"Intel's first preference is to always fill the fabs with their own products," says Moorhead. "Doing so, they're basically double-dipping compared to something like TSMC, where TSMC makes the manufacturing margins and Qualcomm makes the design margins. In-house, Intel captures both of those."
Intel still needs to manufacture hundreds of millions of processors to meet the demand for basic PCs and laptops, even in these glum times. And if Intel ever (hopefully, eventually) becomes competitive in the mobile market -- much less the Internet of Things market -- it will need big volumes there, too. Big volumes mean full fabs.
An expanded custom chip business isn't out of the question, though. Processors-to-order can keep the fabs a-flowing until Intel strikes the mobile mother lode.
Intel does bring unparalleled processor design technology to the table. Then, too, some of its fabs are running slim these days, and the custom chips they've designed thus far have all been of the high-margin variety. Plus, the impending move to larger 450mm wafers means that existing fabs will have even higher capacities in the future.
"If they don't feel they can fill their fabs [with in-house work], that's the time for Intel to do a very broad foundry business," Moorhead says. Dean McCarron, principal analyst at Mercury Research, told PCWorld that eventually 20 percent of the chips coming out of Intel fabs will likely be contract work, no doubt of the high-margin variety.
So, we might see Intel take on more custom foundry customers here and there, but it's basically a side gig -- albeit potentially a billion-dollar side gig.
But hey, it's always good to have a Plan B. Intel has to pay for that $18 billion in R&D costs somehow.