2. A lot of the "Launch a 1U Cubesat for $100k" figures are for the launch itself. That ignores other stuff like engineer wages, legal, etc. and is mostly hyperbole. Launching two for $200k is much more common as the second one take a lot less time to put together once the initial R&D is done. Then these "$100k" Cubesats, capability wise, are fairly useless. Think Sputnik type satellites. Want a working payload? Prepare to do more R&D. Oh and Cubesats have about a 50% rate of actually operating in orbit. First one doesn't work? Now you need to find more money to launch another if you weren't lucky.
3. The number of people who are willing to throw millions after something that currently isn't explained by science and is a lot harder to debug when you don't have someone sitting next to it in orbit, when it can done on the ground for an order of magnitude cheaper, is very small. Most people interested in using this are waiting for someone to pay for the testing first. Hell most satellite engineers I know for now still believe it to either be a hoax or will end up being just a measurement error. So people are riding this out until there is more hard evidence.
Disclaimer: I'm a Cubesat engineer
EDIT: and once I convince Milner to finance this boondoggle can I sign you up to be my lead satellite engineer?
Key word: if
A Falcon 9 can lift 10,000 kg to LOE (or more). Let's say we could get a simple Em-drive and basic satellite and solar panel in for under 100kg. (We can do better, but let's start there). That's 1/100 of the load, so (in theory) we could maybe pay 1/100 of the $60m cost for a Falcon 9 lift, or $600,000.
Half a million bucks plus some change, let's say.
Edit: Less if we can get on board a previously-enjoyed rocket!
I'd be astonished if the cost of this was less than $2-3M.
So not much, but the larger problem is we don't know how this drive works. Miniaturizing it could impact performance, as could other electronics in the satellite... given the unknowns it makes sense to let NASA do some more terrestrial due-diligence before going to space.
Maybe Hawking and Milner should be considering this for Starshot?