Take your pile of investment money. Now, take half of it (or more, if you'd like) and destroy it. Return the other half to your bank account and go home.
Space is a rough business. It's highly regulated. Your competitors are often subsidized entities, sovereign governments, or extremely well politically connected. Testing is hard because you often only get one shot to get things right, and if things go wrong it could set you back years and destroy hundreds of millions of dollars of work (made only slightly better if you have gone to the trouble of spending lots of money on insurance) and could generate a massive amount of bad PR in an instant that will haunt your company for years. Export controls are a bitch. A lot of industry standard parts pipelines are heavily optimized for massively overfunded government contractors so getting your hands on things like radiation hardened CPUs is going to cost an arm and a leg.
And these are all just the problems that aren't related to the core, irreducible complexity of building whatever it is you are wanting to build.
Ultimately any advice is going to be heavily contingent on your goals. Do you want to put colonies on Mars? Hotels in Earth orbit? Telescopes in space? Rovers on asteroids? Do you want to build spaceships, launch vehicles, space suits, scientific instruments?
-
In a broad sense my advice is this: start small (garage project small); work iteratively; explore some problem space that isn't receiving as much attention as it should; acquire domain knowledge, expertise, and experience with bending metal and hands on the hardware; leverage existing domain knowledge (through hiring or personal experience); make connections; and have a vision.
Some ventures to draw inspiration from: SpaceX (of course, http://www.spacex.com/), Armadillo Aerospace (http://www.armadilloaerospace.com/), XCOR aerospace (http://www.xcor.com/), Johns Hopkins University Applied Physics Lab (http://www.jhuapl.edu/), HyperV technologies (hyperv.com), Malin Space-Science Systems (http://www.msss.com/), CubeSat (http://www.cubesat.org/), and the many nanosat missions.
I'll draw special attention to HyperV as an example. They have previous domain knowledge with plasma jet technology as it applies to fusion research and they used kickstarter to fund research into pulsed plasma jet electric thrusters for space applications.
Edit: if there's still interest in this thread tomorrow I'll followup with a list of open problems in spaceflight / space exploration that are amenable to investigations or R&D at a relatively small scale.