Friday, September 01, 2006

Intel needs to layoff 80% of its workforce to avoid BK

Back in 2005, I told Intel execs that they will have to layoff large number of employees once AMD FAB36 ramps up. Back then, I suggested to Intel that it should cut production and raise ASP to maintain constant revenue, as losing unit share is inevitable. Intel took the exact opposite approach. As a result, I projected operating losses for Intel from 3Q06 onward, leading to eventual BK around 4Q07 to 2Q08 timeframe.

It is reported that Intel will layoff as much as 20,000 people. Imagine 20,000 Intel engineers and marketeers become jobless and on welfare overnight. Furthermore, you can expect most of these people have spent their youthful years at Intel and will face a tough situation in the job market. They will have nowhere to go, but join FACE Intel. The result is tragic. and the impact on local economies will be huge.

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.

The fact Intel has to layoff so many people leads us to conclude that Conroe is not saving Intel, on the contrary, Conroe is killing 95% of Intel's sales. I have predicted this long time ago. We concluded previously that Conroe was the final straw that pushed DELL to AMD.

Laying off 20,000 workers will delay Intel's BK time by 1 quarter. If the layoff is confirmed, we should adjust Intel's BK time to 1Q08 to 3Q08.

To avoid the BK scenario, Intel needs to layoff 80,000 workers. AMD has only 10K people, Intel should have no more than 20K.

25 Comments:

Anonymous Anonymous said...

Intel might be in better position when they ramp up Conroe productions and for god's sake stop all Netburst productions. Conroe are now the fastest x86 chips, I can't deny that, even if it is comparing to a 3 years old chip. Therefore they should concentrate fully on that, and leave AMD to the low-mid range.

However until Intel ramp up Conroe productions, my guess is that AMD would have something competitive enough.

11:15 AM, September 01, 2006  
Anonymous enumae said...

Sharikou said...

"To avoid the BK scenario, Intel needs to layoff 80,000 workers. AMD has only 10K people, Intel should have no more than 20K."

20,000 is all Intel should have...

Intel...

90nm = 2 fab facilities
65nm = 3 fab facilities 1 currently in ramping
45nm = 3 fab facilities currently in ramping

AMD...

90nm = 1 fab
65nm = 1 fab

Where do you get your numbers?

I do not have a Phd, but I understand that Intel will nedd well more than 20,000 employes, and we haven't even touched chipsets.

11:31 AM, September 01, 2006  
Anonymous netrama said...

I think conroe is perceived to be on the high end like the FX62.. so ramping it might help ,only if they are planning to dump it on low asp in line with the Netburst crap. This ofcourse is shooting in the foot twice. Either way Intel is screwed. The mindshare and interest AMD has received in the last few years has taken a ir-reversible toll on Intel and its bully like business practices. For a lot of folks deciding to buy a computer , probably their only (corporate) choice was DELL..not because Dell had Intel ..but because Dell was their preferred vendor !! Dell realised this ..but a little late ..But better late then never !!

11:57 AM, September 01, 2006  
Anonymous Anonymous said...

Enumae?

Where do you get your numbers from. AMD has 2 fabs doing both 90 and 65nm. Don't you know FAB 30 went into 65nm production and compleated conversion in may 06?

Fab 36 is doing 65nm as well. Also AMD is being helped from a few other fabs that do 65nm in other buisnesses. So you don't know what your talking about.

12:16 PM, September 01, 2006  
Anonymous Anonymous said...

Intel has been planning these actions well before Conroe came out. They discussed this in their Q2 update.

They have too many resources and need to reduce their workforce. AMD has taken much more dramatic actions in the past (as a accordance to their size.) In fact I bet when AMD does have to lay off because they are currently unable to keep up with AMD that you will praise them - I have read here for just a little while but I believe the name should be changed to "Journal of Hypocritical 64 bit computing" This is just a place where someone who has a grudge against Intel is working through their issues - Counseling is another good option.

1:07 PM, September 01, 2006  
Blogger Kalle said...

"Intel only needs to sell 20 Conroes to feed one worker for a month"
How many old Netbursts does it take to feed a man? How many of those does Intel sell? How many flash chips and embedded CPU's it sells?

"Laying off 20,000 workers will delay Intel's BK time by 1 quarter. If the layoff is confirmed, we should adjust Intel's BK time to 1Q08 to 3Q08."

Is it just me or your math is a bit weurd? 1+1=2, not 3.

1:40 PM, September 01, 2006  
Blogger Pop Catalin Sever said...

Some days ago I was asked to make a custom PC configuration for a friend.

"I want the latest technolgy!" I was told. "It's going to be a Pentium 4, right ?"

"Well Pentium 4 isn't the latest technolgy from Intel, actualy is about three generations old and also Intel is droping the Pentium brand." At. that moment my friend was staring at. me not knowing what to believe or if I was making fun of him.

This is kind of got me thinking. There are actualy alot of people in the world for wich Intel, AMD, Core or Atlhon does't mean anything.

It was a mistake for Intel to drop the Pentium brand because the ne brand "Core" will take lots of time and efort to reach the level of acnolegement similar to the former one for the masses.

If Intel had named Core "Pentium V" it would have been a great victory from start, because that's what non technical people are expecting, the next thing after Pentium 4: Pentium 5 :).

3:28 PM, September 01, 2006  
Blogger Sharikou, Ph. D said...

Intel...

90nm = 2 fab facilities
65nm = 3 fab facilities 1 currently in ramping
45nm = 3 fab facilities currently in ramping


Intel will soon have to shutdown some FABs.

4:12 PM, September 01, 2006  
Anonymous enumae said...

Anonymous said...

"Where do you get your numbers from."

My a$$...

"AMD has 2 fabs doing both 90 and 65nm. Don't you know FAB 30 went into 65nm production and compleated conversion in may 06?"

While I will admit I was under the impression that Fab 36 was already at 65nm, turns out I was wrong, my bad.

But don't you mean 1 Fab doing 90nm, and one Fab doing both 90 and 65nm... read on!

Here is what AMD is saying...

quote...AMD said it is planning to transform the Fab 30, which currently processes 200mm wafers and utilizes 90nm process technology, into Fab 38, which will use 300mm wafers and 65nm process technology. AMD will ramp down 200mm manufacturing in the second half of 2007, with preparation already underway for the ramp of 300mm wafers on 65nm process technology at Fab 38 by the end of 2007... /quote

"Fab 36 is doing 65nm as well."

quote... Fab 36 will be "substantially converted" to 65nm manufacturing by the middle of 2007... /quote

"Also AMD is being helped from a few other fabs that do 65nm in other buisnesses."

Well those would not be AMD employese would they, and that was the basis of my previous post.

"So you don't know what your talking about."

I have admitted making a mistake, will you?

5:50 PM, September 01, 2006  
Blogger sharikou180 said...

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.

This number is fantasy. To get the real number, check it out here.

7:55 PM, September 01, 2006  
Anonymous Anonymous said...

Conroe will mean simply the beginning of intels fall. Why so many workers lost, why so many jobs lost, why so many stock lost, why so little conroe, why not get rid of pentium storage, all simply mean signs of... the end!

The fall of a once great company destroyed by its final last ditch effort to save themselfs eventualy FRAGED them to space dust. I'll be more then happy to forget them. Why? Because too much of conroe is a joke made to bench, not for real world performance but just to show off.

Intel never again! Sounds like a unhappy customer right? Cus I'm very very unhappy. Intels are jokers, stop the lies... just stop the lies!!!

8:32 PM, September 01, 2006  
Anonymous Anonymous said...

"Conroe is killing 95% of Intel's sales"

Sharikou how many times can you still make this error?

How does Conroe (DESKTOP CHIP) hurt mobile and server sales? Or do you think desktop is 95% of Intel sales? Does it hurt sales of NOR flash products? Does it hurt mobile chip set sales?

Do you mean Conroe is hurting 95% of Intel's destop chip sales?

7:34 AM, September 02, 2006  
Blogger Sharikou, Ph. D said...

How does Conroe (DESKTOP CHIP) hurt mobile and server sales?

Conroe, Merom, Woodcrest are the same chip.

7:51 AM, September 02, 2006  
Anonymous Anonymous said...

"Conroe, Merom, Woodcrest are the same chip."

Opteron, Athlon 64 X2 and Turion X2 are the same chip as well, but I've never heard anyone claim that as a disadvantage for AMD.

11:41 AM, September 02, 2006  
Anonymous Edward said...

"
"Conroe, Merom, Woodcrest are the same chip."

Opteron, Athlon 64 X2 and Turion X2 are the same chip as well, but I've never heard anyone claim that as a disadvantage for AMD.

"

Conroe kills 95% Intel desktop sales; Merom kills 95% Intel notebook sales (probably a bit less); Woodcrest kills 95% Intel server sales (probably a bit more). Sharikou was just being brief by saying in general that Conroe and its siblings kill 95% Intel sales.

What AMD sales does Turion64/Athlon64/Opteron kill? Athlon-XP? Athlon-MP? Or Sempron maybe?

4:39 PM, September 02, 2006  
Anonymous Anonymous said...

" Does it hurt sales of NOR flash products? Does it hurt mobile chip set sales?"

Flash is a net loss for Intel as it was for AMD. Thats why they (AMD) spun it off into Spansion. Intel no longer makes mobile chipsets. They sold that to Marvel. (spelling?)

Netburst chips across all product lines, IE Xeon, desktop and mobile are indeed killing sales and ASP for Intel. There is a glut of these chips Intel forced into the channel to keep up numbers in previous quarters. This has been reported all over the internet, not just here. i can't attest to the exact percentage but it is a reality. This was a classic firearm/foot manuver.

5:13 PM, September 02, 2006  
Anonymous Anonymous said...

The stupdity here is astounding. Like the Phd Prentender.

Lets do some math. If indeed AMD has 10K employees and has 25% marketshare then lets do simply math.

INTEL owns 75% market share that is worth 30K employees. Add to the fact they have a chipset business > ATI+NVIDIA togather. That is worth another 2K. Add the fact they have 3-4 8" 130nm factories doing all the manufacturing and that is another 10K. Add the Itanium and all them great blue man comericals and that new fancy swirl log and you come to another 10K.

I'd say a lean INTEL that can do what they do know but with no fat is probably around 55K. Thus if Paul Otellwennie had any balls the right number.

And this guy is as smarth as the host blogger "ntel no longer makes mobile chipsets. They sold that to Marvel"

A real PhD

10:15 PM, September 02, 2006  
Anonymous Anonymous said...

I stumbled onto this confabulation via Google, and it made my day!

Do I necessarily love Intel?

No.

But this author has (pick one)

1) Sold Intel shares short
2) Purchased lots of AMD shares
3) Smoked too much loco weed, losing all math capabilities.
4) PhD obtained online for $29.99.
5 Unlikely to post any posts critical of his methods.
6) All of the above.

Personally?
I'm going with #6

8:34 AM, September 03, 2006  
Anonymous george said...

intel is torn by core 2 duo because it has big invetorys of Pentium 4 and pentium D processors so intel must ether throw away networst processors or paper launch core 2 and lose to amd for most users this means that core 2 backfired because they did'nt stop production of pentium D processors.

2:56 PM, September 03, 2006  
Blogger Scientia from AMDZone said...

I still don't see how you arrive at the idea of bankruptcy in Q2 08. By that time AMD will have perhaps 28% of the X86 processor market by volume. 4% is taken by VIA. That still leaves 68% of the market to Intel even if they do nothing. However, Intel's production should be almost all C2D by then so they should be reasonably competitive. That would be an 8% decline by volume but actually a bit less because of growth in the market. Even if this caused a loss of revenue of 20% this does not sound like enough to cause bankruptcy.

Some of the statements about AMD's production are a bit off.

AMD has one 200mm FAB, FAB 30 which is producing 90nm chips. It produces 30K wafers per month. It will not produce 65nm on 200mm wafers.

AMD has one 300mm FAB, FAB 36 which is just beginning to produce 65nm chips. This FAB will be at 30% 65nm by the end of 2006 and will provide all of the 65nm chips for the 4th quarter that are made by AMD. FAB 36 will be at 10K wpm at the end of 2006.

Chartered will also provide some 65nm chips however my current estimate would be no more than 4K wpm during Q4. Chartered is also a 300mm FAB but this would be only about 1/4 of what FAB 36 will produce.

In mid 2007 FAB 36 will be at 75% (15K wpm) capacity which would be equal to the output of FAB 30 if FAB 36 were producing all 90nm chips. Since it will be 70% 65nm its capacity will be larger than FAB 30 and FAB 30 will begin scaling back for conversion to 300mm. When conversion is complete in early 2008 it will be renamed FAB 38.

At the beginning of 2008, FAB 36 will begin converting to 45nm. FAB 36 will be at 100% capacity however it will begin using capacity in the expansion to bring total capacity to 125% in mid 2008 when the first 45nm chips will be delivered. FAB 38 may lag 45nm conversion by a quarter but should also be producing 45nm by the last quarter of 2008. FAB 38 may be at as much as 75% capacity. This would be a total of about 40K 300mm wpm by end of 2008 and increase to 45K in 2009.

3:12 PM, September 03, 2006  
Blogger Sharikou, Ph. D said...

INTEL owns 75% market share that is worth 30K employees.

Intel had 73% of the market in 2Q06 and that number is falling. Expect Intel to have less than 60% market share soon. AMD+ATI (CPU+chipset+GPU) is a company of 15K people. Intel should be 150% of that (60/40). 15K * 1.5 ~ 20K.

3:32 PM, September 03, 2006  
Anonymous Edward said...

"AMD+ATI (CPU+chipset+GPU) is a company of 15K people. Intel should be 150% of that (60/40). 15K * 1.5 ~ 20K."

I don't really agree with that. Intel needs some downsizing, but not that drastic. Intel still have a big wireless group (WiMax anyone?) and a chipset group larger than Ati+nVidia+Via. Intel have a software/compiler group that's been doing great jobs over the years. If 2/3 CPU market share means 20k employees (actually about 25k), then Intel would use all those resource mentioned above to (legally?) prevent its market share to fall that much.

Intel's earnings and stock price are other stories, though, because the only milking cow in all its businesses is CPU manufacturing. A 1% CPU revenue drop could mean 1.5-2% total revenue decline and even more for net profit. While Intel's market share and profitability can reduce quickly, one thing remain true: there are just too many Intel stock out there.

11:13 PM, September 03, 2006  
Anonymous Anonymous said...

"Conroe, Merom, Woodcrest are the same chip."

Opteron, Athlon 64 X2 and Turion X2 are the same chip as well, but I've never heard anyone claim that as a disadvantage for AMD.


K8 or if you prefer Opteron is a server processors design, where:
- Northbridge, connects to others processors without the need of a chipset.
- IMC, share memory with others processors.
All of the above made it a Multi processor, multi core design.

In single processor (Athlon/Semprom) all of the above are mean less.
Only the IMC is used but not needed outside of the server, IMC was a requirement for the integrated Northbridge (multi processor).
x64 is another "server" feature, that’s why K8 (Opteron) is an server processor design.

Intel has nothing like that yet because is very difficult to do, not only at the processor level but also at the chipset level, not even an very complex mobo and chipset design beat the K8, besides it would be too expensive to manufacture.

Conroe, Merom, Woodcrest is just an Mobile based processor (yohan), where you get very good performance/watt specially because of that, than .65nm, and finally by going desktop/server you can ramp up clock speed and deliver the superb performance.

Conroe, Merom, Woodcrest is still years behind vs K8 in design, at least at the server level.

Resuming, Conroe, Merom, Woodcrest are exactly the same! All have the same capabilities; the only differential factor is the chipset/socket that is used.

3:10 AM, September 04, 2006  
Blogger Scientia from AMDZone said...

In Q2 06, Intel had 74% and AMD had 22% of the volume. So, if we redo your calculation we get:

74/22 * 15,000 = 50,500

However, Intel is further ahead in terms of revenue. Intel had 81% versus AMD's 18%. Using these numbers we get:

81/18 * 15,000 = 67,500

I would say that the above figure will be closer to reality after Intel divests itself of its unprofitable companies and trims a bit.

BTW, the 30% by the end of 2008 number comes from AMD. If you are trying to derive this number on your own you need to take into consideration the expansion of the market.

In 2005, AMD produced 45 Million processors. They plan to produce 100 Million in 2008. That is 2.22X the number of chips. However, with a volume increase of only 50% that would correspond to an estimated increase of 48% in the total chip market from 2005 to 2008. This is an annual growth rate of 14% for the processor market by volume. The actual production capacity of AMD will be at about 5X what it was in 2005 (about 6X with Chartered). However, average chip size will be about 2.5X larger. In other words we switch from single core with some dual core in 2005 to dual core with a lot of quad core in 2008. The average die size would be larger but the cache size doesn't double from X2 to quad core.

5:08 PM, September 04, 2006  
Anonymous Anonymous said...

And this guy is as smarth as the host blogger "ntel no longer makes mobile chipsets. They sold that to Marvel"

Great spelling for a PhD. Also, try a Google search for "Intel" and "Marvel". You'll find numerous stories of they sale of Intels mobile chipset business. That would be the one in cell phones and communications equipment, not laptops. Sometimes the more educated people get, the dumber they sound.

9:56 PM, September 04, 2006  

Post a Comment

Links to this post:

Create a Link

<< Home