I've talked recently about the practice of offshoring parts of a business as a mechanism intended to reduce expense rate. (See Business for Engineers: Offshore Outsourcing.) There are pros and cons to moving part, or all, of a business unit's "non-essential" skills offshore. But the huge emphasis on offshoring manufacturing is often taken for granted without fully understanding the consequences or fully understanding how this is a strategic decision that will have a long-term impact on the company and on the industry as a whole.
Now it's time for some confession...
My career has been focused mostly on the semiconductor industry, with occasional forays into related businesses. When I started in the semiconductor industry, the company that I worked for, American Microsystems, was about a $110 million business -- about the same size as Intel's operation at that time. Semiconductor capital expenses were manageable for most companies, clean-room requirements weren't too strict, nor was vibration a major concern.
Jump ahead to 1989 and a job with Analog Devices Inc. as director of DSP Products. ADI had good process development, but lacked adequate manufacturing capacity for a potentially large-volume digital product. We had manufacturing facilities in Massachusetts and Ireland, but they were dedicated to mostly linear processes.
Expanding manufacturing capacity in-house was out of the question for both time and capital cost reasons. Thus began a search for manufacturing capacity for a soon-to-be-released line of DSPs. You may have determined that I'm not a big fan of offshoring core technology, regardless of the country of origin. The reasons are both tactical and strategic. Tactically, proximity reduces the complexity and cost of managing a more geographically dispersed organization. The strategic reasons are complex, but can be boiled down to controlling one's own destiny.
The search for a contract manufacturer began with US-based companies. ADI already had a foundry relationship with a Chinese company for moderately complex, low-volume digital products, but the logistics of managing an offshore foundry was a greater complexity than I wanted to manage.
The criteria for selecting a foundry included the following:
- We had a preference for a US company.
- The foundry needed to have sufficient capacity such that the new products would not consumer more than 10% of its capacity in the first three to four years of our production.
- A true 1 micron process with a near-term migration to 0.8 micron CMOS was required.
- Since the processor was dominated in performance by logic, we needed a good logic process with competitive SRAM layouts.
- We wanted a foundry that would put a special double poly process in place for our voice-band sigma-delta converters.
The search for a US factory went from Armonk, N.Y., to Santa Clara, Calif., and locations in-between. I felt as if I was reliving the children's story of "Goldilocks and the Three Bears." The New York company made devices with cross sections that looked as if they came out of a semiconductor process textbook, with technical illustrations drawn by artists. Other companies didn't have an adequate process. To further complicate the selection, Texas Instruments had the ideal process but was a head-on competitor with a comparable part, which left it out of the picture.
So what did we decide on? Our first try was to combine Orbit Semiconductor for prototype runs and initial volumes with a very large Santa Clara microprocessor company as our production facility. The choice was problematic as the processor company's demand fell and rose; foundry services were not its main business, and we had some significant variability in production schedules. The production uncertainty doomed this relationship. Re-enter the offshore option.
I had a good friend, John Luke, who was president of TSMC America. John and I sat down and structured a business relationship that gave us access to TSMC's 1 and 0.8 micron processes immediately and allowed TSMC to install our double poly process. The relationship worked well, in large part because ADI engaged TSMC at a strategic time for both companies. TSMC's business was "foundry services," and it did not have products of its own that might compete for capacity. Most of all, my personal relationship with John enabled the other members of my team to develop close working relationships with their counterparts at TSMC.
At the end of it all, ADI hit all but one of our requirements -- US manufacturing was not in the cards.
So, I've been on both sides of the outsourcing and offshoring debate. But I will tell you that having manufacturing a half a world away definitely complicated management and added overhead and inefficiency.
In our case, the decision to offshore was almost a foregone conclusion due to the lack of non-conflicted onshore manufacturing. To this day, however, even though ADI and TSMC proved to be a great fit, I wish the decision could have been made differently.