*AV*--I was in stitches and thought about responding but decided not to. Did you notice that, unfortunately, the membership on SI started on 4/2/98???
Now for something serious. they following is an excerpt from a private message sent to me. the name and some of the comments have been deleted but I wanted to share my response with everyone.
To: Andrew Vance Sunday, Jun 7 1998 9:52AM ET
Hi Andrew,
..........I read 2-3 hours every night to learn more about semi-equips (AMAT, ASYT, PRIA) and biotech (LGND, SUGN, ISIP)....... Could you answer some questions on ASYT which I'm afraid would sound a little too stupid to be posted on SI?
I read on ASYT's homepage that they have pentrated 44 fabs. To put into perspective their market penetration, how many (ballpark) 8" fabs are in existence worldwide?
How many are in Taiwan? How many pieces of frontend process equipment that might need an enclosure are in a DRAM fab or a Taiwan foundry fab? How many SMIF pods might one fab use?? I haven't a clue to any of this or even if its relevant.
....What is the effect of Yen exchange rates on price competion for AMAT's/ASYT's products? Is price really a secondary consideration with this technology or is Tokyo Electron and others going to eat US
semi-equips's lunch?? I found a website with historical currency exchange data which showed we have moved from 80 Yen to the dollar in '95 to 140 now and forecasts are for 150 to (yikes!) 200 Yen to the dollar. At what point does this hurt AMAT with respect to Japanese semi-equip price competition??
......If there is any data in the currency area (historical currency
versus US and Japanese semi-equip sales???) that you think might be of interest on Stockswap I would be glad to try research, follow and post if you have any guidance or advice to offer on what would be relevant or important.
PS FWIW I read on Yahoo that ASYT has cut back orders to a machine shop significantly. Sounded "genuine" but I thought ASYT did their manufacturing in-house???
I could use some help in verifying the number of active 200mm(8") wafer fabs throughout the world and in each region. Any takers???
ASYT does most of its manufacturing in-house. You must separate the ASYT equipment product manufacturing from the contract minienvironment component manufacturing. The minienvironments are nothing more than sophisticated high end jewelry display cases with fan filter units, hinges, panels, doors, and interface hardware or brackets. None of these items are worthwhile for ASYT to have an in house machine shop or lexan cutting shop. This is low end enough to have local shops perform this service. ASYT still makes their own arms, tags, indexers, etal. "robotic and handling and tagging" units which are all covered by patents and are more cost effective to make in house.
The machine shops in question do not run 100% of ASYT's needs either. Sometimes when a project is too far from their home office, they contract shops in the locations where they are going to do the installation. This allows a huge savings on shipping and handling. Such was the case with 2 installations in Colorado. ASYT shipped the
blueprints and drawings to local establishments who then provided the finished pieces for the minienvironment installations.
Unfortunately, the comment you mad was also slightly non specific. Certain cases for the equipment might be bought from external machine shops and could signify a slow down in their business. This may very well be the case but ASYT's business has always been in fits and starts. I have seen lulls as they are winding down projects and trying to close the deal, provide the engineering or preliminary designs, or ramp up a new program. ASYT may have "off the shelf" mechanical units such as ALUs, ARMS, or integrated indexers but the minienvironment and equipment specific requirements makes this part of the process highly customized.
ASYT has penetrated 44 fabs which is not a great number relative to the number of fabs in the world. However, this is really a great situation. I was personally involved in SMIFing the first production fab in the US (not just some feeble pilot line or partial implementation). I also consulted on the first grounds up SMIF facility in both the US and Taiwan. I started in 1986-1987 with my installation and it took a great deal of time due to an inordinate amount of "learning curve". To make a long story short, each installation, retrofit, or new implementation takes a great deal of time and personnel to pull it off. Therefore, ASYT is limited by the number of projects it can really handle.
Luckily, they have not had everyone jump on the bandwagon at once. Luckily, there are still a great deal of non believers that are ripe for conversion. Lucky for ASYT that the talk is that integrated indexers, robotics, and minienvironments are being considered critical (not a luxury) to a successful 300mm implementation.
The slow acceptance of ASYT and the technology has been both a blessing and an issue for ASYT. As we get further into this semiconductor crisis and with 300mm pushed out, the existing 200mm wafer fabs will be looking very closely at improving their yields, production efficiencies, implementing their next generation of technology, while trying to lower overall wafer and manufacturing costs. This becomes very fertile ground for future ASYT implementations. As the number of fabs grow so do the number of satisfied customers and believers in this technology. As the word of mouth successes filter across the industry, the non believers and naysayers may be forced to re-evaluate their opinions. A good portion of the industry engineers would rather buy brand new equipment (toys) for them to play with instead of trying to optimize their existing process, equipment or look at less costly alternatives.
I do not see a full embracement or much regard for Cost of Ownership. Once the upper management starts feeling the pain (raises and bonuses lost) when their business objectives are not met and once someone put their foot down from the Finance side to question engineering decisions and to demand ALL alternatives be presented and financially justified, they will discover the benefits of SMIF and WIT technology.
I do not have the number of existing 8" (200mm) fabs in operation today but it must be somewhere close to 200 individual fabs. This a number should be asked to MEMC Corporation (WFR). I am sure that a nice phone call to them or to SEMI or SIA could yield a very reliable answer. I usually rely on the ICE report that comes out annually to
get a grip on the number but I do not have access to it these days.
The 8" fabs are very highly concentrated in Japan, Korea, and Taiwan. Most of the fabs are probably at this wafer size with the older diameters either retrofitted or idle. As far as how many SMIF Pods might be used, I will give you an idea. It is relative to the number of wafers produced in the fab, though.
If we are talking about a high volume 10,000 wafer start per week facility with a 42 day cycle time, for example, this would mean you have 10K starts * 6 weeks of wafers or 60,000 wafers. Each POD will hold a 25 wafer cassette. Therefore, at a very minimum, you have 2500 PODs in the fab. I would use this as a baseline since we have not added additional PODs for process reworks, spares, transport to other locations within the building, breakage, cleaning, or output indexers to some of the equipment. I also assumed that the lots size remains at 25 wafers while by best estimates the average lot size, after all line yields are factored in, is roughly 22 wafers. We have not deducted the number of PODs that are not needed because all wafers are not in PODs at the same time. Remember, we do have wafers being processed in furnaces and other equipment most of the time. But if you want a good ball park guess, I would use the maximum wafer inventory a fab might experience divided by 22 (average lot size) to get a number. Fabs run from 2000-15,000 wafers starts per week so you need to have a decent handle on how big these fabs are.
I would guess that the average size fab that is not SMIF'd already, has close to 150 process tools that could be retrofitted. Depending on how they are retrofitted (types of ARMs, ALUs or integrated indexers) the cost of the project varies from $3-$20 million. While the number might be slightly off (due to price increases or new products as well as decreases in the cost for minienvironments, engineering, installation, and integration), I like to use the $50K figure as an average price for SMIFing a piece of equipment.
Whoops, question was, how many enclosures for the front end of a DRAM fab. That is really dependent on the talent of the SMIF engineer. This whole thing is still an art form. The really good Engineer, like myself, went way overboard to minimize the number of enclosures used to retrofit my fab. Also, most of the 200mm pieces of equipment today has SMIF integrated indexer options that can be purchased that actually either incorporate an enclosure with their equipment or do not need an enclosure. For instance, the Eaton 6200 implanter needs nothing since its wafer indexer resides in an ultra clean environment. Either you cut a hole in this environment and attach an ARM or your re-engineering the load station to incorporate an integrated indexer. Same holds true for the AMAT8330 metal etcher. All I had to do was enclose the upper ARM an flush mount it to the front door panel of the etcher. The smart engineer could minimize the number of full blown enclosures required. A guess from my own experience in an older fab is that close to 30% of the front end equipment might need an enclosure with most of that being cleaning stations, inspection stations, diagnostic tools, and manual operations.
The currency issue is rather strange. As long as the companies charge and receive in US funds from their customers, there should be no problem. when the company has a foreign office of manufacturing site they are affected by fluctuations in currency. I think ASYT charges in US currency and will fare okay during these financial issues. AMAT has a Japanese manufacturing organization and might have issues with currency changes. AMAT as well as other companies that deal in local currency are forced to raise prices on new orders in concert with the changing currencies to keep margins where they need to be. When in a local market the currency issues also impact their local cost of manufacturing so the offsets in price may not necessarily be 1:1 but rather adjusted to maintain margins. On long lead time items, like most of AMAT's products, companies that deal in local currencies get bit in the butt since they cannot reprice product after the PO is placed. This is where volatile currencies hurt the equipment companies.
Tokyo Electron is trying to eat our lunch as you mentioned. It has always been this way. As they get bigger, stronger and have a competitive product, they will try to dominate the equipment industry as much as they can. This will be no different than how the Japanese have acted in the automotive, semiconductor chip, steel, etal industries. However, we really cannot whine about it. Global competition on an even playing field is healthy. Unfortunately, the Japanese government does not allow an even playing field to exist. US companies have to remain competitive and meet the challenges head on. The real selling "war" is going to be in both Europe and China going into the next decade. (BTW-As much as we whine about the Japanese taking over the DRAM, SRAM, EPROM businesses, the Korean and Taiwanese have been giving the Japanese fits over the past few years and the Taiwan Foundries have been of great support to US based Fabless or partially fabbed semiconductor companies.
The US still is a technological leader in equipment and design. We must not become complacent. None of this was meant as Japan bashing but rather a wake up call to the USA to remain competitive and to lobby the US government to reciprocate with equally unfair import and trade restrictions to counter those of Japan.
Andrew
BTW-My figure of 200 includes the revised number of fabs expected to be built or completed over the next 12 months. I wish I had a better handle on the number, though. I could be off by 50 fabs or more. |
