Ricardo B (ricardo.b.delete@this.xxxxx.xx) on September 29, 2013 4:00 pm wrote:
> RichardC (tich.delete@this.pobox.com) on September 29, 2013 12:28 pm wrote:
> > Ricardo B (ricardo.b.delete@this.xxxxx.xx) on September 29, 2013 8:27 am wrote:
> > > The point is that, in a future dominated by cheap SoCs and with ever growing
> > > NRE costs, there may not be enough revenue to pay off the investment.
> > > If this happens, you may get to borrow money to fund the
> > > development of a new process node or even maybe two.
> > > But if you can't pay it off, then you'll be bankrupt.
> > >
> > > And this will not happen in a day, it would be progressive and it's already happening.
> > > The number of foundries developing new process nodes is reducing
> > > and the time to market for the ones that do is increasing.
> > > UMC is just rolling out 28 nm *now*
> >
> > But it's basic economics is that if Intel can be highly profitable with an
> > investment in fabs at a particular process node, *and* if there is no significant
> > barrier to entry (and the x86 patents used to be such a barrier, but wouldn't
> > be in a future dominated by ARM-based or ISA-neutral software), then *somebody* -
> > not necessarily an existing player - will make the investments to compete.
> >
> > So I can imagine a future where Intel keeps spending money to stay ahead; and I can
> > imagine a future where Intel accepts spending less money and having less advantage
> > and lower margins. I just can't conceive of a future where everyone stops and leaves
> > Intel with enough of a lead to keep its high margins (once x86 dominance ceases to
> > be a big deal).
>
> I was imagining a future where Intel's high profit margins are eroded by good enough cheap ARM SoCs
> and they would, too, stop investing in new process nodes, because they can't afford it anymore.

It is easy to imagine it, and everybody claims that this is how Intel will fail, but what would be impressive is actually narrowing down what would happen and giving some numbers.

Lots of people were predicting the doom of Intel from ARM since the iPhone (earlier for some, but iPhone brought ARM to the minds of the masses). Their margins are still up near 60% and we've been waiting a long time for them to plummet like a rock.

> My question is is weather in that future, TSMC and other foundries,
> could also keep investing in new process nodes.
> And if not, weather Intel or one of the foundries would be left
> as the last man standing with the most advanced process node.