so Jobs gets screwed by IBM over game consoles, thus Apple-Intel ?

[Robert Myers]

(e-mail address removed) wrote:


"PC Chips" would not include processors for embedded applications,
which AMD also makes.

That's a good question actually; remembering that margin = profit/cost
and profit = revenue - cost. AMD doesn't pay for the fab twice, but
the value of the fab has clearly depreciated quite a bit by the time
its used for flash. So there is a question of how to assess the fab
cost. Presumably in the books they "sell it to themselves at a
depreciated cost", but they could practically make up any number (up to
the original cost of the fab) I would think.

There _is_ a great deal of arbitrariness in cost of goods sold figures,
just like there is a great deal of arbitrariness in earnings and
profits. That doesn't mean the numbers are irrelevant.

Other than that, flash costs are much lower. They don't have on-going
design costs related to "cranking the frequency"; flash is a kind of
memory array, so via proper redundancy in the design, there is no
reason their yeilds couldn't be as high as they want; they don't have
to pay for a "compatibility lab" with anywhere near the complexity of
those for their x86 CPUs. They are not paying retail marketing costs,
etc.

Of course AMD is able to play a lot of games with CPU prices via market
segmentation, but all this has to be counted against higher costs and
intense competitive pressure from Intel.

So which unit has actually higher *margins* is not clear to me. Flash
may be lower revenues and lower cost at the same time, versus CPUs
which have higher revenues and costs at the same time.

Either way, figuring out the "margins" of the CPUs seperated from the
flash seems like a silly kind of exercise, since AMD's goal clearly is
to maximize net profits, not individual margins.

Frustrating it may be, but not silly. Manufacturers and investors want
to know which products are carrying their weight and which are not.
Should the product be continued, should more investment or advertising
be put into the product, is the selling price right...all kinds of
issues.

The numbers in Forbes certainly look suspicious, but that doesn't mean
that knowing one manufacturer's gross margin for a particular product
line versus another manufacturer's isn't something investors would and
should be interested in.

AMD is doing okay right now. If they keep doing okay, maybe we'll
never find out what those numbers mean. If there is trouble, we'll
hear more. Intel's gross margin numbers, as I'm sure you know, are
constantly under scrutiny, as is the profitability of its different
product lines. Managers and investors _do_ look at whether individual
products make money and how much. That seems so obvious that it
doesn't seem as if I should have to write it.

I.e., without other
factors, such as total product shipped for each unit, and full
disclosure of costs, and without knowing how AMD is "dividing up the
fab costs" (or other shared costs), some of which I am sure are
available from the SEC filings, its both hard to put "margins for each
product" into perspective or even know what they are.

If you've got serious money to spend, you can buy a report with all
that stuff worked out for you and with the methodology documented.
We're talking about an article in a mass market magazine here. You get
what you pay for.

RM

 

[Travelinman]

That's a good question actually; remembering that margin = profit/cost
and profit = revenue - cost.

Actually, no.

Let's go through this again. The following is simplified, but if people
posting here would learn it, it would eliminate about 95% of the
pseudo-economics that appears here so frequently.

Revenue is the amount of money you receive for your product or service.

Cost of sales is the cost that can be DIRECTLY attributed to each unit.
That is, in principle, if you don't ship anything, cost of sales would
be zero. This usually includes direct labor and direct materials. A
small amount of manufacturing costs (electricity, for example) also goes
here.

Overheads are your fixed costs. That is, costs that you have whether you
ship anything that month or not. Typically, this includes selling costs,
R&D costs, plant depreciation, and admin costs.

There are a very few items that can be included in a couple of different
places. As long as you're consistent, you get a good bit of leeway. For
example, commissions are often included in the overhead bucket, although
I think they should really be in COS. SEC doesn't care which one you use
- as long as you're consistent. This is one of the reasons that you have
to be very careful in comparing gross margins from different companies.
They may be accounting for some overheads differently. That problem
doesn't occur if you're looking at net income.

OK, now for the math:

Gross margin = revenue - cost of sales

Gross margin percentage = gross margin / revenue

Net income (or pre tax profit) = gross margin - overheads
or
Net income (or pre tax profit) = revenue - cost of sales - overheads


It's really not that hard, but people constantly insist on getting it
wrong - like Nasty's confusion between gross margin and profit.

 

[Travelinman]

What is a "bogey" cost? Bogus? There are many.

Cost accounting is a serious discipline, and determining a
valid cost of a manufactured good depends entirely on what
the number will be used for.

The biggest distinction is between marginal cost (the increase
in costs due to producing one more) versus "average" cost,
total costs divided by total number produced. Byproduct credits
(up/down binning) and IP use complicate an already difficult calc.

Not actually correct.

In an appropriate costing system, the cost of sales only includes the
direct costs, not overheads. Within fairly wide ranges, the direct cost
isn't all that dependent on the number produced.

Yes, there are some efficiencies for making larger quantities and you
will get price breaks if you buy enough of some components (although
there are cases where per unit costs actually go UP if you produce
more), but these effects are generally rather modest in normal business
plans. You might increase production from 1000 units to 1500 units -
which will not greatly affect your costs. It is very unusual (other than
a new product launch) to go from 1,000 units to 50,000 units in a short
time.

 

[George Macdonald]

I wish somebody would explain it to me. Where did that number come
from and what does it mean? Even inside a mnufacturer, manufacturing
cost can be pretty arbitrary. Did Forbes manipulate the number?
What's the motive?

Motive?... it's just anal..yst hubris - no motive is required! As for
"explain", how can you fail to have read the accounts, at the end of the
1st quarter, of how AMD had made a profit on processors and been sunk by
losses on flash memory? Did you miss the fact that Hector then decided
that the flash memory unit must be spun off? Kinda hard to make a profit
on a line of products with a 12% gross margin.

If nothing else, George, if Forbes can put that number out there and
expect not to get blown out of the water, it explains a great deal
about AMD's credibility problems. Don't blame it on me. I don't
publish any widely-read financial publications.

But you did cite a source which was more credible than Forbes on overall
gross margin and we umm, knew(?) that flash was a loser - even Intel said
so. Analysts are always making gross errors in their estimates and
fabricating projections which are no better than lies. Are you forgetting
the projections made by IDC about Itanium sales over the years? Did they
correct them? Did they get "blown out of the water"?... nope - they are
still considered by many as a "credible source".<shrug>

 

[Robert Redelmeier]

But so what? Manufacturers rely on cost of goods numbers
because they have to. So do analysts. Knowing how the
number was made is at least as important as the number,
but so what? Unless you can get inside AMD and examine
the books, you, and I, and nearly everyone else, is going
to have to take what appears in the media.

Hardly. Manufacturers need to know all aspects of their costs
and particularly how they vary in order to make good decisions.
Competitors need to be able to read others costs to know how
they can compete. Fortunately, they use much the same techniques
and can read a lot more information from press releases than
less-informed people. But so can good industry specialists.
Unfortunately, they get heavily edited.

For a concrete example, on some measures, a P4EE might
cost about $10 to make. An Itanium on a different cost
basis might cost $2000.

So what? This is all idle speculation on our part. We are
not making manufacturing decisions. We might be trying
to forsee where the market is going, and believe that
cost will tell us. I'm not so certain.

-- Robert

 

[Robert Redelmeier]

In an appropriate costing system, the cost of sales only includes
the direct costs, not overheads. Within fairly wide ranges,
the direct cost isn't all that dependent on the number produced.

"appropriate costing system"? Just where do you think such a
thing exists? Any why would such a jewel ever be made public?
There is a huge spectrum of cost types, all with different
variability [often stepwise] on production. Drawing a single
breakpoint is lunacy because it has to move according to the
decision at hand.

Yes, there are some efficiencies for making larger quantities
and you will get price breaks if you buy enough of some
components (although there are cases where per unit costs
actually go UP if you produce more), but these effects
are generally rather modest in normal business plans. You

I'm not sure what you are saying. In the semiconductor biz,
in the short run all costs are fixed and marginal cost of
production is simply HP silicon, other materials and chemicals.
When lines get full, then you have to add shifts, and when
those get full, more machines and lines. You still don't
need new masks, but those sunk development costs have to be
paid somehow. In the long run, all costs are variable.

-- Robert

 

[Robert Myers]

Motive?... it's just anal..yst hubris - no motive is required! As for
"explain", how can you fail to have read the accounts, at the end of the
1st quarter, of how AMD had made a profit on processors and been sunk by
losses on flash memory? Did you miss the fact that Hector then decided
that the flash memory unit must be spun off? Kinda hard to make a profit
on a line of products with a 12% gross margin.


But you did cite a source which was more credible than Forbes on overall
gross margin and we umm, knew(?) that flash was a loser - even Intel said
so.

Intel makes money on flash, as far as I know, or at least it's
frequently a part of the picture when Intel shows improved
profitability. I don't follow AMD. I probably read the story about
AMD and flash and just didn't remember it. AMD just _isn't_ that
interesting to me.

That's because I don't see them as a long-term player. They're too
small and their fortunes look too dicey to me. Take a look at the
history of commercial airplane manufacturing in the US.

Analysts are always making gross errors in their estimates and
fabricating projections which are no better than lies. Are you forgetting
the projections made by IDC about Itanium sales over the years? Did they
correct them? Did they get "blown out of the water"?... nope - they are
still considered by many as a "credible source".<shrug>

A projection is different from an estimate. Anybody can get a
projection wrong. Happens to the best every day. An estimate that
improperly represents the current operations of a company is a
different matter.

I'm as suspicious of the motives behind the Itanium projections as you
are. Microprocessor Reports has said some things about Itanium that
seem downright weird (Microprocessor of the Year?). What do I know? I
only know what I read in the papers.

On the other hand, maybe the insiders aren't just a bunch of fools.
When the big guys decline AMD as a vendor, it means something. When
the analysts refuse to take the consensus advice of csiphc that Itanium
is dead, that means something, too.

What this "gross margin" number from Forbes means, I don't know.

RM

 

[Del Cecchi]

Intel makes money on flash, as far as I know, or at least it's
frequently a part of the picture when Intel shows improved
profitability. I don't follow AMD. I probably read the story about
AMD and flash and just didn't remember it. AMD just _isn't_ that
interesting to me.

That's because I don't see them as a long-term player. They're too
small and their fortunes look too dicey to me. Take a look at the
history of commercial airplane manufacturing in the US.


A projection is different from an estimate. Anybody can get a
projection wrong. Happens to the best every day. An estimate that
improperly represents the current operations of a company is a
different matter.

I'm as suspicious of the motives behind the Itanium projections as you
are. Microprocessor Reports has said some things about Itanium that
seem downright weird (Microprocessor of the Year?). What do I know? I
only know what I read in the papers.

On the other hand, maybe the insiders aren't just a bunch of fools.
When the big guys decline AMD as a vendor, it means something. When
the analysts refuse to take the consensus advice of csiphc that Itanium
is dead, that means something, too.

What this "gross margin" number from Forbes means, I don't know.

RM

Seems like IBM just added some AMD stuff.
http://news.yahoo.com/news?tmpl=story&u=/nf/20050616/bs_nf/36493

 

[TravelinMan]

In an appropriate costing system, the cost of sales only includes
the direct costs, not overheads. Within fairly wide ranges,
the direct cost isn't all that dependent on the number produced.

"appropriate costing system"? Just where do you think such a
thing exists? Any why would such a jewel ever be made public?
There is a huge spectrum of cost types, all with different
variability [often stepwise] on production. Drawing a single
breakpoint is lunacy because it has to move according to the
decision at hand.

Where did I claim that there was a single breakpoint?

I said that you needed an appropriate costing system - which means
'appropriate for that business'. The costing system for McDonald's is
undoubtedly different than the costing system for Boeing.

HOWEVER, the general principles are the same.

I'm not sure what you are saying. In the semiconductor biz,
in the short run all costs are fixed and marginal cost of
production is simply HP silicon, other materials and chemicals.

Which is why semiconductor companies have such high gross margins.

When lines get full, then you have to add shifts, and when
those get full, more machines and lines. You still don't
need new masks, but those sunk development costs have to be
paid somehow. In the long run, all costs are variable.

You just managed to contradict yourself. The semiconductor fab cost is
not variable, it's fixed - at least in accounting terms.

 

[Robert Redelmeier]

You just managed to contradict yourself. The semiconductor fab
cost is not variable, it's fixed - at least in accounting terms.

Those "accounting terms" [GAAP & FASB} are part of the problem.
A bigger problem is trying to use one set of numbers for everything.

With short economic lifetimes, most of the fab costs are variable.
You decide to build a new one/line depending on projected sales.
Or shut'em down. They vary in a longer timeframe than one year,
but they vary. Everything varys, the question is: How quickly?
and how do we amortise sunk costs?

-- Robert

 

[TravelinMan]

You just managed to contradict yourself. The semiconductor fab
cost is not variable, it's fixed - at least in accounting terms.

Those "accounting terms" [GAAP & FASB} are part of the problem.
A bigger problem is trying to use one set of numbers for everything.

No one ever suggested using one set of numbers for everything.

It IS reasonable to use one set of accounting rules and then apply them
differently for different types of companies. That just happens to be
what GAAP and FASB do.

With short economic lifetimes, most of the fab costs are variable.
You decide to build a new one/line depending on projected sales.
Or shut'em down. They vary in a longer timeframe than one year,
but they vary. Everything varys, the question is: How quickly?
and how do we amortise sunk costs?

Unfortunately, your kind of thought process is why some execs are now
sitting in jail.

 

[Robert Myers]

Seems like IBM just added some AMD stuff.
http://news.yahoo.com/news?tmpl=story&u=/nf/20050616/bs_nf/36493

<quote>

Support for Opteron is part of IBM's open-specification initiative for
offering better custom Blade Center options, Willeford said. "AMD is
now one of 250 companies that have received the specifications and are
using them to create new products."

The advantage of this strategy, said Yankee Group analyst Andy
Efstathiou, is that businesses running systems based on AMD's chips now
can plug IBM blades into those systems without the hassle of converting
their applications. "That means a reduction in costs for those on
legacy systems configured for AMD chips," he said.

</quote>

Note paragraph 2 for those csiphc regulars who have been hassling me
about amd, intel, who cares as long as it's x86.

IBM has, I gather, forced everyone, including Intel, to meet it
halfway. If Intel wannts its chips in IBM blades (and I'm sure it
does), it has to satisfy IBM specifications. That levels the playing
field a bit for AMD.

RM

 

[Del Cecchi]

<quote>

Support for Opteron is part of IBM's open-specification initiative for
offering better custom Blade Center options, Willeford said. "AMD is
now one of 250 companies that have received the specifications and are
using them to create new products."

The advantage of this strategy, said Yankee Group analyst Andy
Efstathiou, is that businesses running systems based on AMD's chips now
can plug IBM blades into those systems without the hassle of converting
their applications. "That means a reduction in costs for those on
legacy systems configured for AMD chips," he said.

</quote>

Note paragraph 2 for those csiphc regulars who have been hassling me
about amd, intel, who cares as long as it's x86.

IBM has, I gather, forced everyone, including Intel, to meet it
halfway. If Intel wannts its chips in IBM blades (and I'm sure it
does), it has to satisfy IBM specifications. That levels the playing
field a bit for AMD.

RM

Yep, and intel is in on it....
http://www-1.ibm.com/servers/eserver/bladecenter/open_specs.html

 

[Robert Redelmeier]

Unfortunately, your kind of thought process is why some
execs are now sitting in jail.

Figures don't lie, but liars figure.

I never said numbers couldn't be misused. It fact it is
very hard _not_ to misuse them. And impossible if you do
not understand their origins and assumptions. Costs are
especially difficult to understand. The question becomes
what is included?

-- Robert

 

[George Macdonald]

Intel makes money on flash, as far as I know, or at least it's
frequently a part of the picture when Intel shows improved
profitability. I don't follow AMD. I probably read the story about
AMD and flash and just didn't remember it. AMD just _isn't_ that
interesting to me.

As I recall, Intel made a loss on flash in the same period - they were
dumping it to buy share of a shrinking market as the flash market was
playing its cyclical accordion. The whole flash market has been badly hurt
by lack of success of 3G.

That's because I don't see them as a long-term player. They're too
small and their fortunes look too dicey to me. Take a look at the
history of commercial airplane manufacturing in the US.

You mean Boeing? They are getting thrashed by Airbus now - it's a global
market and the Euros have decided they'll break the rules just like the
U.S. has been doing for decades. Does this mean Boeing will now collapse?

A projection is different from an estimate. Anybody can get a
projection wrong. Happens to the best every day. An estimate that
improperly represents the current operations of a company is a
different matter.

But this particular projection was repeatedly and grossly overstated to the
point of absurdity.

I'm as suspicious of the motives behind the Itanium projections as you
are. Microprocessor Reports has said some things about Itanium that
seem downright weird (Microprocessor of the Year?). What do I know? I
only know what I read in the papers.

On the other hand, maybe the insiders aren't just a bunch of fools.
When the big guys decline AMD as a vendor, it means something. When
the analysts refuse to take the consensus advice of csiphc that Itanium
is dead, that means something, too.

WTF do you mean "decline"? There are sufficient "big guys" taking up AMD
to make that a downright lie. As you say, you're just not interested in
AMD so you apparently fail to notice what is going on and assume that Intel
is king.

 

[Robert Myers]

You mean Boeing? They are getting thrashed by Airbus now - it's a global
market and the Euros have decided they'll break the rules just like the
U.S. has been doing for decades. Does this mean Boeing will now collapse?

Nah. Too important to the DoD. For now, anyway.

WTF do you mean "decline"? There are sufficient "big guys" taking up AMD
to make that a downright lie. As you say, you're just not interested in
AMD so you apparently fail to notice what is going on and assume that Intel
is king.

Now, George. Temper, temper. Your opinion, or mine, won't make the
slightest difference to AMD's fortunes, and this isn't, at least
ostensibly, a stock-pumping group.

If I enjoyed baiting you, it would be much too easy. I really don't
enjoy baiting you, so I'm sorry if my choice of words, or my mastery of
the facts, upsets you.

RM

 

[Del Cecchi]

Nah. Too important to the DoD. For now, anyway.


Now, George. Temper, temper. Your opinion, or mine, won't make the
slightest difference to AMD's fortunes, and this isn't, at least
ostensibly, a stock-pumping group.

If I enjoyed baiting you, it would be much too easy. I really don't
enjoy baiting you, so I'm sorry if my choice of words, or my mastery of
the facts, upsets you.

RM

Gee, I thought that at the moment Boeing was kicking Airbus' behind.
Weak dollar, strong euro, product cycle etc. Their stock is at a 5 year
high, having doubled in the last couple of years off the march 03 low.

What was that about mastery?

del cecchi

 

[Robert Myers]

Gee, I thought that at the moment Boeing was kicking Airbus' behind.
Weak dollar, strong euro, product cycle etc. Their stock is at a 5 year
high, having doubled in the last couple of years off the march 03 low.

What was that about mastery?

I never predicted the demise of Boeing. The Boeing competition with
Airbus has as much to do with politics as it has to do with business.
If it gets uglier than it already is, I'm sure the US will do what it
has to do.

I was referring to the shakeout and consolidation that have left Boeing
as the last man standing producing full-scale commercial passenger
aircraft in the US.

As to who is kicking whose butt right now, it depends on who you ask
and when. For example:

http://www.przoom.com/view.php?id=817&type=search

RM

 

[ZnU]

You mean Boeing? They are getting thrashed by Airbus now - it's a global
market and the Euros have decided they'll break the rules just like the
U.S. has been doing for decades. Does this mean Boeing will now collapse?

The Boeing/Airbus rivalry is about to get a lot more interesting. The
two companies have a fundamentally different view of the future of air
travel. Airbus wants to focus on the present hub-based system, so they
keep building bigger planes. Boeing thinks the model of the future is a
more flexible strategy based around point-to-point flights (with smaller
planes).

We'll see who's right.

[snip]

 

[George Macdonald]

Gee, I thought that at the moment Boeing was kicking Airbus' behind.
Weak dollar, strong euro, product cycle etc. Their stock is at a 5 year
high, having doubled in the last couple of years off the march 03 low.

Airbus has been crowing about out-selling Boeing and poaching their orders
at the recent Paris airshow, with large orders of single aisle and even
double-aisle mid-range and of course they think they're going to make a
kill as the sole supplier of elephantine-sized long rangers. I have my
doubts myself and tend to think their elephant is going to turn out to be
err, white.