In Physics, a Masters typically means 2 years of coursework and passing some set of exams. Most of us join a research group about 1 year into grad school but it's hard to do real research until your classes are done.
It is probably easier to start a company on the side as a grad student than to wait until you graduate. Grad student salary sucks, but the flip side is that it can be a pretty awesome day job. The guy who pointed out that 50% of the average grad student's 60-hour week is spend playing WoW is pretty much on the money. You can go months at a time doing only 20 hours of research a week, or even 10, and still not get fired.
Of course, you're not getting paid much, but you're surrounded by lots of other smart young people who aren't getting paid much either, so it's kind of fun. And you certainly aren't going to graduate if you don't do a lot of work -- but if your preferred exit plan is to found a startup and drop out, who cares about that?
It's actually rather hard to fire a grad student. The adviser pays the student very little, but in return (s)he is expected to graduate the student. A student who is fired by Professor X three or four years into a Ph.D. program is well and truly screwed. They might sue. They will certainly tell every grad student they know, and they in turn might tell all the first-year students, who will avoid working for Professor X like he had the plague. So, once you've gotten established as a student in a research group, it's not so hard to spend six months or a year doing only 20 hours worth of work per week. Your adviser will roll her eyes a lot, but she's not likely to take action until you've accumulated a really solid, long-term record as a slacker.
...which is why the question rarely arises. The faculty have devious ways of eliminating students that don't make the grade. Why fire people, when you can just collude to make their lives so miserable that they choose to quit?
Do it properly, and the other students will just assume that the problem is a personal issue. If you're a successful researcher, it won't scare anyone away from your lab.
Your PI knows the weaknesses in your research and knowledge better than anyone, and can use this to your disadvantage. I've witnessed faculty who aggressively attack their own students during examinations (a truly disastrous scenario). More often, advisors will simply fail to defend their poor students from unfair attacks by other faculty. Departmental lectures, defenses, oral exams, etc. are remarkably subjective affairs, and one bad comment can snowball into an event that will add years to your time in grad school.
Another very common tactic is for the PI to hand out the best research projects to favored students. There's nothing so demoralizing as watching a first-year grad student get an publication in Science or Nature (by being assigned to an already successful project), when you've worked for years on failing experiments. I've seen this happen too many times to count.
Like I said, there's no way to list all of the different ways that your PI can make your life painful. It's just an extremely unbalanced, unregulated power relationship, and with any creativity at all, an advisor can use that power to destroy you.
So, ultimately, if your startup-company moonlighting starts to eat too much into graduate research time, you're going to have to fish or cut bait: Either keep your adviser happy, or take the necessary leave of absence.
Note that (most) graduate advisers are not evil people. I suspect that many of them will quite happily grant you six months off to pursue your startup-company dreams... if the timing is right (hint: don't ask until just after the Nature article is submitted) and after they explain to you that your publication record and your graduation time will be set back by a year or more. Oh, and you probably won't be given the plum projects ever again, since your adviser knows that you have one eye on the door... those are the breaks of the game.
One very special caution: Your adviser may try to enlist your help in his startup company. Be very wary of this situation, or of any potential adviser with a history of this. You want your adviser to be unambiguously focused on helping you graduate and get a better-paying job... or as unambiguously as possible, anyway.
I don't agree with your statement. I think we hear more stories of dropouts starting companies because they are rare and make interesting stories.
For instance: 1) http://en.wikipedia.org/wiki/John_Warnock 2) http://en.wikipedia.org/wiki/Irwin_M._Jacobs 3) http://en.wikipedia.org/wiki/Henry_Samueli
. .
That is basically what I was driving at in my original post. If you are a PhD candidate 3 years into grad school are you more likely to start a company by dropping out or by staying in school?