Over the past few years, we've seen the foundry business evolve from a single-horse race that TSMC effectively "won" each and every cycle to a two-way competition with Samsung. Now, GlobalFoundries is making a serious push of its own with early 7nm technology availability and volume production currently planned for the back half of 2018.
That's the news(Opens in a new window) from Saratoga today, where GF announced that its 7nm LP node (LP = Leading Performance) is ready for partners to begin planning their designs. The first customer launches on 7nm LP are expected 12-18 months from now, and GF is promising that it can deliver up to 40 percent improved performance compared with 14nm. The company claims that its 7nm work is exceeding its performance and power targets, and is on track to deliver up to 2x area scaling compared with previous 14nm technology. The company has also been hard at work on 5nm in partnership with IBM.
"Our 7nm FinFET technology development is on track and we are seeing strong customer traction, with multiple product tapeouts planned in 2018," said Gregg Bartlett, senior vice president of the CMOS Business Unit at GF. "And, while driving to commercialize 7nm, we are actively developing next-generation technologies at 5nm and beyond to ensure our customers have access to a world-class roadmap at the leading edge."
GlobalFoundries will also have the option to integrate some EUV manufacturing at this node, if its customers want to use it, but we'd be surprised if many do. EUV may be slowly moving into commercial production, but it isn't expected to see broad availability for the next few years and may be reserved for critical masks where other forms of multi-patterning can't be used for some time after that.
What's interesting about GF's announcement is that it's also claiming up to 60 percent reduced power with 40 percent improved performance, using the word and. That didn't used to be significant, but in recent years it's become common to see semiconductor manufacturers using or instead. You can have higher performance at the same power or you get lower power at the same performance, but delivering both simultaneously has become increasingly difficult as the benefits of each successive process node grow smaller.
GF also seems to be leading with a process node that's better suited to higher power silicon designs like GPUs and CPUs, rather than the mobile-first approach TSMC has taken to better suit the needs of Qualcomm and Apple. There's nothing intrinsically wrong with each option, but it reflects how different companies now drive the semiconductor market compared with a decade ago, when GPUs were major drivers of each new node. Now, we see companies like AMD and NV sitting out some process nodes that are explicitly designed for mobile SoCs, and taking a staggered approach to process node deployment.
When GlobalFoundries first spun off from AMD, it had a great deal of difficulty finding its sea legs. The company's original roadmap called for aggressive competition with TSMC across a range of nodes, but it was unable to deliver on that vision. Later, it was forced to license Samsung's 14nm technology due to problems with its own 14nm XM. GF seems to be on much more stable footing now, and is pushing hard to turn the two-way race between TSMC and Samsung into a three-way competition with itself in the lead.
