The Powerlessness Of Fablessness.
Interesting that Globalfoundries is to manufacture for all-comers, ICs made on the ST-developed FD-SOI process.
It seems that STMicroelectronics and ST-Ericsson have developed both 28nm and 20nm FD-SOI processes.
28nm FD-SOI is available in limited quantity production from the sub-20,000 wpm capacity at ST’s fab at Crolles, but clearly a volume second source was needed if the technology was to enter their product line in a meaningful way.
IBM is thought to be pretty full, and Globalfoundries has stepped into the breach, but has apparently insisted on having the right to use the ST process to make anyone’s chips on them..
This is not, of course, as good for ST and ST-E as having sole access to a proprietary, and maybe superior process, of their own.
Unlike Intel who can claim its 22nm finfet technology gives them a proprietary advantage, ST now has to share its advantage with anyone who wants to use it.
If ST had a high volume proprietary source of leading edge process technology, then putting FD-SOI wafers down it could have delivered a vital competitive advantage. The technology is claimed to deliver a 35% power improvement.
But ST has not pursued the 300mm fab building strategy which was in place in 2005 when there was a transition in the company’s leadership.
On the other hand, ST and ST-E are undoubtedly further down the line than anyone else in developing FD-SOI wireless chips and so should have an advantage for a while, but only for a while. ST itself points out that it is a straightforward matter to transfer designs from bulk CMOS to FD-SOI. Rivals, presumably, will also find it straightforward.
So, at the end of the day, it shows that process matters, fab matters and capacity matters.
Without them you lose control over your destiny.