The Little Chip Fab that Could

By Jon Miller | Post Date: October 4, 2006 12:20 PM | Comments: 5

To a person who is not "up" with the latest high tech terminology, a phrase like How AMD Bakes its 65 Nano Barcelona Cakes seems almost like nonsense. Giving near-nonsense a chance can pay off at times, and this article in technology journal The Inquirer contained a neat case study in two very different approaches to operational excellence taken by Intel and AMD.

Intel is the giant of the semiconductor world and AMD is the challenger chasing it. Being the incumbent with resources and market share, Intel dominates through momentum while AMD is more nimble. Says the article:

Intel has a strategy of copy exact. Basically, whatever the technology is, it is rolled out to the relevant fabs exactly the same way. AMD has a more "on the fly" approach and experiments in real time on the production lines.

Intel has many fabs, while AMD has one. Fabs cost billions to build, so neither firm is fab-building their way to operational excellence.

AMD has no luxuries, it has to produce stuff with what it has.

That's a great reason to do kaizen. In an echo of the definition of Just in Time and doing more with less, the manufacturing system called APM allowed AMD to [...]make only the chips needed and continuously tweak them. When your back is to the wall, and any mistake means death, you can do amazing things.

Describing the wafer manufacturing process like baking a cake, the article summarizes the difference between AMD and Intel as one in which AMD has the ability to make changes to settings during the baking process and perform experiments while Intel can not. The quicker, more flexible system that responds to feedback improves faster than a system based on scale. It's a classic case of the fast taking a bite out of the slow.

The article calls this an approach a "continuous small scale improvement program". I can spell it in five letters: kaizen.

The AMD operational excellence strategy lets them make what they need, when they need it in the right quantity (as much as their forecasting is "right"). The article says:

AMD claims it does not start a wafer without an order in hand, if you sell something before you make it, you almost never end up with depreciating parts in the warehouse.

Pretty impressive if their definition of an "order" is an actual paying OEM or end customer rather than something that exists in the production planning cyberworld.

This strategy of speed and flexibility for achieving operational excellence has helped AMD increase productivity by 31% and start 1,000 more wafers per week while lowering time through the process. Stock has fallen 40% also according to the article.

Intel didn't get to be as successful as they are by doing production poorly. But that it doesn't matter how good you are. Time flows and how fast you can improve is the key to survival. Intel's operational excellence strategy of "copy exactly" would be excellent if what you was copied had been through kaizen recently.

Jon,

Another great article and reference - your insight has proven very valuable to me on many occasions as I work with my clients on their improvement journeys.

I am therefore almost ashamed to point out that "kaizen" is spelled with six letters and not five.

Thanks again, and please keep the excellent posts coming,

Jason

Poster: jason | Post Date: October 5, 2006 5:38 AM

I think you're misrepresenting the Intel approach. "Copy Exact" is an extreme version of Standardized Work. They have a process for driving improvement -- then that improvement is driven out to ALL fabs. I'm not sure what's wrong with that. To imply that Intel doesn't do experimentation and improvement seems quite inaccurate.

Poster: Mark Graban | Post Date: October 5, 2006 5:51 AM

Mark,

I have not misrepresented the Intel approach. I have merely presented a summary of The Inquirer article's interpretation, and pointed out that AMD's approach seems to be more kaizen-oriented.

As far as "Copy Exactly" being an extreme version of Standardized Work, it is not. Standardized Work a.k.a Standard Work as defined by Toyota is the "most effective current combination of manpower, material and machinery" built on takt time, work sequence and Standard WIP. It has nothing to do with "standardization" based on copying as Intel does.

I'm sure Intel has a continuous improvement approaches that are less famous but very effective. I've talked to people there who were doing something that sounded very much like Production Preparation Process (3P).

Jon

Poster: Jon Miller | Post Date: October 5, 2006 7:23 AM

So two Toyota assembly line employees doing things the same way is "copying"? If you have a proven "most effective combination of manpower, material, and machinery" at one Fab (which Intel does), why would you not do things the exact same way with an identical Fab? I don't see why you view Copy Exact as something that has nothing to do with standardization.

Poster: Mark Graban | Post Date: October 5, 2006 9:13 AM

Yes it is copying and copying is a good thing. It's better if you copy and at least attempt to kaizen it, which I understand Toyota does and Intel does not (they copy exactly).

Let me clarify an ambiguity. When I said above 'It has nothing to do with "standardization" based on copying as Intel does.' the first word "It" stood for "Standard Work" in the preceding sentence. Standard Work does not equal standardization.

"Copy exactly" DOES mean standardization, not Standard Work. I have no evidence that Intel does Standard Work based on takt time, work sequence and Standard WIP as Toyota defines it.

Poster: Jon Miller | Post Date: October 5, 2006 10:16 AM

Comments are moderated to filter spam and inappropriate content. There may be a delay before your comment is published.