Without wanting to get involved in the purpose of your rhetorical question (the purpose being some kind of analogy to the StarLink situation)...
How many compilers were there back in the 60s? By codebase or by install count, your choice. Hundreds of codebases and tens of thousands of installs?
How about now? Tens of thousands of codebases and billions of installs?
Do you think this could have happened without lowering the cost of transistors pretty hugely? (circa 3e7-fold price decrease in 50 years)
Seems to me that lowering the cost of transistors DOES increase the number of compilers, whether by counting the number of distinct codebases, or by counting number of users, or counting number of installed copies, or number of times executed, or whatever.
Here's the logic: If space launches are expensive, you spend more on your satellite to make it worthwhile. Give it more features, etc. Then spend even more to make it extremely reliable, because replacing it would also be expensive (both the launch and the hardware itself). If launches are cheap and routine, you can just launch a cheap simple satellite for each type of sensor you want, and even if it fails you can launch backups (or just activate stand-bys already in orbit). You can even aggregate the sensor data from a swarm of smaller satellites to generate the view of a single large virtual telescope, potentially bigger than anything you could construct on the ground.