By: Doug S (foo.delete@this.bar.bar), February 4, 2013 11:08 am
Room: Moderated Discussions
Richard Cownie (tich.delete@this.pobox.com) on February 3, 2013 8:28 pm wrote:
> someone (someone.delete@this.somewhere.com) on February 3, 2013 1:44 pm wrote:
> > It is hilarious to hear folks claiming Intel and x86 are doomed because of its
> > huge overhead of fabs etc. Apparently they think ARM silicon is made in fabs
> > and processes subsidized by the tooth fairy or easter bunny or something.
>
> Intel Q4 gross margin 58%, revenue $13.5B, operating income $3.2B
> TSMC Q4 gross margin 47%
>
> So Intel's production cost cost 0.42*$13.5B = $5.67B.
> If they had the same costs, but only 47% gross margin, revenue would be $10.69B
> If other expenses of $4.63B stay constant, that would reduce operating income
> from $3.2B down to 10.69-5.67-4.63 = $0.39B, an 88% drop.
>
> The difference between Intel's and TSMC's margins makes a vast difference to
> profitability, and if Intel worked on a foundry business model it would
> have to become a very different company, with a much reduced cost structure.
>
> That might happen, but Intel's management and stockholders would probably
> view it as a last resort.
>
> [Of course, doing a small percentage of low-margin business has less impact]
Your comparison seems to assume Intel switching from their current business to a foundry business. Unless they spun off the x86 design side like AMD did, that's not going to happen. The PC business is shrinking, but it will be around for a long time. At any rate, the shrinkage is all on the client side, unless ARM servers make inroads Intel's server revenue will continue to grow.
Intel's foundry business is in addition to their current business, which if it becomes large might hurt their overall gross margin but the company as a whole would make more money. If the revenue top line is growing and profit is growing, who cares about margin shrinkage if the reason for it is entering another profitable line of business? If they require the same margins they make now before entering another business, they'll never do so because they'll never get lucky enough to be handed another monopoly.
Intel's problem is the coming 450mm transition. If they want to have anything like the number of fabs they do now, post-450mm, they are going to need to double their chip production (measured in sq mm) and that certainly isn't going to happen with a shrinking PC market, even if they succeed in grabbing some of the growth in mobile for themselves.
Why shouldn't they just accept having half the number of fabs they do now? Because they'll lose out on some of the economies of scale they get from their "copy exactly" strategy the fewer fabs they have. It also makes retooling fabs more difficult when each is the source of a larger percentage of your total output. You also have bigger problems with what to do with your outdated fabs. The chipset will soon disappear, and with it a way for Intel to cheaply fill outdated fab capacity.
> someone (someone.delete@this.somewhere.com) on February 3, 2013 1:44 pm wrote:
> > It is hilarious to hear folks claiming Intel and x86 are doomed because of its
> > huge overhead of fabs etc. Apparently they think ARM silicon is made in fabs
> > and processes subsidized by the tooth fairy or easter bunny or something.
>
> Intel Q4 gross margin 58%, revenue $13.5B, operating income $3.2B
> TSMC Q4 gross margin 47%
>
> So Intel's production cost cost 0.42*$13.5B = $5.67B.
> If they had the same costs, but only 47% gross margin, revenue would be $10.69B
> If other expenses of $4.63B stay constant, that would reduce operating income
> from $3.2B down to 10.69-5.67-4.63 = $0.39B, an 88% drop.
>
> The difference between Intel's and TSMC's margins makes a vast difference to
> profitability, and if Intel worked on a foundry business model it would
> have to become a very different company, with a much reduced cost structure.
>
> That might happen, but Intel's management and stockholders would probably
> view it as a last resort.
>
> [Of course, doing a small percentage of low-margin business has less impact]
Your comparison seems to assume Intel switching from their current business to a foundry business. Unless they spun off the x86 design side like AMD did, that's not going to happen. The PC business is shrinking, but it will be around for a long time. At any rate, the shrinkage is all on the client side, unless ARM servers make inroads Intel's server revenue will continue to grow.
Intel's foundry business is in addition to their current business, which if it becomes large might hurt their overall gross margin but the company as a whole would make more money. If the revenue top line is growing and profit is growing, who cares about margin shrinkage if the reason for it is entering another profitable line of business? If they require the same margins they make now before entering another business, they'll never do so because they'll never get lucky enough to be handed another monopoly.
Intel's problem is the coming 450mm transition. If they want to have anything like the number of fabs they do now, post-450mm, they are going to need to double their chip production (measured in sq mm) and that certainly isn't going to happen with a shrinking PC market, even if they succeed in grabbing some of the growth in mobile for themselves.
Why shouldn't they just accept having half the number of fabs they do now? Because they'll lose out on some of the economies of scale they get from their "copy exactly" strategy the fewer fabs they have. It also makes retooling fabs more difficult when each is the source of a larger percentage of your total output. You also have bigger problems with what to do with your outdated fabs. The chipset will soon disappear, and with it a way for Intel to cheaply fill outdated fab capacity.