First, note that I didn't say redistricting is likely to help the GOP (though that may also be true). Until and unless we know more about who controls the statehouses in 2010, and where the population shifts occur, it's impossible to predict which areas will gain or lose seats. That said, the overall shift will clearly be from Blue States to Red States (at least at the presidential level).
National Journal has the story:
Newly released Census Bureau population estimates show that as many as 13 House seats may shift across the country following the official 2010 Census, according to an analysis by the Washington-based Election Data Services...
The shift of as many as 13 seats for the scheduled 2012 reapportionment is based on several projections by Election Data Services that assume a continuation of current trends, though at various paces. The contrasts between what EDS president Kimball Brace terms "long-term," "mid-term," and "short-term" shifts are based on the current recession, the housing market downturn and related factors...
Under Brace's "short-term" model, 12 districts would shift among 19 states. The winners would be: four to Texas, two to Arizona, and one each to Florida, Georgia, Nevada, Oregon, South Carolina and Utah. Those districts would come from the following states: two from Ohio, and one each from Illinois, Iowa, Louisiana, Massachusetts, Michigan, Minnesota, Missouri, New Jersey, New York and Pennsylvania.
Brace's more robust "mid-term" model would add one district to North Carolina and remove it from California -- which has gained at least one district in every decennial reapportionment since it joined the Union in 1850.
Obviously the shift is not exclusively from Blue to Red, but I count a shift of +8 or +9 electoral votes from reliably blue to reliably red states (counting North Carolina, Texas, Arizona, Florida, Georgia, Nevada, South Carolina and Utah as reliably red). That wouldn't have been enough to put John McCain in the White House, but it may help the GOP in 2012.