In the wake of AMD’s decision to take its 7nm business to TSMC, GlobalFoundries has announced some radical restructuring plans of its own. The company is killing its 7nm deployment and retrenching around both 12/14nm and its investments in product offerings like FDX. According to GF’s press release, this major change in direction took place after Thomas Caulfield assumed the CEO position earlier this year.
This is a significant departure from GF’s previous statements on the topic and its own manufacturing plans. The company is also spinning off its ASIC business as a wholly-owned subsidiary that will operate independently of the foundry business. GF writes:
In addition, to better leverage GF’s strong heritage and significant investments in ASIC design and IP, the company is establishing its ASIC business as a wholly-owned subsidiary, independent from the foundry business. A relevant ASIC business requires continued access to leading-edge technology. This independent ASIC entity will provide clients with access to alternative foundry options at 7nm and beyond, while allowing the ASIC business to engage with a broader set of clients, especially the growing number of systems companies that need ASIC capabilities and more manufacturing scale than GF can provide alone.
I’ll admit to a lack of knowledge when it comes to the specifics of how ASIC design teams work with foundries, but announcing that you’re spinning off a business unit dedicated to helping your existing and future customers leverage the design capabilities of other foundries doesn’t seem like a positive announcement. GF refers to providing clients with access to alternative foundries, implying that this service is primarily being provided to one or more customers that intended to use its 7nm node and now won’t be able to. But even if that’s the case, you’d think this kind of offering would be limited to the impacted customers and not spun off as its own business. At the same time, keeping the unit as a wholly-owned subsidiary encourages customers to think of it as an extension of GF as opposed to a completely separate business — and that’s just weird when you consider that one of the most logical purposes of a foundry’s wholly-owned subsidiary would be to funnel business to that foundry.
We reached out to GlobalFoundries for additional comment. The company told us: “The value of the ASIC business is in the IP and design services it offers to customers. The new ASIC subsidiary will still be offering IP and design platforms on GF’s 14nm and above technologies, and we expect they will create ASIC offerings in more advanced technologies with other technology and manufacturing partners.”
And Then There Were Three
Ever since the 28nm node transition, there have been four foundries on the leading edge — Samsung, TSMC, GlobalFoundries, and Intel. GF has struggled the most with its place as a leading-edge fab, but the company appeared to be righting itself. Its 14nm may have been licensed from Samsung, but GF still had to implement the node in its own factories. The success of AMD’s Ryzen and Polaris GPU families is evidence that the foundry was able to achieve acceptable yields and then improve its own process enough to deliver significant generational improvements to AMD’s second-generation Ryzen CPUs.
But with today’s announcement, that’s (mostly) over. You can, I think, make a valid argument that FD-SOI and the 22FDX and 12FDX deployments are a different type of technology, and one where GF is leading, but it’s also leading by virtue of being the only player on the field. And so far, the applications are fairly niche. As a private company, GlobalFoundries is not required to make public quarterly statements, but the company recently announced that it had more than $2B in design revenue with FD-SOI. For comparison, TSMC’s yearly revenue for 2017 was over $32B. None of this is to imply that GlobalFoundries work on 22FDX or 12FDX (which isn’t ready yet) is meaningless — in fact, given projections for IoT sales, the foundry’s early position in FD-SOI could be quite lucrative in the future — but it’s not a drop-in replacement for the kind of revenue TSMC and Samsung have earned on FinFETs.
That’s probably why GF is making this announcement now, while AMD’s sales of 14 and 12nm products are still robust, and the company has time to wind down its 7nm investment and presumably decide what else to do with its foundry space that was previously dedicated to the ramp. After nine years of trying to compete against the entrenched players of the industry, GF has decided to chart its own course and focus on alternatives.
It’s not clear what this means for IBM’s POWER roadmap, but presumably any future POWER10 and POWER11 CPUs will be built at either Samsung or TMSC. Given that GlobalFoundries bought IBM’s entire foundry business and had pledged to continue working with the company, this twist undoubtedly impacts IBM as well. It’s entirely possible that IBM is one of the clients that GF’s new ASIC business unit will be assisting in porting its projects to another foundry.
GlobalFoundries to Soon Pilot 7nm and EUV Production
GlobalFoundries is pushing ahead with a plan to pilot EUV and 7nm production. We recently got a chance to sit down with the company and talk about its plans for the future.
AMD Needs More 7nm Capacity Than GlobalFoundries Can Provide
AMD is sourcing 7nm from TSMC and GlobalFoundries because it expects to have more need for 7nm volume than GF can supply.
AMD Will Fab Its 7nm ‘Rome’ Epyc CPUs at TSMC, Not GlobalFoundries
AMD is building its 7nm Epyc CPUs, codenamed "Rome," at TSMC — not GlobalFoundries.
GlobalFoundries Departing the Leading Edge is an Ominous Sign for Foundry Industry
GlobalFoundries' departure from the 7nm node isn't a great sign for the overall ability of the foundry industry to navigate these transitions.