phd? I’d say work on the undergrad degree first. Phd is a long way away and you might find after 4 years of college you might be sick of it. Phd is an entirely different ball game to college, you really have to be 100% commited to an idea for 3 years minimum and do it before someone else does it(often they can take many more years, i know people who took 5-6 years to do one).
A phd wont help you really to make games, plenty of people do it without it. Phds really are useful if you want to make a career in research (work at intel labs\hp labs\ at &t\bell labs etc) or to go into lecturing etc.
As for the phd increasing your chances of a succesful start up, people might take you a little more seriously at venture capital meetings but if you talking rubbish they’re still gonna call you on it :) Unless you phd is in business analysis :)
Seems like you have a lot of ambition there, so thats always good but i suspect setting up a games company isnt all that easy (read the cover story on pooka games). As ivan said your gonna need a team behind you to get a publisher etc. but i guess that team could evolve from college mates when you get there. Also start up capital is gonna be a tricky one, i would take ivans advise again here and do anything to pay the bills while you get going (web\i.t consultancy\database work etc). I’m sure thats how the majority of games companies started. Although back in the day they probably delivered pizza, newspapers, worked on building sites etc. Start ups nowadays cost significantly more, especially considering the next generation of consoles. But that doesnt mean you cant prototype now, get the torque engine or something similar.
Finally, damian is right something in the way of distributed computing\parallel computing would be very good, because the amount of multi-threading in games which will be required in the next 3-5 is going to be fierce.