Hands off Intel

Pitzer was later seen leaving the meeting carrying a huge sack slung over his shoulder with a dollar sign printed on it, containing what we can only assume to have been potatoes.

This is an analyst that can even crunch numbers...

Atom might have a pretty hefty markup but Intel is aggressively selling Atom + chipset bundles which INQ also reported on:

http://www.theinquirer.net/inquirer/news/1137394/nvidia-attacks-intel-pricing-advantage

Intel seems to be selling at a lost in case cited above:

Atom $45
Atom + chipset $25

Assuming Atom is marked up by 100% (come on this is supposedly a low end part), it costs around Intel $22.5 to make Atom and most chipsets cost around $10 to make... Even if Atom was marked up by 200%, Intel is making $3.75 on the bundle.

Then again, even if netbook market size hits 100M per year and Intel actually make $20 on average from Atom, that is still "only" $500M/quarter... less than 5% of Intel quarterly revenue...

Really, Atom is going to make or sink Intel! Pfft!

Intel might not be doing so well when they end up shelling out Billions for violation of anti-trust laws in the EU and US.

Intel started about 750 Nm & advertised itself as Biggest Chipmaker in Business. Indeed Intels scale was biggest, NOT Best, or evem Good. Since; mountain of chips to tune refrigerators, balance paint brushes & Water your Lawn @ 2 AM, are being thrown out, ALL Paid for, In Advance, ?BY:YOU.As Us Law Allows Manufacuter to have its Banker Credit Account for Lost sales, even forgein. So While Gallapagous Islanders never Actually Recieved their Chips, YOU Paid For them & Others Wasted Lifetime Making them.

Don't Forgot Those SOCKS that Bark. 5 Pairs Ea.

drashek

Hmmm,
analysts are not always right.

I mean, $22 for an Intel share, that is nothing,
I mean this is almost a free lunch (which does not exist as they say).

Simply consider the growth of this planet's population:

Region 1950 1999 2008 2050
World 2,5 5,9 6,7 8,9
Africa 0,2 0,7 0,9 1,7
Asia 1,4 3,6 4,0 5,2
Europe 0,5 0,7 0,7 0,6
Lat. Am. 0,1 0,5 0,6 0,8
N.America 0,1 0,3 0,3 0,4

From now on till 2050 the world population is not only expected to increase by the sheer number:
the education level, the living standard and the demands of the individuals in existence will increase dramatically.
(Maybe not true for Europe, who knows ...).

So there will be a lot of individuals who demand to have the latest and greatest Communication Devices a la Starship Enterprise to take advantage of Facebook, Twitter and online-gambling and porn and also the other more business-realted stuff.

(When it comes to Starship Enterprise I would prefer to food-replicator, but not progress in this space yet :-)

Anyways, electronics in general and Intel (and AMD) plus cohorts will benefit greatly.
For all societal, scientific and personal advancements you do need IT technology.

So $22 is really peanuts.
$44 is much more appropriate number for the Intel shares.

You also need to take in consideration:

Because of the current (or is already past?) crises there is a very urgent need to grow beyond what we've seen up until now in order to compensate for all the new money that have been printed recently.
Growth (of and in other regions of the world) is the only way out.

Therefore new markets need to be opened and developed in order to provide more people with the goodies of modern life.

You will see this happen.

$22 to make an atom chip? Any source info on that? (it's ridiculously off)

A 45nm wafer costs anywhere from 3200-5000 per wafer in production cost (Sematech benchmark). There are several THOUSAND atom die per wafer... even if you assume POOR yields (which wouldn't be the case on a product so small) and assume <50 s and revenues (which is what news folks tend to focus on), but from a profit/earnings perspective these are better than the low end chips they are displacing even if you assume 100% cannibalization (which is another flawed assumption)

My main point was, Atom is not going to make or break Intel. I would not buy Intel stock based on the outlook of Atom alone.

$18.86 is lower than the 52week high and certainly lower than $25~$28 when Intel was doing better in recent years (and whole lower than $75 during the tech bubble but that was the tech bubble). Now that might make me consider buying Intel stock... That is if I believe Intel's current outlook. Atom? No.

$500M/quarter in should be gross margin and not revenue. Assuming Intel makes $20/unit after cost for Atom bundles and netbook market growing to 100M units/year (which is a large stretch). That actually translate to a hefty 20% of net income for Intel assuming they get back to about $2/quarter level. Still this is highly unrealistic and 5% is more likely.

I was being hasty and admittedly the $22.50 (100% mark up) is pretty far off, but $11.25 (200% mark up) is not. Wikipedia lists N230 at $29 and 25.96mm^2 and N330 at $43 and 51.92mm^2... doing some not so realistic math of: [(300mm/2)^2*PI]/(die size) * 50% yield gives ~1361 and ~680 dies/wafer respectively. 50% yield seems low but you never get to use 100% of the wafer at the edge and spacing gaps (actually a die/wafer estimator gives lower die/wafter). That translate to ~$3.67 and $7.35. The $45 is closer to the N330 so $7.35 and $11.25 is not that far off. So Intel makes like ~$5 on bundles ($25 - $10 for Atom - $10 for chipset)? Not that much over $3.75. Again, I used $20 profit per unit: $20/unit * 100M units/year * (1 year/4 quarters) = $500M/quarter.

$20/unit is 4x $5 and netbook market size is projected to hit 50M units per year in 2010 and "only" 25-30M this year so Intel would be lucky if they made $125M/quarter in profit from Atom. Which is about 5% of gross margin (assuming $2B/quarter without Atom).

Even at $500M/quarter, if Intel does poorly in its core business and that bleeds into this profit, it will not be enough to give Intel a positive spin to even keep stock prices flat let alone drive it up.

Funny name though. :p

OK, the inq's now is a part of an evil hegemony. SpIntel is everywhere now to infested many high rating clicks news websites to make favourable news to them. I'd like to see current and future editorial policy since SpIntel have forced theincisive to fired Paul Hales.

Basically every assumption Lans is making is based on shaky ground.

Assumption A:
An Atom/chipset bundle is indeed $25. Using an INQ link seems to strengthen this #, but the # is based solely on some quote from the Nvidia CEO, who would never distort or misrepresent the facts to make things look favorable for his company. :) We also have no idea what specific Atom he is referring to (single core, dual core? If he was sayint netbook than it's most likely single core) But let's assume this is all 100% accurate for the time being.

Assumption B:
A chipset is $10... based on... well it just is. It has to be around here because someone wants the Atom to cost in the 10-11 range and if a chipset is 10 then that gets you to the desired outcome (~$5/bundle profit)... Does a chipset cost $10 to make? Who knows, but it sounds good, right?

Assumption C:
Atom is in the 10,11, or whatever range... again just because... when confronted with some actual #'s, the #'s come out to $3-4 for a single core and $7-8 for a dual core... so let's just assume we're talking dual cores and then lets also round the WORST case # up a little more , let's say just a little 50% to get it close to $10 or the original 11.25 claim (you know the predetermined, desired outcome)

Assumption D:
Use every unfavorable calculation to get the cost # as high as possible. 50% is too low for yield, well we'll talk about partial die and say well 50% takes that into account and then we're justified in using it.... and poof 1360 die per wafer, we'll then use the highest wafer cost possible to get the cost/die as high as possible.

Well is sounds good until you use a actual full die per wafer calculator and come up with 2350-2550 FULL die per wafer, even when factoring in manufacturing constraints (edge exclusion, spacing between die, elimination of partial die). Now the yield on full die is NOT 50% so you will end up with far more than the 1360 die per wafer... You could also not choose to divide this by the highest wafer cost possible in the range I gave (Intel, which does not use SOI substrates and is helped by their large economy of scale is actually far more likely to be on the low range of what I gave). So you could just as easily be talking a $3500 wafer and 1750 die per wafer or ~$2 for single core Atoms (roughly double for dual cores). Again throw packaging cost into this and I'm still not even getting close to the $10/atom assumption.

Bottom line - given the complete lack of any grounding of any of Lans #'s and his strong desire to fit everything into a nice neat $5/bundle package... well there's 3 main variables in the math... and folks can decide for themselves:
1) The price is based on the say so of Nvidia's CEO who is trying to use this as basis of a lawsuit

2) The chipset cost - completely plucked out of the air

3) The Atom cost - which even Lans acknowledged was 'off' from his initial (gut?) calculations and even the revised #'s are stretched more than the skin on Joan River's face to tweak it to $10, but it still remain off and base on really poor inputs with constant use of worst case assumptions.

PS - I love the translates to 3.67 and 7.35.... so that's pretty close to the original $11.25.

OK if by 'pretty close' you mean at least 50% off and possibly up to 200% off. The difference of 7.35 and 11.25 would roughly DOUBLE the net profit from your original math... but hey.. pretty close...

And I'm not even an analyst... but I'm also not working every # to fit it to an initial gut feeling.

"Does a chipset cost $10 to make? Who knows, but it sounds good, right?"

I haven't gotten out a ruler and measured the thing, but I'm thinking more than 1000 good chips per wafer and a less than $2500 wafer cost. 65nm is a prehistoric process, TSMC practically gives it away, so Intel probably internally charges much less than this. I'm also thinking about two bucks for packaging and test.

Lets just round it up to $5.

First of all, care provide your numbers on how Atom will save Intel if its core business turn sour and how Atom will keep stock prices flat or even drive it up in such scenario? Attack me/my numbers all you want but you have never addressed the most crucial point of my posts.

Second, all that calculation was to establish a FLOOR. From which I based that $20 profit probably a good CEILING. Do you want to argue $20 profits is too low for the average profit per unit from Atom? Go ahead! :-)

My point is there is plenty of all this "selling at a loss" and "killing margin" armchair analysis on the web and folks haven't taken the time to put in some realistic #'s to understand how cheap these things are to produce and you need to stop making up #'s under the guise of analysis when it is clear you have no insight in that area. Your #'s aren't a "floor" they are just made up and plucked out of the air. They are not supported with any factual information and flat out wrong... you don't even know what the $25 bundle price is (it's just some quote from the Nvidia CEO with an axe to grind and you have no clue if it is referring to single core or dual core atom bundles or even if the # is real and accurate)

Coming to some grand conclusion that if 90+% of your business goes sour the remaining 10% (or less) won't make up for it is truly stunning insight. Next you will be predicting the direction the sun will rise in tomorrow.

If you re-read the article, you may realize that while Atom was mentioned, he also talked about overall demand for chips (which personally I'm skeptical about).... of course you used the Atom comments and twisted it to become the entire basis of his conclusion and then made up some #'s to.... why again were you just randomly plucking #"s out of the air... oh that's right to establish a "floor"

He wasn't saying Atom would fuel Intel's overall growth - you just chose to interpret it that way. He is one of the few analysts who does seem to understand the difference between low ASP and low margin and that while atom may lower ASP's, that doesn't mean it is hitting margin and profits.

I have to disagree on Atom not fueling Inte's overall growth. I view that as the whole point of the article.

Intel already claimed demand was picking up so you either believe them (i.e. implies you believe stock prices go will up) or don't (i.e. implies you believe stock prices will not go up).

Article clearly state he is talking to "the naysayers".

Based on that implicit proposition, Intel's core business will either be flat or down slightly moving forward (i.e. what I said by "core business goes sour").

Also, my estimates may be off but is not unrealistic:
http://www.tomsguide.com/us/intel-atom-silverthorne,news-937.html

Though, I'll stop arguing with you on that until you are provide your own numbers to the contrary.

You can also see that the $500M/quarter net income is reasonable ceiling if you look at Intel's Q1 and Q2 2009 quarterly earnings report:

Q1: Revenue from Intel® Atom™ microprocessors and chipsets was $219 million

Q2: Revenue from Intel® Atom™ microprocessors and chipsets was $362 million

They had a gross margin of ~50%... so that means $1B/quarter in revenue from Atom or if you preferred, 100% margin which still requires $500M in revenue.

How should this convince "the naysayers" that stock prices will go up if Intel's core business doesn't improve as they claim?

The analyst did not say Atom will fuel Intel's growth by itself. I did not say Atom will fuel growth either... apparently you keep reading into this somewhere. HAVE YOU EVEN READ THE ANALYST REPORT? Or are just just taking 2-3 sentences the INQ has decided to include and assume they are linked the way the INQ has done?

The analyst points out the margins on these chips are very good. I agree and gave you some #'s to show your complete lack of knowledge in this area with your RIDICULOUS and NON_GROUNDED cost estimates. Numbers which you no longer are disputing despite your creative 50%-200% rounding up things as being "close"...

Something that makes up 5% of Intel's revenue is obviously not going to fuel 10 or 20% overall growth - thanks for pointing that out Captain Obvious. Noone said Atom alone would, the analyst didn't say it either (if you had taken the time to read his report you may have saved us the trouble)... yet you seem to think and suggest that people are.

"He said that the company is certain to "outperform" the naysayers and appears to be on the cusp of getting back into growth. In July, the company posted better-than-expected quarterly results, indicating that the demand for chips is improving, he claimed."

The naysayers he's referring to is people/other analysts expecting Intel's overall revenue&profit to be flat or below expectations and is not tied explicitly or implicitly to Atom growth only. He is talking about people who have the stock at market perform or worse...Again you read the next 2 sentences of what the INQ decided to cut and paste and assumed everything is tied together - read the whole analyst opinion and then get back to me on that.

"Intel seems to be selling the bundle at a lost"[sic] and
"Even if INTEL is marking the atom up at 200% they are still only making 3.75 a bundle"

Yeah... you seem to know what you are talking about... you don't even know the bundle price :) (you just take some article quoting the nutty Nvidia CEO as some sort of "fact")

No, I have no read the full analyst report... and I don't have to after reading if all his argument is:

1.) "In July, the company posted better-than-expected quarterly results, indicating that the demand for chips is improving, he claimed."

2.) Pitzer said that Intel was going to make huge profits on its Atom chips, which it makes for next to nothing and still charges a good mark up on.

#1:
What two-bit (one-bit? half-bit?) analyst can justify demand is improving based on that?

If you read it, how about quoting something he said that actually sustaining this in a real sense?

There is nothing "the naysayers" don't already know? Again Intel claimed demand is improving and anyone calling himself an analyst can read a quarterly earnings report?

To which you allegedly also doubt.

So all that is left is to argue how Atom won't make a difference which you have blown way out of proportions. Yeah, if is so obvious, how come we are arguing about for so freaking long?

#2:
Main point is Atom won't matter and it SHOULD be obvious but that what "the naysayers" would claim. So I explored how likely is it for Intel to make enough profit from Atom to offset the effect of "flat or slightly lower than expected reveune/profit" from its core business?

"In a worst case scenario, Atom could yield a profit margin of 150% (Z500, Z510), and in a best case scenario 1588%."

This is a gross margin percentage of 60% to 94%... The vast majority of netbooks are lower end (i.e. more heavily weight towards 60%) so even a 50-50 average of 77% is generous let alone 100% like I used.

Still, Intel will have a hard time even reaching $500M/quarter in revenue from Atom. Let alone $650M~$1B or MORE in revenue from Atom based on previous quarterly results and projected growth of netbook market. So no, it is unlikely Atom would matter.

You been agreeing all along that Atom does not matter even though you have also been attacking my numbers at the same time. So are you actually going to make a case to the contrary on this?

So you basically claiming my numbers are off and "ridiculous" but we are arguing about nothing else *yawn*? :-)