Where does the semi-conductor industry auction fab go?

Japanese semiconductor manufacturers finally awakened and began to face a reality - embracing the "fab-lite" strategy will be their best hope for survival.

But the big question is: Who would like to buy an old fab in the "clear-sell auction" of the Japanese semiconductor industry?

The answer is probably no one. Malcolm Penn, Chairman and Chief Executive Officer of Future Horizons, a market research organization, put it straight. He said that those Japanese fabs "are all using the old process," and whoever buys the fabs: "There will be management and The logistic nightmare."

But from another perspective, because your bet is on the side that no one is looking for, it means that the winnings that you may win will be doubled.

Recently there was a gossip in the streets of Tokyo, Japan, saying that overseas companies were planning to bid for a fab that a Japanese chip maker had tried to abandon. However, most people in the industry can't understand that there are really companies that are so bold - or so stupid - that they will do it.

Penn said: "If they are no longer good and they are useful for R&D, why buy them? From a market economy point of view of a pure textbook, it is better to let them go offline and disappear." But the point is: The cost of closing a fab is not small, including spending money to hire people to demolish the factory.

In view of this, it is not so far-fetched to allow an overseas chip maker to buy the giant monster at a discounted price and allow it to continue operating. Standing on the view of “a Japanese wafer maker”, inviting an overseas company to come in and use their fabs to manufacture products while meeting the needs of their existing customers would be better than shutting down the entire plant at a loss of two to three times.

Joanne Itow, director of Semico Research, another market research organization, said that the price of the old fabs will vary greatly depending on the size of the plant, the type of technology, and the products manufactured - but if you close down the cost of a fab About 300 million U.S. dollars, why not sell it at a price of 200 million U.S. dollars or even 100 million U.S. dollars?

Itow believes that the acquisition of an old fab in Japan is a potential business opportunity. In fact, there is a precedent: “Tower has acquired Micron's fab in Japan.” And she does not think that the first-line foundry will have Interest in operating wafer factories in Japan refers to pure wafer foundries that focus on specific technologies such as MEMS and power management, and are more likely to become buyers.

In addition, Itow pointed out that if you do not buy the entire wafer fab, first-tier foundries may be interested in acquiring second-hand equipment in those fabs in Japan, such as GlobalFoundries' recent acquisition of ProMOS (ProMOS). )'s fab equipment; she thinks this is possible: "With TSMC intending to expand CIS, HV and eFlash capacity, they may be involved in the used equipment market."

The above examples of Itow are all about the potential interest of pure wafer foundries for potential Japanese fabs; what about fabless chip companies from the United States? Will someone want to buy it?

The idea of ​​buying a fab is almost impossible at the moment when many chip makers choose to go to the fabless or fab business model—unless such transactions are attractive to buyers and sellers. This requires not only extraordinary imagination but also great courage for the buyer.

A few semiconductor manufacturers in Silicon Valley in the United States may occasionally say that fabless chip makers need to have unimpeded pipelines for fab capacity, especially if the company is developing new component structures or structures that cannot be controlled by the company. Conducting conceptual experiments, not to mention understanding how to build new architectures in the development process; and more importantly, foundries are often reluctant to cooperate with fabless chipmakers who have not yet signed major customers. .

So who can help those fabless chip makers test new concepts? In Silicon Valley, there are technology service companies like SVTC Technologies that can provide semiconductor product development and commercialization process technology; the company has wafer fabs in San Jose, California, and Austin, Texas, but has scaled back business last year.

In response to this, Itow said: “I doubt whether the small fabless makers will acquire fabs in Japan because the capital expenditure burden is not small.” She also pointed out that fabless companies: “There are still many products available Additional process service options for R&D and new component architecture testing, including the University of California at Berkeley, Arizona State University, etc., can provide prototype manufacturing services."

Future Horizon's Penn said that IMEC, a European research and development organization, also provides such services; however, Itow counters the claim that IMEC cannot be considered as a fab: "But they do serve as options for providing process development and product development services. "She said that there are also the CNSE, the University of Albany University of Science and Engineering (CNSE).

In addition to pure-wafer foundry manufacturers, MEMS manufacturers may also be interested in old fabs in Japan; Itow said: "MEMS manufacturers have started to use larger-size wafers and are shifting from 6 inches to 8 inches." MEMS supplier Kionix owns its own fab in Japan.

Itow pointed out that MEMS manufacturers may adopt a fabless business model to adopt a fab-fab strategy; Alpha Omega Semiconductor is one example. This company specializes in discrete/power chips. Last year, it mainly used Hua Hong NEC (HHNEC). Process: "In 2011, Alpha Omega acquired IDT's 16-year-old wafer factory in Oregon, and was listed as a fab wafer maker."

A survey of the fabs of Japanese semiconductor companies Fujitsu Semiconductor has transferred its plant in Iwate, Japan, to Denso Corp. last fall; Fujitsu is located in Aizu, Japan. The packaging production lines of Miyagi and Kyushu changed hands to J-Devices.

Last December, Fujitsu’s 12-inch wafer fab in Mie, Japan, still lacked interest. The company said that they are currently “thinking to resell it to foundries, including TSMC.”

At the same time, although Renesas has already announced the organization of reorganization, the company's semiconductor front-end manufacturing department has seven production sites and nine production lines; the latter-stage manufacturing department has two production sites.

Renesas manufactures Renesas High Components in the latter part of Japan's Aomori Prefecture and has already switched to Aoi Electronics in January; the other three are located in the post-manufacturing bases of Hakodate, Fukui, and Kumamoto respectively. It will be sold to J-Devices in early June.

Although Renesas has not yet indicated that the remaining production sites will be resold, closed or scaled down, earlier this year, the company stated that it plans to use the depreciation cost in the fiscal year ending in March 2014. And lease fees) decreased by 25% from the previous fiscal year.