Option 1: Split a few hundred developers into groups of 5-10 each, and throw money at them until you generate a few hits (we'll assume hardware costs are neglig

Option 1: Split a few hundred developers into groups of 5-10 each, and throw money at them until you generate a few hits (we'll assume hardware costs are negligible because for this group they are). Offer low salaries and 'equity' in return, putting as much risk on the shoulders of your employees as you can - after all, for the successful ones you can always screw them out of their shares later. For now, if you do this at scale, you can be reasonably certain of profitability. Option 2: Form a single company where you have to manage that entire group toward a single goal, take all the risk, probably pay some non-negligible hardware costs if the problem is interesting enough, etc. Then you either solve the problem or you fail utterly, and even then you still get to find out if anyone really wanted the problem solved in the first place, so you can still fail utterly. Then once you get past that, you can deal with patent trolls and copycats, and for each case you have about 50% probability at best of getting any kind of justice at all. Option 3: Do the first to fund the second. Sadly there is only one Elon Musk, though.