Funding European Fabs

“Fab-sharing is in the ELG report,” said Bryant referring to the European Leaders Group set up by Kroes to implement her plan to restore European semiconductor manufacturing competitiveness.

“Four years ago, the idea was resisted,” said Bryant, “so there’s a big change. It makes huge commercial sense because it allows for larger, more efficient fabs.”

With so many of the Big Three’s products targeted at the same markets – power, microcontrollers, automotive, STB, MEMs, sensors, smartcard and security – a joint fab would allow what Bryant called “natural load balancing” – when one of the companies loses a slot, one of the others will, most likely, fill it. So the fab will not lose load.

“Missing from the ELG report is memory – 25% of the world market,” said Bryant, “most of the world’s leading fundamental memory research takes place at Imec. That could be commercialised.”

The concept of ‘Euro Memories’ has been discussed, said Bryant, adding that there has already been one attempt to raise funding for it which has failed.

So the problem remains: How does Europe fund fabs?

The ELG gives a hint: ‘By working with the member States and Regions, further leverage can be achieved to establish the critical mass to build on pilots through to full deployment and getting the necessary volumes for mainstream production,’ says the ELG report.

Last November, Jean-Marc Chery, COO of ST, called for support for early-stage fabs. This, he said, is the missing piece in the ELG jig-saw for reviving European chip manufacturing.

“Still missing,” said Chery, “is support for investment in first production capability,” asserting that support for the first production phase of manufacturing is “in line with revised state aid rules.”

“The next step is to move from support for R&D and pilot lines towards a focused investment package for R&D and first production via a dedicated IPCEI (Important Project Of Common European Interest),” said Chery, “we need another funding instrument besides ECSEL, EUREKA is the right framework for it.”

An initial production fab shared by the Big Three and maybe others as well, financed by public and private sources could be the starting point. There is a precedent for this which delivers a warning for Europe.

In 2005 the Japanese semiconductor industry considered a plan drawn up by MITI to set up a joint foundry company. A company, called the Advanced Process Semiconductor Foundry Planning Company, was set up to implement the plan.

In 2006, the CEO of the company, NEC’s Hirokazu Hashimoto, told me: “Japan has 11 integrated device manufacturers at the moment but their fab capacity is not big and they can’t compete with TSMC which has 60,000 wafer-per-month fabs.”

Hashimoto’s plan was to raise an initial $1 billion for a 10,000 wafer-per-month fab and then another $5 billion to extend it to 60,000 wafers-per-month.The Japanese government was right behind the plan.

The first implementation of the plan called for a 65nm fab to be built. However the companies said they had had settled their 65nm processes and design rules, and could not see a way in which variants of a 65nm processes could be effectively run in one fab

So a second implementation was proposed – that a 45nm fab should be built. Fujitsu, Renesas, NEC and Toshiba set out to define a standard process technology for 45nm and defining a standard to allow them easier access to each others’ IP.

Agreement among the companies would have substantial benefits, said the companies. “It will promote cross-use of IP and libraries, improve fab operating rates and facilitate large-scale capital investments among the companies.”

Agreement wasn’t reached. The last best chance at keeping the Japanese semiconductor industry as a significant world force was lost. Instead of revitalising its chip manufacturing base around a shared fab, the Japanese companies went into a long series of more or less disastrous mergers and jvs. From 50%+ market share in the late 1980s, Japan’s share slipped to under 14% last year

Hopefully, Europe’s companies won’t make the same mistake.