When a stock market index suddenly trends downward, or house prices plummet, it is very easy to focus on the sellers and talk about a sell-off, or people "getting out" of equities, commodities, property etc. But in every sales transaction there is a buyer.
This is one reason why the sale of a Hillsboro, Oregon wafer fab for $26 million offers an interesting microcosm of the global semiconductor industry.
With this move the vendor Integrated Device Technology Inc. moves from being a fab-lite chip company to being fabless. At the same time the buyer, Alpha & Omega Semiconductor Ltd., moves from being fabless to being fab-lite, as it boot-straps itself into a more prominent place in the industry.
So what is the difference between the two companies?
IDT was founded in 1980, by solid-state design process engineers mainly from Hewlett Packard, and fundamentally linked into Silicon Valley. The company's initial push was to develop high performance circuits on the then emerging CMOS process variant. The company developed its own proprietary CMOS process and chose military-grade SRAMs as its market entry vehicle and quite quickly moved up to offering high-speed RISC processors. In the absence of a mature foundry industry at that time manufacturing was a fundamental part of IDT's early value proposition.
However, as Malcolm Penn, chief analyst with Future Horizon's would say, fabless is usually the eventual conclusion of the adoption of a fab-lite strategy and IDT is now reaching that point.
But A&O is going in the other direction. A&O was founded in 2000 as a fabless company with its origins in China, although it is now officially registered in Hamilton, Bermuda. The company stalled a Hong Kong stock exchange listing in 2008 but went public on Nasdaq in 2010 and has its main office in Sunnyvale, California. The company made a net income of $37.8 million on sales of $361.3 million in the year to June 30, 2011.
So who is the smarter: A&O which is spending $26 million to acquire a 200-mm wafer fab so that it is not always blown about by the whims of its supplying foundries, or IDT who is gaining $26 million while letting go the responsibility of feeding a fab?
At one level the answer depends on the detail of the fab, its equipment, staffing and on the price. But on another it reflects a philosophical position as to whether controlling the means of production is an inherently vital part of business.
I have had a go at answering your two questions but it was too long to post here.
I have put it under its own heading http://www.eetimes.com/electronics-news/4219627/Why-the-British-got-out-of-fabs
Would Peter Clarke rather answer why the British had to get out of Fabs and end up just with design / IP a la ARM ?
Why Fabs still survive ( if not exactly thrive ) in the UK's traditional rival France ?
Globalization seems to have reduced the US to the status of the UK
And the answer is...the two companies are in very different areas of the semiconductor industry. IDT, as its name implies, is in the functional integration business. These days, the manufacturing technologies necessary for IDT's products are readily available from wafer foundries. A&O, on the other hand, is in the specialty device business - high power transistors and diodes, power converters, etc. The semiconductor processes necessary for these kinds of products are not as readily available from foundries - because in many cases these kinds of devices are for lower volume applications that don't demand large wafer volumes. And foundry pricing for these kinds of wafers is by no means set at commodity levels, as the foundries offering these technologies typically do so on an opportunistic basis.
This deal wasn't about IDT dumping a fab to A&O for pocket lint. IDT's business takes it towards fabless, while as A&O continues to grow it needs more control over its technology and cost model.
Well, contrary to what the title indicates, this article did not answer why the two companies took different paths, but simply pointed out the difference, and left the "why" to the "detail of the fab,..."
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.