Slashdot Mirror
Intel's Per-Chip Cost Averages $40
Fedorpheux writes "According to a report by the analysts at In-Stat, Intel's average cost per chip is about $40. These same chips, such as the Pentium 4s, can cost consumers up to $637. This $40 average cost has remained rather steady since 2003. This cost does not include money spent on marketing or development, but it does explain how Intel can continue its profits even in this era of quickly dropping prices in computer hardware."
The first couple cpu's of a line cost millions and millions to make, they need to charge for the money lost on the first few.
You're kidding right?
I work for a certain chip company too, and I know that we spent over $500 million on the R&D for a single new chip architecture. There's thousands of man-years involved in bew chip design and architecture, and these 'gouged' margins you speak of are the only way to recoup that cost.
Yes it is quite obvious you're not an accountant, no need to actually state that.
$719 - xeon 3.6ghz 604
straight from your site. and while it doesn't say, it's likely that you can find higher end xeons (with gobs of cache) for a few grands
Well, software is a different animal than hardware. For the same product, hardware decreases in price over time. For example, when I purchased my laptop about 3 years ago, a 512MB DDR266 SODIMM stick cost over $250. Now a pair will get you $10 back from your Benjamin. However, Windows XP still costs the same $299 for a full Professional installation CD that it did in 2001 when it launched. And Windows 2000, which the OfficeMax near me still stocks, costs $259. Software is IP while hardware is mostly "nuts 'n bolts." That makes a HUGE difference.
Just "gittin-r-done," day after day.
These are not the same costs. The $40 average is the manufacturing cost. The "UP TO $637" represents the retail price to the consumer. An analysis of any manufacturing industry will yield similar markups (profit margin), although the percentage varies for every industry.
Amigori
"The quality of life is determined by its activites."--Aristotle
Some numbers from their financial report...
For 2004, Intel had a net income of US$7.5 billion on revenue of US$34.2 billion.
Overall tax rate expected for 2005: 31%
(With 2004 earnings as a guide, taxes will be US$10.6 billion)
Their expected R&D budget for 2005 is: US$5.2 billion
Capital spending for 2005: US$4.9-5.3 billion
Overall, Intel pays 31% of their revenue in taxes. 30% in Capital spending and R&D, which leaves 39%, or US$13.4 billion, to pay salaries, benefits, cost of fabrication (not including the facility itself), cover the cost of their bad chips/wafers, and sending some cash to their stockholders.
Reading code is like reading the dictionary - you have to read half of it before you can go back and understand it.
I have a friend who works in IBM's fabs for nVidia chips and he was explaining to me that when you buy a top end chip, you have to pay for all the failed chips produced in order to get a good one. In the case of the 6800, he mentioned numbers along the lines of 20% when the 6800 was new. Obviously, as the 90nm (or 120, I forget) matured, this number goes up, but even so, they have to offset 4 other failed chips for every chip they ship.
This is probably not as bad for x86 chips, as they can just underclock less well fabed chips, but the point remains: at $40 a pop failure can get expensive fast. The article mentions that the $40 figure doesn't take this into account...it is a fairly big omission, IMHO.
Coupled with them ignoring other huge expenses like the entire cost of the design of the chip, $40 seems kind of high. I wonder if it takes into account the creation, operation and maintenance of the fab facilities. I get the feeling they are simply pricing the cost of raw materials here, and the article is skimpy on details about what exactly IS included.
Take with a healthy dose of salt, I'd say.
Alot of people are correctly pointing out the sloppy 'news' reporting by slashdot these days by pointing out the costs or R & D, marketing, etc... that should also be factored into the per-chip cost.
Well here's some Intel Financial Data. Please use it responsibly. Surely somebody with some smarts can use this to determine a 'real' per-chip cost.
Look at Intel's profit margin, it's below 25%. http://finance.yahoo.com/q/ks?s=intc/. There are lots of expenses in running a company. Only the fastest P4s are in the $600+ region, many are below $100 retail. http://www.pricewatch.com/. The cheap ones cost every bit as much to make as the expensive ones; it's just a matter of supply and demand.
Contribute to civilization: ari.aynrand.org/donate
It's more than marketroids, although they do seem to excede the scientists in cost. If you look at the last reported quarter, Intel grossed $9.2 billion of which $2b was profit. Of note $1.17b went into R&D (1.34b into "selling and adminstrative"). That's still a decent profit -- about 22%, but it isn't like they profit $540 on every $40 of expenses (1350%). Intel Balance Sheet
... but those lower prices definitely show on the bottom line.
Compare that to AMD's rather ugly results. Only $11 million in profit (and two of the last four quarters were losses). Still, the last Intel chip I "bought" (it was part of a laptop) was a 486sx20. I love AMD's stuff and their prices
What changed under Obama? Nothing Good
"We make higher margins on those than the US Treasury does making dollar bills".
The margin on today's chips is nowhere near that high.
Seriously, though the comments about R&D and marketing costs are on track, but leave out an important one: for each new generation of chip, one or more entire fabs (manufacturing lines) need to be built. Lately this costs $2bn (yes, billion) or more. When the next chip process comes along, the whole plant is essentially thrown away (yes, in reality it gets used for down-rev chips, but the lifetime isn't long). The difference between the actual capital deprecitation of these and the real cost/lifetime is another "hidden" component of chip cost. This applies pretty much equally to anyone making cutting-edge chips, including AMD.
One of the reasons AMD stayed so far behind for so long was that its chips, generally a generation behind Intel (in the 1990s) didn't generate enough profit to build these truly leading-edge fabs. The "treadmill" as it was known at Intel, ran too fast for them to catch up. When the market hiccuped in 2000, things changed. Before that was a truly fine time to own lots of Intel stock options.
-- gnet
Close, but to be quite accurate, it's normally around $5.4 billion (I work for a big european pharma). And around 10 years. Also, bear in mind that most project tend to be binned in the last few years (stage three clinical), so most of that money been spent by then. But, yeah, the analogy is a good 'un.
Give a man a fire, and he's warm for a day. Set a man on fire, and he's warm for the rest of his life. (Terry Pratchett)
Intel did not give away their patents, they were forced into a cross-licensing agreement during arbitration with AMD.
"So what's the per-chip cost WITH all of the overhead?"
No need to ask. We can deduce this using some basic Internet research skills and some junior high-level math.
If they amortize overhead equally across all products, you can guesstimate it this way (we'll use a part that costs $600 at retail for an example):
So, for that $600 part you buy:
Sitting in my day care, the art is decopainted.
i'm assuming this isn't limited to pentium 4 chips, but includes all chips that intel makes.
I don't think you fully appreciate the size of Intel, or the sheer number of chips they make...
P4s count for less than 1% of their sales, in terms of volume. CPUs in general almost certainly make up less than 10%.
The vast majority of Intel's output consists of things like opamps, ethernet controllers, simple logic chips, and other trivial (compared to a modern CPU) ICs that mostly cost well under a dollar (to buy, not to make) each.
No, their average cost per chip, over their entire product line, does not come out to $40. Not even close. That would bankrupt them in a week, selling X chips at $600 while selling 95X at $1.
That said, the $40 figure certainly does not take the total cost into consideration. Perhaps the raw materials, electricity, and immediate labor to produce them once everything has fallen into place. But just the cost of building a new fab (in the low billions), or retooling an old one for a new process (hundreds of millions) far outweighs the ongoing per-unit production costs.
Sorry,
They don't make simple logic chips, nor do they make op-amps.
Intel is not in the business of making discreet components.
If you anre interested the Adjusted gross margin of the company (based on previous SEC filings) is roughly 48-52%. That would indicate that a $600 wholesale chip cost the company about $300 to produce, with the other $300 going to expansion, investment in new tech, shareholders dividends, into a bank account for a rainy day, etc.
I would tend to think that most of the profit is mature tech for which the R&D has been ammortized, such as the P3 chips, Xscale, etc.
I have a fairly reliable way of thinking that the P4 division, while a profit center, is not where the big money comes from. Most of that $400-$600 you spend on a CPU is covering other people's costs. (remember retailer markup)
whois gawk date unzip strip find touch finger mount join nice man top fsck grep eject more yes exit umount sleep dump