Edit: Seems I misread the Starship capacity. This is probably only to LEO. Its hard to estimate but going by the geostationary payload, it would be one third of that, so triple the amount of flights. At a cadence of 2 launches per week it would take 4.5 years to launch all of it. ( 450 launches / 2 launches per week / 50 weeks a year)
I'd guess it would be far more efficient to launch more of these into regular orbit than trying to assemble them out in L1 and keep them from drifting off, even though you would need to launch at least twice as many for the same amount of coverage.
I wonder if there's a solution that uses the dust and rocks. Maybe you nudge some near-Earth asteroid near L1 and then blow it up or something.
Technically you need a bit more than that, for its umbra to block out the entire photosphere over the entire Earth:
https://en.wikipedia.org/wiki/Tangent_lines_to_circles#Outer...
Let's see - Earth's radius is 6,378 km.
Area exposed to the sun therefore would be 6,378 * 6,378 * 3.14 = 127,731,695 [1]
If you wanted to block just 1% of the light, that would take a shade of 1 million square kilometers, which would be equivalent to a shade that is 1,130 km on a side.
Seems like it would be tough to do given all the dust and rocks floating around the sun. Also, you're only covering 1%, which is much less than cloud cover.
I'd assume you'd aim to block at the equator since that would have the most direct impact on the intensity of the sun over the course of a year.
[1] - I'm calculating the projection as a flat surface here, so the numbers aren't exact but close enough.