22nm Finfet May Extend Intel’s Two Year Lead In Process.

Intel has a two year lead in process technology over the rest of the semiconductor industry and could get further ahead if yields are satisfactory on its finfet-based 22nm process due for introduction later this year, said Mike Bryant, CTO of Future Horizons, at last week’s IFS 2011.

Intel has been shipping 32nm in volume since 2009, said Bryant, “while the others are only just getting into volume production.”

 

TSMC has issued its 28nm design kits but, said Bryant: “This tends to happen a year ahead of full production’.

 

“If Intel’s 22nm process yields well,” reckoned Bryant, “then they will move further ahead.”

This is what is giving Intel CEO Paul Otellini the confidence to say, earlier this week: “Our job is to make sure that we can out-perform, out-battery, and out-compatibility the ARM guys on Win 8 tablets.”

Intel has always suffered in competitiveness with ARM for two reasons: ARM is multi-sourced and inexpensive; ARM uses less power.

If finfet 22nm eliminates the second of these disadvantages, then Intel will have a powerful argument to persuade OEMs to use x86 in tablets.

That’s in default of using the powerful argument of MDF, which the US FTC has told Intel not to use any more.

Other prizes for Intel, if it can crack 22nm finfets, are that processors might get to 4GHz, and Apple may use Intel for foundry to get an advantage over Android competitors.

If that happens, one wonders:

Would Intel be so keen to get Apple’s foundry business that it would manufacture ARM processors?

Or:

Would Apple be so keen to get Intel’s 22nm finfet process that it would use x86 in iPads?

Tags: arm processors, future horizons, semiconductor industry, tsmc

Related Tech News

5 Comments

  1. David Manners
    July 26, 2011 07:51

    Great story Robert

  2. Robert
    July 26, 2011 03:57

    I agree with Scunnerous. There was a story about 6 months ago that Intel presented a mobile Atom concept design to a major cell phone chipset company. Apparently, when the power analysis foil was shown, the CTO leaned back and laughed so hard he broke the conference room chair, falling over backwards. Needless to say Intel’s presentation suffered the same fate.

  3. Mike Bryant
    July 25, 2011 20:42

    Emphasis on the ‘might’ !!! :-)
    But they’re both obsolete in 2 generations of Moore’s Law anyway.

  4. David Manners
    July 25, 2011 11:54

    Yes indeed, Scunnerous, there’ve always been two ways of looking at these process stratagems like high-k. low-k, SiGe etc – one way is that they’re technological advances improving the product; the other is that they’re acts of desperation to keep delivering products which show some improvement over the previous generation but not the traditional threesome of more density, less power and higher performance. Is 22nm finfet a technical advance or an act of desperation? The proof of the pudding will be in the eating but, as Mike Bryant points out, fully depleted SOI (FD-SOI) on SOI wafers – the approach of IBM, AMD, GloFo and ST – might prove to be a better bet than 22nm finfet.

  5. Scunnerous
    July 25, 2011 06:46

    I find it funny that Intel talks of “out-compatibility” in a market where they have no current presence… compatible with what? Have they been asleep for the past 10 years? And as I’ve said before, isn’t it a bit of a confession that they need 22nm Finfet to out-compete current mobile devices at 45nm and 32nm?
    Can they really come up with a home-brewed multi-faceted SoC based on x86 which has the flexibility to hit the market points of all the flavors of mobile chips with embedded ARM? Maybe I’m missing something here but I don’t see how – their whole business model just doesn’t fit. There’s also nothing to proprietarize here.
    It almost seems that it hasn’t yet sunk in with Intel that the CPU doesn’t matter anymore as long as it does its job reasonably efficiently along with all the other bits in a mobile SoC. A CPU is no a longer a mystery – it’s just another embedded plugin module with known, documented parameters and an affordable licence fee. Not to over-simplify of course, but the trick is more in the integration.
    Tick-Tock-Tick-Tock…. Doink.

Share your knowledge - Leave a comment