Without detracting from the significance of Intel opening its first Chinese fab in Dalian, the company is certainly not new to offshoring. Andy Grove lamented the concept recently, but global competitiveness demands it—at least for commodity products.
Intel continues to innovate in spite of its maturity and size. Intel is systematic in its approach to looking beyond the next fiscal quarter or even the next year as it devotes resources to long term research and development. Engineers and scientists have more influence than the accountants in the company. I’m certainly not suggesting any lack of fiscal responsibility, though. In fact, the business savvy PhDs at Intel deliver value for shareholders without allowing the search for efficiencies and cost savings to deflect their focus away from the ultimate goal—pushing the frontiers of semiconductor technology. Otherwise, the Dalian fab would be just the latest Intel fab in Asia, not the first.
Intel media relations announced the opening of the Dalian fab in the same analyst conference as a multi-billion dollar investment in domestic production at four U.S. sites.
What I like about the approach is that Intel does not make commodity and leading edge products an either/or proposition. They stay competitive in the chipset game by building the China fab.
Considering the relative costs of U.S. versus Chinese production, their options were to either give up on this business or outsource the wafer production to TSMC. They did neither. What they also did not do was send anything near the leading edge of process technology beyond the confines of the 48 contiguous states. 65-nm is cutting edge for China, but it’s two generations behind current 32-nm production. The investments announced for two Arizona and two Oregon fabs target 22-nm and beyond.
Whether Intel management feels it is putting its intellectual property at risk in China, or if they will migrate more advanced processes to Dalian or future Chinese fabs are open questions. However, any doubt about their commitment to domestic production appears to be answered by the investment of several billion dollars to gear up four U.S. fabs for 22-nm production. The reported range of between $6 billion and $8 billion also includes the brand new D1X development fab in Oregon.
The situation is best summed up by the transition of Arizona’s Fab 12 from 65-nm directly to 22-nm. Meanwhile, those trailing edge 65-nm tools and recipes go to China.
Intel provided a couple of nuggets about the Dalian fab during an analyst call a couple of weeks ago that also spoke directly to the offshoring question:
Dalian is Intel's first newly sited fab since 1992
It is the company's first fab in Asia
Dalian offers 1.75 million square feet of Class 10 clean room
The outside air in Dalian approaches Class 3,000,000
Intel is showing it has the chops to manage state-of-the-art in parallel with legacy technology factories to produce commodity products. They should at least be congratulated for not following the herd.
I suppose Intel has no choice if it wants to remain cost competitive on the low end. Though it's a few billion dollars not spent in the US and it is creating jobs in China, the alternative is probably no investment at all and no jobs anywhere.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.