In other words: basic economics disagrees with your assertion.
Microsoft could invest 40-50k a year into people who qualify for retraining either in house or at local universities and still come out ahead.
You brought up the fact that there are 12 million workless, as a counter to Microsoft's desire to hire H1-B labor. If that's not to suggest they could fill those positions, what's the relevancy?
> Microsoft could invest 40-50k a year into people who qualify for retraining either in house or at local universities and still come out ahead.
Uhm, 40-50k a year is more expensive than full-time Harvard tuition. That does change the argument a little bit.
Whilst that may be okay if you're in your early 20s or from a country with very limited options, for an older person with greater skills living a relatively good life (e.g. a high-level contractor in South Africa) it's not particularly appealing as the cost-benefit analysis isn't very positive.
Ironically, these are probably precisely the people you want emigrating to your country - people with a proven track record who are ready to settle down and want to create a future for their immediate family, spending and investing everything they earned in the same country -- not young people who have nothing to lose and will probably send the bulk of their earnings back out of the country to their families in poorer nations, and who will potentially bring the rest of their (non-skilled) family in on family visa arrangements. (Of course this is somewhat a generalization, but I think on average it holds true.)
[1] http://en.wikipedia.org/wiki/H-1B_visa#Criticisms_of_the_pro...
This basically buys Microsoft six thousand wage slaves, in the very most literal sense of the term.