GlobalFoundries drops push for next-gen 7nm semiconductor tech

Posted:
in General Discussion edited August 2018
California chip firm announces a halt to its development of 7-nanometer fabrication technology, leaving Apple partner TSMC and rival Samsung as the only remaining players in the space.

GlobalFoundries headquarters


GlobalFoundries announced Monday that it is backing away from its ambitious plans to develop next-generation semiconductor technology, and will instead refocus its investment on existing technologies.

According to AnandTech, the shift means GlobalFoundries will drop five percent of its staff. The move also means the only remaining players in the next-generation semiconductor space are a pair of current and former Apple suppliers, TSMC and Samsung, respectively, and that the consumer electronics industry as a whole whole may very well find itself more dependent on those companies.

"Demand for semiconductors has never been higher, and clients are asking us to play an ever-increasing role in enabling tomorrow's technology innovations," GlobalFoundries CEO Tom Caulfield said in a statement released by the company. "The vast majority of today's fabless customers are looking to get more value out of each technology generation to leverage the substantial investments required to design into each technology node. Essentially, these nodes are transitioning to design platforms serving multiple waves of applications, giving each node greater longevity. This industry dynamic has resulted in fewer fabless clients designing into the outer limits of Moore's Law. We are shifting our resources and focus by doubling down on our investments in differentiated technologies across our entire portfolio that are most relevant to our clients in growing market segments."

GlobalFoundries and Samsung have a joint technology agreement, though it appears the deal will not extend to the fabrication of 7nm chips.

While a difficult feat, shrinking die size increases the density of a given chip to fit more transistors into a smaller space, thereby presenting a number of performance and efficiency improvements. For consumers, benefits typically include a reduction in power usage that can help increase battery life, lower heat production, and a boost to processing capacity.

AMD announced that it would move its 7nm production to TSMC, Reuters reported Monday.

In 2013, Apple reportedly tapped GlobalFoundries to fabricate what would become its A9 system-on-chip for iPhone and iPad, but backed away from doing so at the last minute, in favor of TSMC.

GlobalFoundries' decision comes months after TSMC ramped up production of its own 7nm chips, a healthy portion of which are likely bound for iPhone and iPad. As a fabless manufacturer, Apple designs its A-series silicon and outsources production to firms with foundries, such as TSMC.

Samsung was an original fab partner, but lost exclusivity starting with the A8, which for reference was built using a 20nm process. TSMC has been Apple's lone A-series chip fabricator from the 16nm A10 Fusion forward.
«1

Comments

  • Reply 1 of 26
    Shame. Competition is healthy despite what some here think. 
    chasmavon b71983
  • Reply 2 of 26
    I think he's saying that they are not moving forward with cutting edge technology so they can better invest in cutting edge technology...OK. I'm not a chip expert so maybe what he said makes sense. It seems a little to 'spiny' to take seriously. 
    1983
  • Reply 3 of 26
    I think he's saying that they are not moving forward with cutting edge technology so they can better invest in cutting edge technology...OK. I'm not a chip expert so maybe what he said makes sense. It seems a little to 'spiny' to take seriously. 
    There aren’t too many ways to effectively spin “we can afford neither to develop an expensive next-gen process node nor to build the fab capacity required to make that investment ever turn a profit.”  

    Given their resources, it’s probably the best road to profitability, for them. Developing chips for fabrication at 7nm (and beyond) is extremely expensive, and most customers won’t be going there anytime soon. I suspect they’ll re-visit 7nm three to five years from now when it’s less expensive for them to implement, and for their customers to utilize. 
    edited August 2018 radarthekat1983watto_cobra
  • Reply 4 of 26
    netroxnetrox Posts: 699member
    It looks like we're already meeting the Moore's Law limit. It's time for them to figure out innovative solutions.
  • Reply 5 of 26
    nunzynunzy Posts: 662member
    Is this good for Apple?
  • Reply 6 of 26
    Well, I can't say I blame them. There are only so many customers for the latest process. The majority of chips used today are on older, more mature processes. There's no need for a small microcontroller (found in everything including your toaster) to be made on 7nm when even at 40nm it's going to be ridiculously small and power efficient already.
    PickUrPoison1983watto_cobra
  • Reply 7 of 26
    hentaiboy said:
    Shame. Competition is healthy despite what some here think. 
    What are you talking about? Appleinsider & its readers are not stupid. Competition is always healthy. 

    Like Steve Jobs said ‘For Apple to succeed, Microsoft does not have to lose’ 
    edited August 2018 bb-15
  • Reply 8 of 26
    JinTechJinTech Posts: 323member
    I think he's saying that they are not moving forward with cutting edge technology so they can better invest in cutting edge technology...OK. I'm not a chip expert so maybe what he said makes sense. It seems a little to 'spiny' to take seriously. 
    Sounds to me like they are trying to get bought out.
  • Reply 9 of 26
    polymniapolymnia Posts: 895member
    hentaiboy said:
    Shame. Competition is healthy despite what some here think. 
    What are you talking about? Appleinsider & its readers are not stupid. Competition is always healthy. 

    Like Steve Jobs said ‘For Apple to succeed, Microsoft does not have to lose’ 
    Will you bring this point up after the next Apple vs. Sonos article?
    1983[Deleted User]
  • Reply 10 of 26
    Rayz2016Rayz2016 Posts: 4,556member
    hentaiboy said:
    Shame. Competition is healthy despite what some here think. 
    If you’re talking about Nunzy then I’m afraid you’ve been had. He’s just here to give the community a bad name. 
    nunzylkruppStrangeDayswatto_cobrajony0
  • Reply 11 of 26
    May be we have to start to look at the innovation in the chip market in a different way. Shrinking have limits, and the current trend show that shrinking is slowing down, even for big pocket actors. Innovation in high end chips will be very very different in 5-10 years.
  • Reply 12 of 26
    Rayz2016 said:
    hentaiboy said:
    Shame. Competition is healthy despite what some here think. 
    If you’re talking about Nunzy then I’m afraid you’ve been had. He’s just here to give the community a bad name. 
    Not just Nunzy. Just look for comments by many other people on Samsung/Google/Qualcomm etc. Having said that, this forum is far better than many other forums, so it is not as bad as it might look like.
    [Deleted User]gatorguynunzy
  • Reply 13 of 26
     Sure, for now there may be ways to innovate and stay relevant without reducing size.

    But my guess is that eventually this will catch up with them.  Given how difficult the 7 nm process is, you can’t just show up a few years down the road and suddenly decide you want to participate after all...
    edited August 2018 bonobobwatto_cobra
  • Reply 14 of 26
    blastdoorblastdoor Posts: 1,888member
    nunzy said:
    Is this good for Apple?
    I love your posts :-)

    but I think there is a serious answer to that question.

    The cost of new process nodes is exploding. We will soon be at a point where new nodes yield more expensive transistors, not less expensive. The only companies that will pay for chips fabbed on these new nodes are those that highly value performance per watt and have very deep pockets. I just described Apple.

    so in a way, this is possibly good for Apple in the sense that it is bad for smaller companies that can’t afford to fund massive investments in capital and R&D. 

    But this is clearly bad for consumers — it means less competition. 
    nunzywatto_cobrajony0
  • Reply 15 of 26
    In other words our 7nm process isn’t good enough for the customers that require next-gen processors, and without them we’re unable to advance our 7nm roadmap.  

    TSMC and Samsung have pulled ahead, and rather than spend boatloads of money to try to catch up and still fail to secure the lucrative contacts, we’re going to wait on the sidelines for the scraps.

    Fabs for the scraps can still be profitable because there’s little competition...
    watto_cobra
  • Reply 16 of 26
    tipootipoo Posts: 1,007member
    Another leading edge fab down. The drop in competitors is both remarkable and not just a little disheartening, capitalism got us this far, but after, it's going to be...Interesting. Everything consolidates or dies. 


    edited August 2018 Rayz2016blastdoorwatto_cobra
  • Reply 17 of 26
    GG1GG1 Posts: 217member
    Wow, in a few years' time, we have this progress:

    14/12 nm: Intel, GloFlo, TSMC, Samsung
    10 nm: TSMC, Samsung (Intel delayed)
    7 nm: TSMC (expected), Samsung (expected)
    5 nm: TSMC (roadmap for 2019); Samsung (on their roadmap)

    Yes, I realize that "X nm" is more of a marketing term than a technical comparison, but this technology shrink is taking its toll on the market. Intel is the biggest surprise (to me).
    watto_cobra
  • Reply 18 of 26
    blastdoorblastdoor Posts: 1,888member
    tipoo said:
    Another leading edge fab down. The drop in competitors is both remarkable and not just a little disheartening, capitalism got us this far, but after, it's going to be...Interesting. Everything consolidates or dies. 


    Thanks -- I love that figure (as I've mentioned elsewhere). 

    And we can mentally superimpose the cost curve of investments needed to fund new processes with the same steepness, just going up from left-to-right rather than down. 

    The big question -- who falls out next? My guess is that it won't be TSMC, but between Samsung and Intel it's a tough call. 
    watto_cobra
  • Reply 19 of 26
    melgrossmelgross Posts: 31,335member
    I think he's saying that they are not moving forward with cutting edge technology so they can better invest in cutting edge technology...OK. I'm not a chip expert so maybe what he said makes sense. It seems a little to 'spiny' to take seriously. 
    There aren’t too many ways to effectively spin “we can afford neither to develop an expensive next-gen process node nor to build the fab capacity required to make that investment ever turn a profit.”  

    Given their resources, it’s probably the best road to profitability, for them. Developing chips for fabrication at 7nm (and beyond) is extremely expensive, and most customers won’t be going there anytime soon. I suspect they’ll re-visit 7nm three to five years from now when it’s less expensive for them to implement, and for their customers to utilize. 
    You beat me to this. Each generation of fab costs at least 50% more than the previous generation. Several years ago AMD said that 14nm would cost more per transistor than 22nm would. That trend continues. When computer sales rose every year, that expense would be covered by more sales. But pc sales peaked several years ago at about 360 million units a year. It’s been dropping since then. Last year sales were around 260 million. Meanwhile ARM sales continue to grow, though those are slowing as well as smartphone sales worldwide are slightly down to flat, and tablet sales are dropping. Other devices are coming out that need the small, low power and cheap chips.

    GF has always been a more minor player, even when they were still part of AMD. With a 10nm fab costing at least $15 billion, and closer to $20 billion, not all can play in the space of 7nm, which cost much more.
    edited August 2018 muthuk_vanalingamwatto_cobra
  • Reply 20 of 26
    melgrossmelgross Posts: 31,335member

    blastdoor said:
    tipoo said:
    Another leading edge fab down. The drop in competitors is both remarkable and not just a little disheartening, capitalism got us this far, but after, it's going to be...Interesting. Everything consolidates or dies. 


    Thanks -- I love that figure (as I've mentioned elsewhere). 

    And we can mentally superimpose the cost curve of investments needed to fund new processes with the same steepness, just going up from left-to-right rather than down. 

    The big question -- who falls out next? My guess is that it won't be TSMC, but between Samsung and Intel it's a tough call. 
    These are all for CPU related fabs. There are several dozen fabs that do other work that are doing well.
    muthuk_vanalingamwatto_cobra
Sign In or Register to comment.