While I agree with most of your statement, the OP's original point was that the profit should yield a better return than can be gotten with other projects. For example, if the XBOX project (could) realise(s) an ROI of 3% and just putting the money in some sort of savings account gives you a net return of 4%, then you should pick the savings account, especially if there is less risk involved in other possibilities to use that money.
Of course nobody is certain about how risky the console market is, nor does anyone know how big the return on investement will be in a few years, but the man has a point if he says that just throwing money at something isn't really the best way to invest. It seems to me that if you need 10 more "halo-size" hits, somebody didn't really do his or her homework properly.
On the other hand, one shouldn't account for sunk costs when calculating current future profits, which supports your point of "not closing down a now profitable division".
Just keep in mind that the profit of that division (and its possible growth scenarios) might be lower and more riskier than some other investments.
To end, I'd like to point out that Microsoft "sticking to their knitting" might be seen as a way to lower the risk involved, thus benefting the shareholder once more.
I think you both have a point, and that, like always, the truth is probably somewhere in between.
