GLOBAL FOUNDRIES HAS finally announced customer #2, AMD being customer #1. While this is significant, there is a hidden gem in the press release, an unannounced new process node.
First the headlines, GloFo has announced STMicroelectronics as their first customer. Unlike most of the fabless houses that foundries like GloFo usually targets, ST has several fabs on relatively recent process nodes.
The movement to GloFo signals that ST believes spending billions on fab development is not a sane way forward for the company. Spending that many billions of dollars on a bleeding edge fab that they are unlikely to fill doesn’t seem like a good use of money, so they are doing the smart thing and outsourcing. It is better to pay a slight premium for an effective small slice of a fab than a huge premium for empty space and duplicated R&D.
While ST will make the finance wonks all giggly with anticipation, the geeks will be much more interested in the new process called 40LP. It is something that wasn’t on the roadmaps when we talked to them here.
At Semicon a few weeks ago, the 40LP process, for 40nm, low power bulk, was strongly rumored to be in the cards. The idea is that the competition, lets just call them xSMC to hide their real identity, hasn’t exactly been inspiring confidence with their 40nm process, even if things are back on track now.
Global Foundries seems to have noticed that the vast majority of potential customers are wanting 40nm parts soon, and 28nm ones after that. 32nm for some reason seems to be a non-starter in the foundry business. Noticing that, GloFo seems to have lessened emphasis on 32 while pumping up 28 and adding a 40nm node. If someone will pay you for it, why not?
Given that GloFo has a fully baked and high volume 45nm, albeit SOI, and access to other 45nm bulk processes from the IBM alliance, the 40LP process doesn’t seem to be much of a stretch. If they are actually going to de-emphasize the 32LP/32 bulk processes, it could actually speed up 28nm development.
In the end, ST moving to GloFo is a clear win for the new foundry. It shows that big semi houses have enough confidence in them already to trust the family jewels to a ‘startup’, but also brings in cash. What it isn’t is a huge loss for TSMC. ST is an oddity, since they have fabs of their own, they are essentially a new customer for foundries, additive to the sector. GloFo, by being nimble and thinking on their feet, won the business. There are surely more to come.S|A
Latest posts by Charlie Demerjian (see all)
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018
- Intel can’t supply 14nm Xeons, HPE directly recommends AMD Epyc - Sep 7, 2018
- AMD reintroduces the Athlon name with two CPUs - Sep 6, 2018