Having vented in the post below, I thought I'd quickly sum up what I've gleaned so far about Intel's re-structuring.
Projected savings - 2 billion in 2007 & 3 billion in 2008
Impact:
1. Workforce reduction of 10,500 - 5,000 already done through sale of businesses, attrition and manager layoff. Another 2,500 principally from marketing and IT this year. Another 2,500 by mid-2007 at which point will extend to manufacturing due to efficiencies gained through changes to process, etc. All these will be a combination of layoffs and attrition.
2. Reduction of merchandising and other expenses. Intel has already reduced the amount of marketing funds they allow customers to accrue under the Intel Inside program from 6% down to 5% earlier this year. Even on a base of 1.5 billion $ for that program, each basis point gives you about 250 million $ back so I would definitely expect them to save a significant chunk in their marketing expenses by scaling back this program.
3. The incremental 1 billion $s in 2008 will come from manufacturing. I guess they've found some serious inefficiencies that will allow better use of infrastructure, materials and manpower over time.
On the severance cost, I was pretty surprised at the number being only 200 million. But here's the rub. First, 5000 people of that 10,500 already gone. Also, they continue to attrit so naturally those folks will not receive severance. Let's assume that's a conservative 1% per qtr which reduces the number even further if you count the 3 quarters from now till mid 2007. Next, it's not clear from their press release whether that 200 million is cost between now and mid 2007 when layoffs finish or strictly within 2006 or 2007 financial year so I'm assuming it's overall. All in all it looks like the severance may be more in the range of 50-100k $ per person vs the 20k it looks like on the surface. Here's an interesting story that Intel employees in Ireland were receiving 8 weeks salary for every year served. I wonder if that's the legislation? Whatever...that's a pretty good handshake.
At the end of it, I still think this was a pretty weak finish to all that big talk about re-focusing the company. I won't repeat myself - just read my other post.
Subscribe to:
Post Comments (Atom)
11 comments:
I do apologize for the off subject post, but I can't help but mention it in a forum where I don't think it will be immediately erased. Has anyone noticed besides me a very distinct change in Sharikou's writing style? It's as if more than one person has been writing the blog, and in particular either a less educated or younger individual is currently at the helm. It's a bit odd, but look at the writing change over the course of time since the blog began.
No problem - does it feel to you like when his blog started, it was a little tentative, become more agressive and showed deeper industry insight and has now seemed to become softer in tone and content again?
Frankly - I was expecting to hear a more vociferous response to some of my posts too. But I'm not complaining.
Okay, here's the story on Sharikou, after observing his behavior for several months now I'v taken it on as a pet project to analyse him/her (I assume he's a male).
Sharikou is, from a professional standpoint, exhibiting symptoms of obessive/compulsive behavior.
It would very much surprise me if he was NOT obsessive/compulsive. He spends an inordinate amount of time thinking about and talking about issues which, unless he was the CEO of AMD, practically speaking, shouldn't have the importance in his life that he displays it to have.
Even a CEO with the weight of an entire corporation on his/her shoulders will be able to separate themselves from their work proportionally, or at the very least, tire enough of certain thoughts and be able to detach themself from their 'crusades'.
Sharikou does not talk much about anything other than his dislike for Intel.
I would venture to say that he is not employed (another telltale sign of ocd is some people allow their disorder to consume them to the point they are unable to work or live in a normal fashion), as he posts messages relentlessly on most days of the week.
I once followed him through his haunts and simply counted his posts, which were over 100 posts of varying length over the course of a day - and my search was confined only to a few blogs, there may have been far more that I didn't even know about.
You would be noticing (as I have), changes in his moods from time to time. More wide and sweeping mood changes rather than sharp and abrupt changes. OCD is quite often associated with other anxiety disorders such as depression, which judging from the lengths in his tonal variances, is what he may also be suffering from.
We see that for several weeks he will be more passive, then gradually shift over to an agressive persona in a few more weeks. Sometimes his tone is very disturbing, but as far as it goes, not TOO disturbing (enough to worry about).
I believe it's the same person authoring his blog, as you don't see sudden changes but rather amplified, predictable changes.
I would expect him to start becoming aggressive and very short tempered again very soon...he seems to be due...expect a lot of comments like 'idiots' and 'retards' for the next two weeks or so when that happens.
I would also expect 180 sharikou to be a favorite target of his when this happens. Enjoy!
Good detailed assessment. It will be interesting to observe. Perhaps this dialogue will trigger something. Absolutely no need to apolgize for being off post. After all, I did label the blog Sharikou 180° for a reason. As long as folks are not making personal attacks it's cool to have an intellectual exchange. Now just to prove I'm not Sharikou I must head off to get some work done.
cheers -:)
I actually snorted when you posted that....
...the thought HAD crossed my mind. Seems like a good way to generate traffic!
This is pure speculation on my part but I think this is the first step of the process with additional divestitures (spelling) and changes coming.
1) It is only a matter of time before flash (and it's ~5000 employees) is spun off - this has been a mainly non-profitable division for many years and I think the board/stockholders will be calling for Otellini's head if this isn't done. There's already speculation about ST being a potential buyer.
2) I think some additional areas will be scaled back/eliminated - there is a farily large effort in digital health and itanium contnnues to be a resource draw (will this be put in a sustain only, no future development mode?)
3) There will be changes in the fab world (although I doubt the speculated pushout of either AZ oor Ireland plant will happen)
I agree announcements were weak (especially after 5-6 months of "evaluation") - I suspect Otellinin is being conservative and more announcements will trickle out over time.
1. Agree.
2. Agree - there have to be other things that were not announced. Perhaps those will be done quietly under the radar. I've seen articles that people at Ireland and Malaysia have been given voluntary seperation. But the Intel press release mentioned nothing about these so I'm suspicious.
3. Agree - seen these articles?
Malaysia cuts
India news
I think/hope you're right.
Here are a few thoughts you may be interested in thinking about and or correcting:
I read the link to the Ireland paper in regards to compensation of any laid off personnel. What I would like to address is how they phrased things to lead you to the point you made. For example:
"It is understood that the departing staff received eight weeks’ pay for each year of service, as well as other bonuses and pay for shift work.
The phrase "it is understood" doesn't conjure up proof or fact. Essentially it is offering an opinion or hearsay. Based on information I have, Intel's separation package is more along the lines of 8 weeks of base salary with 2 weeks of additional pay per year of service. For example: An employee with 5 years of service would receive 18 weeks (~4.5 months) of base salary; 8 weeks + (2 weeks x 5 years). This fails in line with other severance compensation packages offered by a variety of corporations and what I read was offered to the 1,ooo managers that were cut in July. Obviously the variable is time with the company, but an average of $20k paid in compensation per employee lost would be just about right and fail in line with the $200 million announced; $20k x 1o,ooo+ reduction.
The incremental billion dollars you mentioned being saved in 2008 is something of a mystery. From what I read in the press release, it sounds to me like Intel is more apt to get a little bit more bang for their buck by using equipment already installed in their factories and not rushing out to re-tool factories. Successful semiconductors have always been the ones who spent tall sums on new equipment from generation to generation. I would expect Intel to be one that spends more than others. Judging by the number of fabs they have and the cost of equipment used in manufacturing semiconductors, I wouldn't think it would be hard at all for Intel to save a billion or more by trying to reuse as much existing equipment that has already been expensed via depreciation in other technologies. Granted, this wouldn't be possible for entire factories as certain processes require cutting edge technology for improvement. That being said, what could be re-used or repurposed is just extra money in the bank for Intel.
All in all... I enjoy your perspective and spin. Keep up the good work, Gumshoe.
Sheepshagger - my thoughts (and I'm not pretending to have all the answers):
1. I was thinking like you were too until I read a comment that this included attrition too. Attrition should be around 1% per quarter (if not more which takes ~2.5 - 3k people out w/o anything more than notice from now till mid-07. Secondly, Intel has already taken 5000 people out of the system through the sales to Eicon and Marvell and earlier lay offs. So even if we disregard the attrition, we're talking 200 million over 5k people. If I add in the attrition we can look at a 50K per head number. Now, you're right and not everyone will get the same severance. But in some markets there are laws on severance. Intel also has an employee base with long tenure in the company. I've heard of VSPs in the past with 3 months base and 1 month for every year. All in all, my calculation is not fact I can prove but I think 20K is not adding up for me and even 40k is a more likely number if you think 50-60k is too high.
However, this is relatively a non issue. Your comment below on the 1 billion $ saving is of more interest.
2. On the extra billion - I think you're broadly right. I also went back and read the press release. Have a look at this:
In addition, Intel expects to achieve a capital expenditure avoidance of $1 billion by better utilizing manufacturing equipment and space.
SPACE - why mention space. How does space help in semiconducting manufacturing. Are they talking smaller fabs - or bigger? Is there a connection between tool utilisation and size of fab? Can you get better utilisation if the fab is bigger? Perhaps you can add more product lines into the same fab by adding some incremental machines but sharing some that you couldn't if the fab was dedicated to limited product lines. Would that mean a bigger fab is better? I'm making this up. If anyone out there knows the answer do jump in. Intel has a lot of fabs so assuming they could replicate the saving from one fab to all using "copy exactly", that means they would need to find a 100 million $ saving and just replicate it 10 times.
Or maybe bottomline they discovered they were at the bleeding (not leading) edge of manufacturing and decided to copy their competitors. Though I doubt they could reap benefits so quickly.
-:) At this point I expect the real Sharikou to jump in and say "I told you so - APM is superior and Intel is going to copy it".
Thanks for the encouragement...and thanks for keeping the discussion intelligent.
Re: the cost savings of reducing headcount, the standard figure bandied about is a burdened cost per employee (salary, benefits, overhead such as office space) is $150-200k. If you want to divide by 3 for offshore work, and assume 30% of layoffs are offshore (SWAG):
7000*175k=1.225B
3000*58.3k=0.175B
Total annual employee cost savings= 1.4B/year
SPACE - why mention space. How does space help in semiconducting manufacturing.
Space is the most important thing. All manufacturing takes place in the FAB's highest grade clean room. The numer of tools you can fit into this space determines how many wafers you can process.
I'm making this up. If anyone out there knows the answer do jump in.
Actually, Intel is just talking about not building additional clean room space. Capitol avoidance just means that their steady pattern of building new FABs is over. They will use what they currently have plus the two 45nm FABs they are building.
Or maybe bottomline they discovered they were at the bleeding (not leading) edge of manufacturing and decided to copy their competitors. Though I doubt they could reap benefits so quickly.
This is way off. Intel's manufacturing process is less sophisticated than AMD's. All that has happened is that Intel has revised its estimates for future capacity requirements because of its losses to AMD. They've decided that they can cover the expected volume of chips without building more clean room space.
Specifically, this is the opposite of AMD. In spite of AMD's new FAB 36 they are building an additional cleanroom space so that they can move bump and test equipment out of FAB 36 to allow additional tooling to boost FAB 36 from 20K wspm to 25K. If we add in FAB 38's 20K wspm and this does represent 30% by volume in 2008 then Intel only needs the three 300mm FABs it has now plus the two it is building. Intel doesn't need 6 FABs because its FABs have more cleanroom space than AMD's. One question is whether they still intend to upgrade D1C.
-:) At this point I expect the real Sharikou to jump in and say "I told you so - APM is superior and Intel is going to copy it".
APM is superior to what Intel uses. Intel has a production and scheduling application that tries to balance the production at all FABs. However, this software requires the engineers to manually intervene to adjust any production problems. APM will adjust the process by itself.
Another difference is that Intel's "Copy Exact" method takes 2 months to duplicate on anther FAB, in other words, a single test run before production. With APM, AMD can go to Chartered (and FAB 38 once it is completed) in the same day. This ability doesn't help much with FAB 30 since it is still at 90nm.
Post a Comment