so the satellites launched in 2005 had picked CPUs, memory, and cameras, YEARS before they launched. they went into orbit with computers and cameras from maybe 2003 or 2000 or whatever. certainly not ones from 2005. and thats what they used for 10 or 20 years

so you can have a satellite up there right now that has computers and cameras from 2010 or earlier and thats what its stuck with, right. you can't just improve it once its launched

but if you're Planet Labs, and you're launching MUCH lower cost satellites every year -- which, again, they ARE doing -- you can improve the specs every time you launch, putting better cameras up every year, better CPUs, more RAM, more harddrive space, faster comms

so your satellites get better and better, or if you stick with the lowest cost options for the task, they get cheaper and cheaper

but there's another factor at work still: with dramatically smaller satellites, you don't have to buy an entire launch, you can piggy back on other launches. this was commonplace by 2015, where nanosatellite launches routinely piggy backed on SpaceX launches

rather than paying the FULL price of a FULL rocket, you just have to pay the fractional price of the launch. its still expensive per pound to low earth orbit (current about $1500 per pound to LEO), but prior to that, you couldn't divide the price like that. it was all or nothing

or close to, at least. sometimes you'd share a ride with one or two other satellites. but the price tag was enormous. nowadays there's standardized cubesat launch services.

its not hobbyist cheap, but its still pretty cheap compared to the whole rocket

so nowadays you can launch a 1U cube sat for something like $100k. thats a bunch of money for the launch, to be sure!

but if thats the low-end price tag, lots of companies can launch satellites, and lots do

before, the low end was tens of millions

and so because of the smartphone war, satellites can now be small enough that their launch costs can be accessible to tiny companies, and we can now get basically daily images of the entire surface of the planet because the we can do so much better than in 2005

the math of mere multiplication, using 2005 costs for satellites, is NOT APPROPRIATE for calculating whether its feasible to get daily whole planet imagine, because external changes shift the OPTIONS, and therefore shift the costs