GLOBALFOUNDRIES Announces "First" New Partner
Subject: General Tech | July 29, 2009 - 04:00 AM | Josh Walrath
While the current (and only) client of GLOBALFOUNDRIES is AMD, they have finally officially announced their first 3rd party partner for production. STMicroelectronics is choosing GF to be a partner for 40 nm low power devices, which can possibly power future cell phones. This is an exciting development for GF, as many around the industry were not entirely certain they could pull off being a successful foundry. Well, certainly one new client will not bring in billions of dollars of revenue, but it is a beginning. And an important one, considering that GF is still nowhere near providing a product line for other clients than AMD.
The 40 nm low power product line is interesting in that it will not be available until late 2010. Currently GF is producing parts comprised 65 nm SOI and 45 nm SOI technologies, but that is the extent of their lines. Oddly enough, 32 nm SOI and 32 nm bulk will be available before the 40 nm bulk and low power products. This may seem a bit backwards, but when looking at the industry, not every company needs to have a cutting edge process for their product designs. In this case, it appears as though the chip family that STM has in mind for production at GF will be pretty small, and going to a smaller process node could prove to be problematic to packaging said part (not to mention the redesign aspects of going from 40 nm LP down to 32 nm SOI or bulk.
Hopefully for the industry and consumers alike, this will be the first of many such announcements from GF. Having more players in the foundry field, and especially players which have a proven track record of advanced process tech development, will benefit both consumers and designers alike. This means faster rollouts of advanced product lines, more cost effective and performant parts in a timely manner, and competition for the hard earned dollars from designers who need a partner to actually produce their ASICs.
Currently GF is not talking about other companies who could be jumping on the bandwagon. We do have a short list though. It is no secret that the graphics division of AMD will be going with GF once the 32 nm bulk process is open to manufacturing. NVIDIA of course is likely to jump on board, as they cannot afford to be at a manufacturing disadvantage to AMD/ATI. For NVIDIA, changing foundries is not a big deal. When things look better at another place, they have jumped ship before. IBM was a partner of theirs for the GeForce FX 5700 and 5900 series of cards. UMC is a consistent partner for lower end models. If GF's 32 nm bulk process is clean, then the obvious interest is graphics, but the second big interest would be getting Tegra onto 32 nm ASAP. NVIDIA still believes that Tegra is the breakout part for the company to hit the real big time, and if they can get out a good, fast, feature packed Tegra design at 32 nm and take advantage of the size and power features of that process, then they are poised to offer a cutting edge part to phone and MID manufacturers that should be leading edge.
TSMC is not taking this lying down. They have significantly boosted their R&D staff, and are making a huge effort to not only fix their ailing 40 nm process, but also aggressively pursue 32 nm nodes and beyond. Time will tell if they actually succeed, but I personally think that GF has a big leg up on the competition due to their process war with Intel at the high end of manufacturing technology. If GF can get out their 32 nm bulk process as promised in 1H 2010, then they will have customers streaming to them. Fab 1 Module 1/2 are big fabs, and they can put out a lot of wafers a month. Still, if a lot of designers come to them for parts, they could hit a wall in terms of how much they can actually handle. It will be interesting to investigate this again once GF has both 32 nm SOI and bulk going at the same time next Spring.