Writing in the San Francisco Chronicle, Tom Abate compares the fabless-foundry model in the semiconductor industry to the division between content creators and printing presses in the publishing business. It's a good analogy, even if I don't quite see why you would want to compare two industries when the general public doesn't understand either of them. Better to describe the semiconductor industry more clearly and leave publishing out of it.
Anyway, his underlying point is that the movement of semiconductor manufacturing out of the United States risks "hollowing out" the US semiconductor industry. Where manufacturing goes, he argues, R&D and the rest of it eventually follows. Unfortunately, SIA president George Scalise's suggestion that tax incentives for US fabs are the solution completely misses the underlying dynamics. As Big Steel learned a generation ago, protectionist tactics may slow global trends, but can't stop them.
A better way to look at it might be to realize that profit and growth are more important than control of specific market sectors. American metallurgists began to come back when they turned to specialty metals, like titanium. These metals are more difficult to work with than steel, so companies could differentiate themselves by their capabilities rather than their costs.
Similarly, the foundry model works to the extent that semiconductor manufacturing is commoditized, and commodity markets inherently favor the lowest cost producer. Rather than struggling to hold on to such a business, the US semiconductor industry might be better off looking for areas where technical capability is, so far, inadequate. Don't think of it as abandoning manufacturing, think of it as akin to Intel's transformation from a marginal DRAM company to an enormously successful microprocessor company.
(I keep meaning to write about these ideas for the Back Story section. That, like much else, is on hold pending the move.)