Intel promises 10nm chips still coming, despite being years late

Posted:
in General Discussion
An open letter from the interim CEO of Intel advises some progress is being made for its 10-nanometer processors, but volume production of chips using the process will still wait until 2019, alongside a notice of $1 billion in 14-nanometer production investments and a warning it will also be prioritizing production of high-performance chips over processors for entry-level devices.




Intel CFO and interim CEO Bob Swan's "Supply Update" starts by noting the growth of the processor industry, highlighting the 25 percent increase in its data-centric business in the first half of 2018, as well as 43 percent growth in cloud revenue. The PC industry as a whole is also improving, referencing a Gartner report advising second-quarter PC shipments grew globally for the first time in six years.

"We now expect modest growth in the PC total addressable market this year for the first time since 2011," writes Swan, "driven by strong demand for gaming as well as commercial systems."

The uptick in the PC market also introduces a challenge to Intel, applying pressure to its factory network. To cope, production of Xeon and Core processors will be prioritized to "serve the high-performance segments of the market," which would likely include Apple's Mac and MacBook lines.

At the same time, the prioritization is warned to cause supplies of processors for entry-level systems to become "undoubtedly tight." Even so, the company believes it has enough of a supply that it will be able to meet its full-year revenue outlook.

In answer to the "challenge," Intel are taking a number of steps to capitalize on these increases, including the investment of a record $15 billion in capital expenditures in 2018, up $1 billion from last year's spending. Part of this includes an investment into its 14-nanometer manufacturing sites in Oregon, Arizona, Ireland, and Israel, which is intended to increase supply of chips from those facilities.

A report from earlier in September claimed Intel was outsourcing some of its 14-nanometer production to TSMC, including the H310 and other 300-series chipsets. It was alleged Intel is short of meeting 14-nanometer processor demand "by as much as 50 percent," with outsourcing being one of a number of solutions.

The billion-dollar investment into its own sites will certainly help improve its own production, but it may take time for the sites to be upgraded to meet demand. Outsourcing remains a viable short-term solution for Intel until the investment pays off.

It is suspected the cause of the 14-nanometer supply crunch is due to Intel's continued delays in moving to 10-nanometer chips. Mass production for chips made under the process was originally expected to commence in 2016, but it has repeatedly been put off over yield issues, and is currently expected to occur at the end of 2019.

Swan offers a brief update about the 10-nanometer chips, advising the company is "making progress," with yields improving and volume production still anticipated for sometime in 2019.

Under the name "Cannon Lake," the 10-nanometer processors offer a number of benefits, but for MacBook Pro users it represents a potential memory expansion. While the 15-inch MacBook Pro was updated in July to include a 32GB RAM option, Apple did so by changing the design from using LPDDR3 memory to allow it to run the more power-dependent DDR4 memory, while continuing to use LPDDR3 on the 13-inch model and limiting it to just 16GB of RAM at maximum.

The Cannon Lake generation of processors all include LPDDR4 compatibility, a type of memory that is less power-hungry compared to DDR4, making it ideal for use in MacBooks.

So far, Intel has released only one processor in the Cannon Lake range, and even then with limited supply. The Core i3-812U is a dual-core chip with a 2.2GHz base clock speed, boostable to 3.2GHz, and a notebook-friendly 15-watt thermal design point.

Apple's iPhone A12 Bionic chip is using TSMC's 7nm process.

Comments

  • Reply 1 of 14
    thttht Posts: 3,099member
    Never thought I’d see the day when Intel’s leadership in fab goes away. It was the ace card that could always play even when they bombed on a CPU architecture.

    They have a big capacity lead over AMD, so the position isn’t threatened much, even if AMD Zen-based TSMC 7nm chips end up being faster than Intel chips. That capacity lead will only last so long though. Clocks ticking.
    curtis hannahlarrya
  • Reply 2 of 14
    Yeah keep baiting and dangling hope so PC manufacturers don’t bail. Keep promising and stalling so no one jumps ship prematurely. Though Apple for all intents and purposes is done with Intel and will cut them off as soon as their own chips are ready. Can’t happen soon enough.
    libertyforall
  • Reply 3 of 14
    SoliSoli Posts: 8,678member
    tht said:
    Never thought I’d see the day when Intel’s leadership in fab goes away. It was the ace card that could always play even when they bombed on a CPU architecture.

    They have a big capacity lead over AMD, so the position isn’t threatened much, even if AMD Zen-based TSMC 7nm chips end up being faster than Intel chips. That capacity lead will only last so long though. Clocks ticking.
    This did seem like the one thing Intel had going for it and something that could help keep them ahead of the game. Come on ARM-based Macs!
    macxpress
  • Reply 4 of 14
    nhtnht Posts: 4,429member
    According to Gary, 7nm is on track for early 2019 production which is in line with the other foundries except TSMC and Apple of course. Apple is always first to production with TSMC to make the fall iPhone launch.

    Intel is on the same track with 10nm. Based on people who actually know, 10nm yield is steadily improving and should be at Intel acceptable levels by the end of the year for early 2019 mass production. Remember, Intel went through a similar exercise at 14nm. Yield delays were standard practice in the history of the semiconductor industry up until Apple joined our ranks. TSMC and Apple work jointly on a customized process that must be in production in time for the yearly iPhone launch. The trade-off made of course is performance for yield. Intel on the other hand will not sacrifice performance for yield thus the 10nm delay.

    The result being that the Intel processes are faster and denser than the same named foundry processes.
    https://www.semiwiki.com/forum/content/7602-semicon-west-intel-10nm-gf-7nm-update.html#comments

    So part of this is Apple driven and why TSMC is moving quickly.  

    The discussions on that forum goes over my head quickly.




  • Reply 5 of 14
    Time to bring back Andy Grove!  /s
  • Reply 6 of 14
    Right! Bring manufacturing back to the US. 

    Result. Nothing gets done. 

    But, always make sure the CEO's get paid whopping salaries, and investors get paid dividends for doing nothing.

    Make sure the engineers and scientists get paid as little as possible, and make do with old manufacturing equipment. 

    The American Way.


    GeorgeBMacprismatics
  • Reply 7 of 14
    larryjw said:
    Right! Bring manufacturing back to the US. 

    Result. Nothing gets done. 

    But, always make sure the CEO's get paid whopping salaries, and investors get paid dividends for doing nothing.

    Make sure the engineers and scientists get paid as little as possible, and make do with old manufacturing equipment. 

    The American Way.


    I'll take it you don't care for the United States.
    williamlondonSpamSandwich
  • Reply 8 of 14
    Intel clearly needs better or new leadership...
  • Reply 9 of 14
    larryjw said:
    Right! Bring manufacturing back to the US. 

    Result. Nothing gets done. 

    But, always make sure the CEO's get paid whopping salaries, and investors get paid dividends for doing nothing.

    Make sure the engineers and scientists get paid as little as possible, and make do with old manufacturing equipment. 

    The American Way.


    Yeh, if you study the American steel and auto industries you would know how true this is.   Short term profit, executive bonuses and stock dividends took the place of innovation and reinvestment.

    Fortunately, the auto industry was saved with some government bailouts and seems to be doing better.
  • Reply 10 of 14
    nhtnht Posts: 4,429member
    larryjw said:
    Right! Bring manufacturing back to the US. 

    Result. Nothing gets done. 

    But, always make sure the CEO's get paid whopping salaries, and investors get paid dividends for doing nothing.

    Make sure the engineers and scientists get paid as little as possible, and make do with old manufacturing equipment. 

    The American Way.
    Yeh, if you study the American steel and auto industries you would know how true this is.   Short term profit, executive bonuses and stock dividends took the place of innovation and reinvestment.

    Fortunately, the auto industry was saved with some government bailouts and seems to be doing better.
    Or maybe going to 10nm is hard and Intel may or may not have made a wrong technical decision to go cobalt using plating vs ccs techniques giving them yield problems or the fact they are 36nn vs 40nm SAQP like TSMC/Apple.

    The only clear conclusion from all of this is Intel and Samsung both screwed the pooch by either turning down Apple as a partner or stole from them. GF wishes it had that missed opportunity and TSMC has run with the ball given Apple’s large investments to catch up to Intel.
  • Reply 11 of 14
    nht said:
    larryjw said:
    Right! Bring manufacturing back to the US. 

    Result. Nothing gets done. 

    But, always make sure the CEO's get paid whopping salaries, and investors get paid dividends for doing nothing.

    Make sure the engineers and scientists get paid as little as possible, and make do with old manufacturing equipment. 

    The American Way.
    Yeh, if you study the American steel and auto industries you would know how true this is.   Short term profit, executive bonuses and stock dividends took the place of innovation and reinvestment.

    Fortunately, the auto industry was saved with some government bailouts and seems to be doing better.
    Or maybe going to 10nm is hard and Intel may or may not have made a wrong technical decision to go cobalt using plating vs ccs techniques giving them yield problems or the fact they are 36nn vs 40nm SAQP like TSMC/Apple.

    The only clear conclusion from all of this is Intel and Samsung both screwed the pooch by either turning down Apple as a partner or stole from them. GF wishes it had that missed opportunity and TSMC has run with the ball given Apple’s large investments to catch up to Intel.
    Perhaps.   I'm hoping that intel doesn't go the way of the American steel and auto industries.

    But too, on their side, as you suggest, tech is hard and its risky.   Many companies are one mistake away from catastrophe.  My fingers are crossed that Intel pulls  it off.  

    The same could not have been said for American steel and auto industries.  That was a problem of too many people with their fingers in the pot while nobody paid sufficient attention to business. 
  • Reply 12 of 14
    wood1208wood1208 Posts: 1,867member
    Do you ever wonder Intel might have 10nm good to go but why introduced now when Intel can easily milk out more of it's 14nm,14nm+,14nm++ process for some time.. Moreover, demand for Intel chips are more than Intel can produced. One reason Intel planning to farm out some 14nm low end PC processors to TSMC.
  • Reply 13 of 14
    thttht Posts: 3,099member
    wood1208 said:
    Do you ever wonder Intel might have 10nm good to go but why introduced now when Intel can easily milk out more of it's 14nm,14nm+,14nm++ process for some time.. Moreover, demand for Intel chips are more than Intel can produced. One reason Intel planning to farm out some 14nm low end PC processors to TSMC.
    No. Someone name checked Andy Grove who is an Intel founder. He wrote a book called “Only the Paranoid Survive”. Intel’s culture is to be paranoid about competition, and they want to outrace everyone to the next node. This is the de facto reason for Intel’s dominance. At times, they were a full 2 years, a full node, ahead of their manufacturing competition. That’s an automatic 2x advantage in costs and performance. Giving up a lead like that is suicide.

    They already fired CEO Brian Krzanich in June - ostensibly he resigned because he broke company rules by having a consensual relationship with an underling, but that sounds a lot like he was fired and was allowed to save face - and this announcement is from the new CEO. Monopolies have near impenetrable, but years of neglect can make those walls crumble all on their own.
     


  • Reply 14 of 14
    wood1208wood1208 Posts: 1,867member
    tht said:
    wood1208 said:
    Do you ever wonder Intel might have 10nm good to go but why introduced now when Intel can easily milk out more of it's 14nm,14nm+,14nm++ process for some time.. Moreover, demand for Intel chips are more than Intel can produced. One reason Intel planning to farm out some 14nm low end PC processors to TSMC.
    No. Someone name checked Andy Grove who is an Intel founder. He wrote a book called “Only the Paranoid Survive”. Intel’s culture is to be paranoid about competition, and they want to outrace everyone to the next node. This is the de facto reason for Intel’s dominance. At times, they were a full 2 years, a full node, ahead of their manufacturing competition. That’s an automatic 2x advantage in costs and performance. Giving up a lead like that is suicide.

    They already fired CEO Brian Krzanich in June - ostensibly he resigned because he broke company rules by having a consensual relationship with an underling, but that sounds a lot like he was fired and was allowed to save face - and this announcement is from the new CEO. Monopolies have near impenetrable, but years of neglect can make those walls crumble all on their own.
     


    Well said. Brian Krzanich was out because of consensual relation ship with employee is no reason to invite him to leave but if he was out due to his lack of leadership about so much delay in moving rapidly to 10nm is OK with me. Staying ahead of curve in semiconductor is not just making money but survival. At this poit, Intel is OK with 14nm process and it's chip demand on hand but in rapidly changing tech landscape, no lead can last long unless the company has in it's DNA “Only the Paranoid Survive”.
Sign In or Register to comment.