ChipWatch: The Fall And (Possible) Rise Of Western Chipmakers
By Matt Grimshaw
It's sometimes depressing to be in the US and working in the chip industry, well at least for the last 10 years at any rate. This week's news that IDT is shutting its Oregon fab and becoming a Fabless chip company is about as welcome as a chocolate bar at a Diet Club; I have a lot of friends that work in that fab.
Chip manufacturing in the US has been eroding and shifting to the Far East where government incentives and (initially at least) cheaper labour costs have attracted almost all the new Fabrication Plants. There is of course another reason for this shift - the Fabless/Foundry model which was born in Taiwan.
To those not versed in this concept it's about fairly simple economics: Fabrication plants are horrendously expensive propositions, with the latest cutting edge fabs being in the region of $4.2bn US. That's more then the gross domestic product of a number of countries.
With the Fabless model, the focus on manufacturing is removed. It makes sense for one big company to make the chips, while other companies design them, and then rent the facilities to build them on a common production line.
This has become the dominant business model in the chip industry over the last 15 years where now more than 50% of all chips are manufactured in Foundries like Taiwan's TSMC (Taiwan Semiconductor Manufacturing Corp), the oldest and largest foundry.
The US and Europe has not had a serious pure play foundry entrant in this market since GlobalFoundries, which was recently born from the manufacturing division of Advanced Micro Devices (AMD) when it too became a fabless chip company. It was Intel's brute strength that was a factor in this decision as AMD simply couldn't keep up with it's rival's R&D spending.
At first, I have to say there was much skepticism about how successful the GlobalFoundries venture would be - it looked to as if AMD was jettisoning its manufacturing liability.
We'd already seen two manufacturing spin-offs from once successful companies eat dirt: Spansion Semiconductor, the one time Flash joint venture of AMD and Fujitsu; and Qimonda which was once the DRAM division of Infineon Technologies (itself the old Siemens Semiconductor).
Certainly, GlobalFoundries would ensure AMD's survival against it's much more powerful rival Intel, but was a US/Europe based foundry company viable? Would they bomb in a similar manner to Spansion and Qimonda?
Where would they get their customers from when competing with such a successful monster as TSMC, let alone all the others like UMC, Chartered Semiconductor, and an army of smaller players?
Well a lot of skeptics got their faces slapped by the wet kipper of incredulity the other week with the announcement of GlobalFoundries' first 'proper' customer in ST Microelectronics.
With this announcement the cat is really amongst the pigeons and this time the cat seems to have a Taser and means business.
With GlobalFoundries having leading edge fabs in Dresden Germany and the soon to be constructed 300mm fab in New York; expertise in leading edge manufacturing; and financial backing from Abu Dhabi's Advanced Technology Investment Co. (i.e. a near bottomless pit of cash); there really is a new player in town that appears to be not only talking the talk, but also walking it's walk with a certain hint of swagger too.
The question is: does GlobalFoundries signal the re-birth of chip manufacturing in Europe & the US? I'm certainly looking forward to finding out.
[Matt Grimshaw is the Editorial Director of the Semiconductor Technical Journal; Future Fab International (www.future-fab.com).]