A couple of commenters have pointed out that Obama's demographic coalition in South Carolina wasn't much more impressive than his coalition in Nevada, where he lost to Clinton by about six points last week. There's something to this: Obama won a roughly comparable share of white and black voters in both states (at least once you factor in Edwards's improved performance in South Carolina). The big difference was that Hispanics made up a large bloc of voters in Nevada, and he lost them overwhelmingly, while African Americans made up more than half the electorate in South Carolina. (You can peruse the Nevada entrance polls here, and the South Carolina exits here.)
I'd point out a couple of things in response. First, Obama didn't do that badly in Nevada--slight improvements in a few demographic categories would have given him a shot of winning. And he did improve slightly in some of those categories last night. For example, he lost white men 46-40 to Hillary in Nevada, but basically tied her among white men (28-27) last night.
More importantly, though, the big concern in South Carolina was that the contest would become racially polarized in the extreme--that, thanks to the state's fraught racial politics, overwhelming support among African-Americans would cause Obama's support among whites to drop, possibly near single digits. That seemed like a real risk heading into yesterday, at least depending on which poll you consulted. And Obama's showing vaporized those concerns. So in that respect the pundits are right to tout the breadth of his coalition.
The broader point is that you can't look at the demographic data in a vaccuum. Taking a similar share of the white vote in a western state with a relatively small African American population and a relatively benign (as these things go) history on race and a Southern state with a large African American population and an extremely tortured history on race are just two completely different things.
Update: Several of the commenters are missing my point here. We all agree that South Carolina is demographically different from (most of) the rest of the country. We also agree that Nevada is probably closer to what the rest of the country looks like. My point was just that, since white voters in South Carolina were going to be a lot tougher for Obama to win over than white voters in Nevada, winning a roughly comparable share of whites in South Carolina (he got about 40 percent of the non-Edwards white vote in both states, though reasonable people can disagree about what "roughly comparable" is) represents progress for him.
Or, put differently, you might have expected Obama's 40 percent share of the non-Edwards white vote in Nevada to translate into, say, a 20 or 30 percent share of that vote in South Carolina (again because of the poisonous racial dynamics of that state). That Obama got the same 40 percent share in South Carolina certainly exceeds my expectations and suggests he can do even better among whites in upcoming states.