could be prohibitive if you couldn’t automate the process at a reasonable cost, but overall cost could be lower.
This decrease in overall cost could make it available to more companies, opening production up to more economies of scale, which is a positive feedback loop to a point.
Also the ability to spin up a new satellite remotely and relatively quickly could save companies like one web billions in otherwise unearned revenue at scale, and reduce the ammount of time that startups have to employ expensive engineers before they startup making money.