commitment to foundry, shutting down fabless talk

By: Daniel B (fejenagy.delete@this.gmail.com), March 23, 2021 4:46 pm
Room: Moderated Discussions
Rayla (rayla.delete@this.example.com) on March 23, 2021 3:45 pm wrote:
> Jon Masters (jcm.delete@this.jonmasters.org) on March 23, 2021 3:31 pm wrote:
> > The "Intel Unleashed" event this afternoon featured the announcement of
> > a new "IDM 2.0" model including a second attempt at doing a foundry:
> >
> > https://www.anandtech.com/show/16573/intels-new-strategy-20b-for-two-fabs-meteor-lake-7nm-tiles-new-foundry-services-ibm-collaboration-return-of-idf
> >
> > Have to give Intel kudos for today's news. Second to a full spin-out this is exactly the
> > right thing. The proof of independence will be in the pudding, etc. but it has potential.
>
> Why is this any more likely to work today than several years ago, when Intel's process
> was ahead of TSMC and Samsung and their foundry initiative still failed?


It starts with perceptions. Intel needs to change the unfavourable ways in which it's viewed by potential foundry customers (inflexible, costly, not committed) and Wall Street (distractng talk of going fabless which is quite nonsense but there you end up when all you see is maximizing cash-flow, margins etc, like what Intel needed to solve its problems is actually even more money). Gelsinger, with his credibility, did just that, putting bed all this dysfunctional speculation of recent years.

I don't for a second believe Intel foundry (or semi-custom silicon) failed in the past for technical reasons, it has failed for political and mismanagement reasons. In my experience as an industry observer, when a company of any discipline is organized around developing standardized high-volume products, it is arduously difficult to incorporate customer feedback let alone run a custom shop. Invariably, this results in mutually frustrating experience for both sides where the vendor doesn't understand why the customer is so adamant at minute details that doesn't add business value (often quite rightly, still), and the customer doesn't understand why the vendor is not amenable to change requests. You need an entirely different culture and organizational processes, even before you investigate any changes needed to the design and manufacturing processes (customer-friendly design tools, libraries, design co-operation and semi-custom services, smaller wafer batches, rocket batches etc.). The foundry business is much broader than only cutting edge, and for the vast majority of the market, the performance and density of Intel's 14nm processes are quite advanced still, mature, strong yieldds, and most importantly, all the 14nm fabs paid for 2 times over, that must help with cost competitiveness too.

True, Gelsinger cannot overnight fix 10nm or pull in 7nm all that much either. But process tech is not the sole issue at Intel, even without that cockup, Intel is facing tectonic shifts in the market it cannot fight - some of its buyers are really really big and they really don't like being forced to buy expensive standard products. They expect a full tailoring service, and retain their own tailors too, like royalty.

Gelsinger as the new CEO he has the ability to reset relationship with both major customers and the Street, and set a clear direction and how, strategically speaking, Intel will get there. He's achieved that much already. Execution is another matter, quite right, but we will need to wait a little for that to unfold. I will not be surprised if Apple ends up fabbing some M-parts with Intel in a few years. If for nothing else, to keep TSMC honest.
< Previous Post in ThreadNext Post in Thread >
TopicPosted ByDate
Intel's IDM 2.0 announcementJon Masters2021/03/23 03:31 PM
  Intel's IDM 2.0 announcementRayla2021/03/23 03:45 PM
    Intel's IDM 2.0 announcementdmcq2021/03/23 04:30 PM
    commitment to foundry, shutting down fabless talkDaniel B2021/03/23 04:46 PM
      commitment to foundry, shutting down fabless talkKester L2021/03/23 04:56 PM
        commitment to foundry, shutting down fabless talkanonymou52021/03/24 12:57 AM
          commitment to foundry, shutting down fabless talkme2021/03/24 03:25 AM
            commitment to foundry, shutting down fabless talkJoe2021/03/24 05:56 AM
              commitment to foundry, shutting down fabless talkMichael S2021/03/24 06:47 AM
                commitment to foundry, shutting down fabless talkme2021/03/24 07:04 AM
      commitment to foundry, shutting down fabless talkJames2021/03/24 03:39 AM
        Intel gutted 7 nmme2021/03/24 03:48 AM
          Intel gutted 7 nmDoug S2021/03/24 08:20 AM
            Intel gutted 7 nmanon2021/03/25 12:43 PM
              Intel gutted 7 nmeh2021/03/26 02:36 AM
                Intel gutted 7 nmAdrian2021/03/26 08:58 AM
                  Intel gutted 7 nmwumpus2021/03/26 09:53 AM
                  Intel gutted 7 nmme2021/03/26 10:11 AM
                  Intel gutted 7 nmMaxwell2021/03/26 02:21 PM
                    Intel gutted 7 nmYoav2021/03/29 01:06 AM
              Intel gutted 7 nmDoug S2021/03/26 09:43 AM
                Intel gutted 7 nmanon2021/03/26 03:39 PM
                  Intel gutted 7 nmNoSpammer2021/03/26 08:46 PM
                    Intel gutted 7 nmAdrian2021/03/26 11:57 PM
                      Intel gutted 7 nmAndrey2021/03/28 06:18 PM
                    Intel gutted 7 nmanon2021/03/27 05:52 AM
                    Intel gutted 7 nmwumpus2021/03/27 12:16 PM
                  Intel gutted 7 nmDoug S2021/03/27 05:08 PM
                    Intel gutted 7 nmanon2021/12/13 03:05 PM
                      Intel gutted 7 nm---2021/12/13 04:31 PM
                      Intel gutted 7 nmDoug S2021/12/13 09:34 PM
                        Intel gutted 7 nmanon2021/12/13 10:41 PM
                          Intel gutted 7 nmDoug S2021/12/14 11:38 AM
                            Intel gutted 7 nmanon2021/12/14 11:58 AM
                            Feed the Troll less (unless you are enjoying it!)Mark Roulo2021/12/14 01:52 PM
                Intel gutted 7 nmblue2021/03/27 10:16 AM
                  Intel gutted 7 nmme2021/03/27 12:22 PM
                  Intel gutted 7 nmDoug S2021/03/27 05:14 PM
                    Intel gutted 7 nmDavid Kanter2021/04/07 09:06 PM
                      Intel gutted 7 nmDoug S2021/04/08 12:38 PM
        commitment to foundry, shutting down fabless talkDaniel B2021/03/24 08:08 AM
      commitment to foundry, shutting down fabless talkDoug S2021/03/24 08:12 AM
      commitment to foundry, shutting down fabless talkMaxwell2021/03/25 11:16 AM
        commitment to foundry, shutting down fabless talkme2021/03/25 11:38 AM
          commitment to foundry, shutting down fabless talkAdrian2021/03/25 01:01 PM
            commitment to foundry, shutting down fabless talkme2021/03/27 12:57 PM
    A lot has changed since thenDoug S2021/03/24 08:01 AM
    Intel's IDM 2.0 announcementGabriele Svelto2021/03/25 03:17 AM
  Intel's IDM 2.0 announcementBeastian2021/03/23 05:39 PM
    Intel's IDM 2.0 announcementJames2021/03/24 03:14 AM
      Intel's IDM 2.0 announcementme2021/03/24 03:37 AM
        Intel's IDM 2.0 announcementJames2021/03/24 03:53 AM
      Intel's IDM 2.0 announcementanonymou52021/03/24 08:39 AM
    Intel's IDM 2.0 announcementJukka Larja2021/03/24 06:05 AM
    Intel's IDM 2.0 announcement - Hyper-scalers as customersBjörn Ragnar Björnsson2021/03/24 06:19 PM
  5nm the end of DRAM?Brett2021/03/25 05:17 PM
    What is next after DDR5?Ganon2021/03/25 05:56 PM
      What is next after DDR5?wumpus2021/03/26 09:10 AM
        What is next after DDR5?Brett2021/03/26 04:39 PM
          What is next after DDR5?wumpus2021/03/28 10:44 AM
      What is next after DDR5?Brett2021/03/27 04:27 PM
        What is next after DDR5?Doug S2021/03/27 05:18 PM
      What is next after DDR5?Anononymous Coward2021/03/30 07:03 AM
        What is next after DDR5?Andrey2021/03/30 09:09 AM
          What is next after DDR5?David Hess2021/04/04 08:52 PM
      What is next after DDR5?David Hess2021/04/04 08:45 PM
        What is next after DDR5?Doug S2021/04/05 06:57 AM
          What is next after DDR5?David Hess2021/04/05 10:45 AM
          Same costAnon2021/04/05 11:31 AM
            Same costDavid Hess2021/04/05 08:26 PM
            Same costMaxwell2021/04/06 07:17 PM
              Same costDavid Hess2021/04/07 02:35 PM
          What is next after DDR5?wumpus2021/04/07 09:06 AM
      L4 is not a memory technologyHeikki Kultala2021/04/07 12:38 PM
        DRAM cachewumpus2021/04/08 08:24 AM
        Intel iGPU memory was eDRAM, not SRAMMark Roulo2021/04/08 11:31 AM
          I am not talking about the eDRAM. I am talking about the cache that is L3 for CPUHeikki Kultala2021/04/08 11:09 PM
            I am not talking about the eDRAM. I am talking about the cache that is L3 for CPURobert Williams2021/04/09 11:24 AM
              This STILL has NOTHING to do with the eDRAM. I am talking about the SRAM LLCHeikki Kultala2021/04/09 12:57 PM
                This STILL has NOTHING to do with the eDRAM. I am talking about the SRAM LLC---2021/04/09 01:40 PM
    5nm the end of DRAM?Rayla2021/03/26 09:28 AM
      5nm the end of DRAM?riy2021/04/08 06:26 AM
        5nm the end of DRAM?Michael S2021/04/08 11:51 AM
    5nm the end of DRAM?Doug S2021/03/26 09:52 AM
      5nm the end of DRAM?wumpus2021/04/07 09:10 AM
        5nm the end of DRAM?Robert Williams2021/04/07 11:05 AM
    Totaly misunderstanding of DDR SDRAM memory interfaces and DRAMHeikki Kultala2021/04/07 12:36 PM
      Totaly misunderstanding of DDR SDRAM memory interfaces and DRAManonymou52021/04/07 01:13 PM
Reply to this Topic
Name:
Email:
Topic:
Body: No Text
How do you spell tangerine? 🍊