Richard Cownie (tich.delete@this.pobox.com) on February 1, 2013 6:35 am wrote:
> anon (anon.delete@this.anon.com) on February 1, 2013 5:56 am wrote:
> > They have been trying with x86, not ARM yet.
>
> Which indicates that they don't have sufficient advantage from their
> non-cpu design expertise, plus their manufacturing, to make a dent
> in the market. Why would substituting ARM for x86 fix that ?
>

No need to fight an uphill battle against software installed base.

> > You seem to be saying that other companies could make good profits, but Intel >inherently cannot.
>
> I think it's a big challenge to make good profits in the phone/tablet SoC
> market against the current, very competent, winners. Even for Intel.
> And apart from manufacturing, I don't think Intel's expertise is particularly
> well matched to that market. They haven't done much with SoC's;

They did SOCs well before the name SOCs became popular. In particular, 8096/196/296 family. Even early on, 80186 was sorta SOC.
And, of course, SandyB/IvyB are SOCs.

> they haven't done much with handheld devices;

They did XScale.

> they're not great at GPUs;

They are no worse than market leaders that all use either PowerVR or Mali. In fact, they are somewhat better, since apart from using PowerVR they have in house GPU, while less power efficient, but higher-performing.

> they seem to
> operate on long-term product cycles which may not match the fast-moving
> mobile market (the same problem they had with GPUs).
> >

I thing, their product cycles cadence is the same as Apple and Samsung.

> > >
> > > > If x86 does not work out, they could design their own ARM cores, they could fab cortex cores like Samsung,
> > >
> > > There's more to a good phone/tablet SoC than the ARM cores.
> >
> > Yep, and Intel seems to be more than capable of producing such devices on a technical level today.
>
> I think they've demonstrated the ability to make something functional.
> They haven't demonstrated the ability to make something clearly superior.
> And "clearly superior", on some dimension, is what you need to grab
> market share from strong incumbents.
>
> > Neither is *anything* in these ARM markets, for *any* player,
> > a way to get the kind of huge profits that x86 generates.
>
> Correction: that x86 used to generate. Last quarter Qualcomm alone made
> $1.9B profit; Intel made $2.5B. Not a big gap. And Intel's desktop/laptop
> business shrank a lot YoY (a gap which was partly made up by strong growth
> on the server side).
>

I think of Qualcomm as first and foremost RF and basedand company with application processors being the distant third. Am I wrong about it?
And there were reports that on application processors front Qualcomm had serious troubles with foundry partners.

> > However it's growing of course, and every bit they take can add to their bottom line,
> > and take away from TSMC, or Samsung, or nvidia, etc. So the assertion that the ARM ecosystem
> > will grow to threaten Intel with no recourse for Intel to take is just flawed.
>
> Well, I started out on this thread with the suggestion that *either* ARM-based
> products would take server share, *or* Intel would have to cut server-product
> margins to fight them off. And either way would be a big change in the
> server business (and good for consumers). So I've accepted all along that Intel
> has "recourse to take" - it's just that the form of that "recourse" is a major
> restructuring into a business with lower margins in a more competitive market.
>
>
>