Saturday, September 02, 2006

20 Conroes to feed a worker???

Sharikou claims Intel needs to sell 20 Conroes to feed an employee for a month which I thought was just too plucked out of the air to let pass. Let's use his previously stated assumptions to analyse this.

He's told us in the past Intel's cost of manufacture of 1 cpu + chipset is 100$. I'm gonna make the assumption that the Conroe CPU cost is 80$ - for a couple or reasons. First - Intel chipsets sell for around 40 bucks. I assume Intel doesn't give them away so mfg cost out of a hundred bucks should be 20-30 bucks. I'm also going to assume a higher CPU cost for conroe right now as yields are probably not yet stable. I then assume out of 20 CPUs the mix is C2E - 5%, 6700 - 15%, 6600 - 20%, 6400 - 25%, 6300 - 35%. Frankly, I think assuming the 224$ upward price point will constitute 65% of the mix is generous. My gut is more 80% will be the 6300 & 6400 parts but let's go along The math gives us a profit of ~4600 $.

From this, we need to remove the operating expenses per employee. From Intel's 2005 balance sheet this is ~11 billion $. Amortising this over ~100,000 employees gives us a monthly operating cost of ~9,150 $ per month. So Intel would need to sell at least 40 Conroes a month to feed a worker.

Now I understand this calculation isn't scientific or would hold up to any accounting standards. However, it's a lot more scientific than yanking a number out of my back pocket.

8 comments:

180 Sharikou said...

Azary - I'm just guessing (as is anyone else). What do you think would be a better number?

"Mad Mod" Mike said...

So basically this is going to be a place where you counter every post Sharikou makes? Hmm...

Anonymous said...

If Intel neesd to sell 40 Conroes per month, instead of 20 Conroes to employ each worker, this is TWICE as bad.

You're actually giving Sharikou more credit for giving Intel the benefit of the doubt.

180 Sharikou said...

Actually not. Here is Sharikou's compelete statement:

Intel only needs to sell 20 Conroes to feed one worker for a month. For 20,000 workers, Intel only needs to sell 400,000 Conroes. But the reality is, Intel can't sell so many Conroes for whatever reason.

First, we all know this is a ridiculous over-simplification of why Intel is laying off so many people. However, what this means is Intel needs at least 2X the Conroes Sharikou says they do per month. Now...let's also combine this with Intel's statement that they will ship 1 million Conroes in Aug-Sep and we know that Sharikou's comment doesn't have further discussion value.

Joshua said...

lol this is funny, some biased like Sharikou, but some of it is actually true!

Joshua said...

Oh yeah, I just updated my Blog! www.tech-blog.blogspot.com

Sharikou, Ph. D. said...

I'm also going to assume a higher CPU cost for conroe right now as yields are probably not yet stable. I then assume out of 20 CPUs the mix is C2E - 5%, 6700 - 15%, 6600 - 20%, 6400 - 25%, 6300 - 35%. Frankly, I think assuming the 224$ upward price point will constitute 65% of the mix is generous. My gut is more 80% will be the 6300 & 6400 parts but let's go along The math gives us a profit of ~4600 $.

Even the 6300 is selling at $200+. The average Conroe price should be around $380. 20 Conroes lead to $6000 profit. When you consider the $9150 cost, you need to note some of the cost doesn't depend on head count, such as rent and utility. So the variable cost is lower than that. $6000 profit is definitely more than enough to feed an average Intel worker -- a lot of Intel's workforce are in lower paying countries.

The problem is, Intel can't sell that many Conroes for whatever reason.

180 Sharikou said...

Even the 6300 is selling at $200+. The average Conroe price should be around $380. 20 Conroes lead to $6000 profit. When you consider the $9150 cost, you need to note some of the cost doesn't depend on head count, such as rent and utility. So the variable cost is lower than that. $6000 profit is definitely more than enough to feed an average Intel worker -- a lot of Intel's workforce are in lower paying countries.

The problem is, Intel can't sell that many Conroes for whatever reason.


Let me address these one by one:

1. I have used Intel's list price. If Newegg is selling CPUs above list price that's profit for them...not Intel. Please substantiate how you arrived at 380$ average price as I've shown how I arrived at my number.

Seperately - the 6300 is at 190-195 $ on Newegg, TG Direct or Zip Zoom Fly.

2. The 6000$ profit is like above a number yanked out of thin air based on a 380$ average price.

3. If you notice, I said remove operating cost per employee. Neither you nor I know how those costs break up. However, if we're going to calculate revenue per head to approximate profit per head then it's fair to remove operating cost per head. Remember - we're saying what does it take to feed those 20k workers about to be laid off. To make the sales to earn the revenue you need to incur the operating cost per head.

In addition, you pointed out to Dr Yield rather vociferously when he was trying to point out deficiencies in your calculation of average price that you were only approximating to arrive at mfg cost per cpu + chipset. That you did not have the break up by variable and this was sufficient to make your point.

Hence, I'm going to disregard your comment on removing items like rent and utility.

3. Nobody knows where those workers who will be laid off are - in high cost markets like the US or low cost markets like India. Hence, removing operating cost per head is a fair way to average the cost.

4. Finally - Intel has clearly admitted they will do 1 million units in the first 2 months. That's 500k per month. Remember, they have ~100k employees so even if your number was accurate they actually need to seLl 2 million units a month to feed their employees.

All this is moot. My point is please put some dilligence and some base level rationalisation when you make a claim like this. You did it to calculate average price of mfr so please keep doing that and don't pull numbers out of thin air. It's clear to me you have plenty of insight and knowledge of the semiconductor business.

BTW - thanks for coming to my humble blog.