Apple embeds teams into Intel

[Yousuf Khan]

There's been rumours lately that Apple might be able to get its first
x86 Mac out by January 2006, ahead of its original projections of May
2006. This story doesn't seem to think that it's going to be ahead of
schedule. Also it's mentioned that there are both engineering *and*
sales teams assigned from both companies to each other. This is the sort
of treatment Intel reserves for major corporate customers, of which
Apple now seems to be one.

IGM: Apple Transition Team at Intel
http://www.insanely-great.com/news.php?id=5496

Yousuf Khan

 

[nobody]

There's been rumours lately that Apple might be able to get its first
x86 Mac out by January 2006, ahead of its original projections of May
2006. This story doesn't seem to think that it's going to be ahead of
schedule. Also it's mentioned that there are both engineering *and*
sales teams assigned from both companies to each other. This is the sort
of treatment Intel reserves for major corporate customers, of which
Apple now seems to be one.

IGM: Apple Transition Team at Intel
http://www.insanely-great.com/news.php?id=5496

Yousuf Khan

If the rumors about Dell finally going AMD are true, Steve may aim to
take over Mike's place of the last Intel-only "major" (I guess AAPL
volume is less than 10% of DELL) 1st(???!!!) tier PC vendor
;p
NNN

 

[Yousuf Khan]

If the rumors about Dell finally going AMD are true, Steve may aim to
take over Mike's place of the last Intel-only "major" (I guess AAPL
volume is less than 10% of DELL) 1st(???!!!) tier PC vendor
;p
NNN

It might actually be less expensive for Intel to do it that way. If
Intel cuts Dell loose, it doesn't have to pay for all of those "special"
discounts anymore. It can give all of those special discounts to Apple
instead, and because Apple is so much smaller than Dell, the overall
amount spent on the discounts will be much less for Intel. Plus, Intel
gets a partner with a better name recognition. I hate to say it, but
this makes it extremely unlikely that Apple will go with AMD after this.
Though Jobs is more likely to be heavy-handed in his negotiations with
Intel, and therefore more likely to go with an alternative supplier like
AMD, it doesn't seem likely that Intel will want to let go of Apple.

Yousuf Khan

 

[nobody]

It might actually be less expensive for Intel to do it that way. If
Intel cuts Dell loose, it doesn't have to pay for all of those "special"
discounts anymore. It can give all of those special discounts to Apple
instead, and because Apple is so much smaller than Dell, the overall
amount spent on the discounts will be much less for Intel. Plus, Intel
gets a partner with a better name recognition. I hate to say it, but
this makes it extremely unlikely that Apple will go with AMD after this.
Though Jobs is more likely to be heavy-handed in his negotiations with
Intel, and therefore more likely to go with an alternative supplier like
AMD, it doesn't seem likely that Intel will want to let go of Apple.

Yousuf Khan

If Apple manages locking down the OS to Apple-only hardware, there
will be no embarrassing side-by side comparisons of A64 wiping the
floor with twice more expensive P4EE.
Besides, it's doubful that any major Apple customer would ask tough
questions such as "Why you don't offer Opteron servers? Do you want me
to go acrosss the street to HP?" While Apple may offer multi-socket
servers (at least, AFAIR, they used to), they probably have only one
customer of noticeable size for these, and that customer is...
....keep scrolling...
....
AAPL

NNN

 

[Yousuf Khan]

If Apple manages locking down the OS to Apple-only hardware, there
will be no embarrassing side-by side comparisons of A64 wiping the
floor with twice more expensive P4EE.

Yeah, but there is still a way to directly compare Apple to something
else. Just install a non-locked OS, such as Windows or Linux. Apple has
said that they won't prevent anyone from installing Windows on their
machines (and I assume that means Linux too).

Besides, it's doubful that any major Apple customer would ask tough
questions such as "Why you don't offer Opteron servers? Do you want me
to go acrosss the street to HP?" While Apple may offer multi-socket
servers (at least, AFAIR, they used to), they probably have only one
customer of noticeable size for these, and that customer is...
...keep scrolling...
...
AAPL

Yeah, the average Apple flowerchild doesn't know or care what servers
do. However, even if they don't ask why there's no Opterons, they might
indeed ask why there aren't any Athlons or Turions, or even Semprons.

It's possible that as Apple customers get used to lower prices, they
might be demanding even lower prices later.

Yousuf Khan

 

[Felger Carbon]

Apple traditionally prefers compact cases with no room for massive (or
noisy) heatsinks. The current AMD line of cool-running 90nm CPUs
would seem ideal for Apple's traditional packaging purposes. Intel
will not have production quantities of its next cool-running desktop
CPUs until the end of 2006, a year after Apple will have to have
frozen its x86 packaging.

It is a puzzlement.

 

[Keith]

It might actually be less expensive for Intel to do it that way. If
Intel cuts Dell loose, it doesn't have to pay for all of those "special"
discounts anymore. It can give all of those special discounts to Apple
instead, and because Apple is so much smaller than Dell, the overall
amount spent on the discounts will be much less for Intel.

Intel would trade Dell's volumes for Apple's? That doesn't sound like a
good idea (though neither is Itanic .

Plus, Intel
gets a partner with a better name recognition. I hate to say it, but
this makes it extremely unlikely that Apple will go with AMD after this.
Though Jobs is more likely to be heavy-handed in his negotiations with
Intel, and therefore more likely to go with an alternative supplier like
AMD, it doesn't seem likely that Intel will want to let go of Apple.

This is going to be interesting to watch. Jobs is rather good at the
negotiation table.

 

[Yousuf Khan]

Intel would trade Dell's volumes for Apple's? That doesn't sound like a
good idea (though neither is Itanic .

Yes, that's exactly my point, it's precisely because of Dell's volumes
that Intel might want to cut it loose! If it's starting to lose money on
Dell, then those high volumes would be a *bad* thing for Intel. Large
volumes and positive numbers, give you large positive numbers. However,
large volumes and negative numbers, give you large negative numbers!
With Apple's small numbers, if it loses money on Apple, then it'll lose
much less money than it would've with Dell.

Yousuf Khan

 

[Keith]

Yes, that's exactly my point, it's precisely because of Dell's volumes
that Intel might want to cut it loose! If it's starting to lose money on
Dell, then those high volumes would be a *bad* thing for Intel. Large
volumes and positive numbers, give you large positive numbers. However,
large volumes and negative numbers, give you large negative numbers!
With Apple's small numbers, if it loses money on Apple, then it'll lose
much less money than it would've with Dell.

I can't imagine them voluntarily losing money on *any* customer. Are you
saying that Dell isn't paying at least the marginal cost of Intel
products? That sounds like an anti-trust action waiting to happen.

 

[Yousuf Khan]

I can't imagine them voluntarily losing money on *any* customer. Are you
saying that Dell isn't paying at least the marginal cost of Intel
products? That sounds like an anti-trust action waiting to happen.

Hmmm...



Yousuf Khan

 

[Keith]

Hmmm...

Do you *really* believe Intel is losing money on ever Dell system? I've
accused Intel of being pretty stupid (many times, AAMOF), but never that...

 

[Yousuf Khan]

Do you *really* believe Intel is losing money on ever Dell system? I've
accused Intel of being pretty stupid (many times, AAMOF), but never that...

No, but they probably lose money on specific Dell systems, such as
Celerons. And right now, it looks like Dell is moving a hell of a lot of
Celerons, not a lot of high-end stuff like P4's, and their Xeon business
is under heavy pressure now. Intel can get away with losing money on
Dell's Celerons (assuming Dell doesn't lose money on them), as long as
Dell can push lots of those higher-end stuff too. But these days, Dell
is looking like a Celerons only; Dell even admitted to it a few quarters
ago.

Yousuf Khan

 

[Tony Hill]

No, but they probably lose money on specific Dell systems, such as
Celerons.

I *HIGHLY* doubt that. Remember that it costs Intel VERY little to
produce their low-end chips. Celerons probably cost them only about
$20-$30 to product at most. Dell might not be paying much more than
that, but I would say with a high degree of certainty that they are
paying at least a dollar or two over Intel's cost.

That being said, Intel doesn't like low-margin products like that.
Their average gross margins are up in the 50% range, so when they're
selling lots of chips with gross margins of 10% or less, management
(and shareholders) aren't going to be very happy.

 

[HenryNettles]

I *HIGHLY* doubt that. Remember that it costs Intel VERY little to
produce their low-end chips. Celerons probably cost them only about
$20-$30 to product at most. Dell might not be paying much more than
that, but I would say with a high degree of certainty that they are
paying at least a dollar or two over Intel's cost.

That being said, Intel doesn't like low-margin products like that.
Their average gross margins are up in the 50% range, so when they're
selling lots of chips with gross margins of 10% or less, management
(and shareholders) aren't going to be very happy.

A lot of this is going to depend on their bookkeeping. As I understand it
(and I'm sure someone will correct me if I'm wrong), a very large part of
the cost is setting up the fab in the first place. If you get a large
profit margin in the first few months, then towards the end-of-life, you
can set your books to show that the fab is already paid for, and the cost
of the chips is then really, really low. Alternately, you could allocate a
smaller, but even, amount of that overhead to every chip you produce on
that fab. This would take to make the profit margin higher at the
beginning, but very low towards end-of-life.

I remember a few years back (maybe 12 to 15 years), one of the drive
manufacturers (perhaps Conner?) admitted publicly that they were not
testing drives returned under warranty. Their rational was that it was
cheaper to run the assembly line a few extra hours per week to turn out
drives for RMA, then it was to hire and train people to check the warranty
returns. If you factored in the costs associated with designing the drive
and building the plant, then it wasn't cheaper. But they had already spent
that money, and the assembly line was going idle at night and on weekends,
so why not just run the line a few extra hours?

The way that I recall this, the whole thing became public when someone dug
up a large number of the (still working) returned drives from a dumping
site / landfill. Can anyone with a better memory recall the details?

 

[Del Cecchi]

A lot of this is going to depend on their bookkeeping. As I understand it
(and I'm sure someone will correct me if I'm wrong), a very large part of
the cost is setting up the fab in the first place. If you get a large
profit margin in the first few months, then towards the end-of-life, you
can set your books to show that the fab is already paid for, and the cost
of the chips is then really, really low. Alternately, you could allocate a
smaller, but even, amount of that overhead to every chip you produce on
that fab. This would take to make the profit margin higher at the
beginning, but very low towards end-of-life.

I remember a few years back (maybe 12 to 15 years), one of the drive
manufacturers (perhaps Conner?) admitted publicly that they were not
testing drives returned under warranty. Their rational was that it was
cheaper to run the assembly line a few extra hours per week to turn out
drives for RMA, then it was to hire and train people to check the warranty
returns. If you factored in the costs associated with designing the drive
and building the plant, then it wasn't cheaper. But they had already spent
that money, and the assembly line was going idle at night and on weekends,
so why not just run the line a few extra hours?

The way that I recall this, the whole thing became public when someone dug
up a large number of the (still working) returned drives from a dumping
site / landfill. Can anyone with a better memory recall the details?

The rules for amortization of the capital in the fab are set by the
Government through the IRS and FASB. But what would be in it for Intel
if they accepted abnormally low margins on products they sold to Dell?
I mean, other than possibly an anti-trust case?

 

[Robert Redelmeier]

The rules for amortization of the capital in the fab are set
by the Government through the IRS and FASB.

And those rules must be followed in the accounting done for
taxes and sharesholders (respectively!). The accounting
done for management can (and should) be completely different.
That's more of an economics calculation of alternative uses
and profix maximization (managment science/opations research).

But what would be in it for Intel if they accepted abnormally
low margins on products they sold to Dell? I mean, other
than possibly an anti-trust case?

Ah, there's the huge rub in Adam Smith's marketplace: there
is pressure to sell down to the marginal cost of production.
But how do you cover fixed costs which today loom large?

Intel doubtlessly has some idea about Dell's AMD switchover
point, and will price up to it. Now over, it seems, probably
because the point moved down. It's a poker game.

-- Robert

 

[Robert Redelmeier]

I remember a few years back (maybe 12 to 15 years), one of
the drive manufacturers (perhaps Conner?) admitted publicly
that they were not testing drives returned under warranty.
Their rational was that it was cheaper to run the assembly line
a few extra hours per week to turn out drives for RMA, then it
was to hire and train people to check the warranty returns.

True, but extremely shortsighted. You need to check RMAs
to spot and fix defects before they spiral out of control.
Then sell any false RMAs as "reman".

If you factored in the costs associated with designing
the drive and building the plant, then it wasn't cheaper.
But they had already spent that money, and the assembly
line was going idle at night and on weekends, so why not
just run the line a few extra hours?

The old marginal cost vs fully-allocated cost dilemma.

-- Robert

 

[Keith]

True, but extremely shortsighted. You need to check RMAs
to spot and fix defects before they spiral out of control.
Then sell any false RMAs as "reman".

At $.145/GB, who's going to bother with refurbs? ;-)

The old marginal cost vs fully-allocated cost dilemma.

Rather like sunk cost (AKA losing money but making it up on volume ;-).