*AV*--With a treasure chest of $1.6 Billion, any upstart company that poses a threat to AMAT would be gobbled up in a second. IMO. AMAT has a history of going after what it wants and succeeding along with licensing stuff, if necessary. As you sated, it is very diffcult to get the required capitalization to compete with the big boys.
Don't get me wrong. I would love to see an upstart give AMAT or the others a run for their money but it is unlikely. Look at ISI. They couldn't go it alone, couldn't sell themselves to AMAT, and let themselves be acquired by UTEK. The enormous costs to develop new equipment for 300mm or to market the equipment and gain customer acceptance is getting more difficult everyday. You are seeing more and more partnering as time goes on. Having the best thing since bread means nothing if you cannot support it with parts, field service, technical support, applications support, etc. It is very difficult to do this while looking down the road for enhancements and new products from R&D. And finally, and more unfortunately, the rush for self gratification and wealth will prpbably cause most of these upstarts to sell their companies earlier in order to reap financial rewards instead of gutting it out for a few years on their own. The money might be in developing and implementing process modules, as you suggest, on the big boys equipment and then licensing the technology to the big buys. that counds very plausible and it has occurred before.
I think we could see USA and Europe firms putting in new plants in the depressed Asian sector just because the talent is there, and these
new Fabs are so goddamed expensive. But most of that cost is equipment from the West. So maybe not. But even if they were, the profits would go to USA companies? Hmmm. Dunno.
The US has been doing this for a great while. Fabless companies as well as fab'd companies have partnered with the Pacific Rim IC producers for year in either joint ventures or purchase of foundry capacity. A good deal of the equipment does fall to the bottom lines of the equipment companies in the US but a great deal of $$$ is lost when the output ICs come from the Pacific Rim. This is almost the root of our present crisis. The unbridled expansions in Asia caused and over demand for equipment when the sector really could not sustain it. Lower interest rates and overbuilding of plants came back to haunt them. Building of Foundry relationship and JVs contributed to more equipment being shipped there than was necessary. A basic lack of foresight based on shrinking die, increased density/complexity, lower defect densities, higher yields and larger diameter wafers and a decrease in the amount of chips required all created a mess that they were TOO STUPID to see. The attitude of, "if we build it, we can and will fill it," backfired on them. foundries are capable of running high volume 0.25u devices but none have been ramped up to satisfactory levels.
Lets go back to the SIMM/DIMM memory modules as an example.
1. Oldest motherboards had 8 slots for 8 SIMMS. Each SIMM had 8 or 9 chips, depending on parity. This meant 64 to 72 DRAM chips for 8 Meg of memory total (8 1 meg SIMM modules). You had to add 4 modules together in 2 banks.
2. 4 Meg DRAM chips came out and we saw 3 chip modules consisting of 2 - 4 Meg chips and 1 - 1 Meg chip. This brought the 8 slot total to 24 ICs.
3. 4 Meg SIMM modules make their debut. Automatically you wither had to fill up your empty second bank or replace 4 of the 1 Meg SIMM Modules. If you replaced the 1 Meg Modules, you recycled the 1 Meg modules through friends or through companies that accepted them as trade in. Market for 1 Meg Modules drops off. Your system now probably has 20 Meg of mwmory and capable of having 32 Meg is you get rid of all the 1 Meg SIMMS and replace them with the 4 Meg variety.
4. Next you get the 1 bank of 4 SIMMs and 1 bank of 2 SIMM motherboard popping up. The original 30 pin modules fit in the first bank of 4 slots but the second bank is now a 72 pin module that is added in pairs. Now we have a reduction in the number of DRAMs chips needed per system. We had 64-72 in the beginning but now we have reduced it to 32-36 on the first bank and 16 to 18 in the new 72 pin module variety.
Fully loaded you have 48-54 DRAMS per max'd out configured system for a whopping total of 24-48 Meg of memory since you are able to get 8 Meg or 16 Meg modules in the 72 pin variety.
5. We then moved onto the 4 - 72 pin modules, with 2 banks of 2 modules. Using 16 Meg SIMMS you have lowered your DRAM IC requirements to 16-18 chips to get 32 Meg on just 2 modules and 64 Meg if all banks are filled. Low and behold, you get the 32 Meg Modules boosting your IC needs a bit but maxing your 4 bank system out at 128 Meg. Meanwhile, you only have to switch out baks of 2 SIMMs to do an upgrade. I won't even get into the 64 Meg Modules here.
6. Finally, you move to the combo of 2 or 4 72 pin SIMMs and 168 DIMM module or the most recent systems with just 2 DIMM slots.
It doesn't take a rocket scientist to figure out that we went from 8 Meg of DRAM requiring 72 ICs to 256 Meg of DRAM system memory using 16 chips. This coming at a time when we were shrinking the die to get more die per wafer AND increasing the diameter of the wafer to get more die per wafer AND improving the process to improve yields AND expanding the industry output of the number of wafers.
I do not even want to discuss the 64 Meg chips in production and the introduction of the 256 Meg DRAMs over the next few months. When you sit back and reflect on this you have to realize that STUPIDITY reigned supreme here and the DRAM prcing wars creamed everyone. You can't spend $2 billion on a wafer fab expecting $50 die and then ramp it up and find out the prices have dropped below $5, an order of magnitude. Unless we find a use for a great many DRAMs, these producers are in dire straits. Even as DRAM prices look to be firming up, all it takes is 2 companies to ramp production and we are back in deep doo doo.
The long winded response above was to illustrate the point that these high volume, low mix facilities need a huge attitude adjustment and use these very expensive fabs to run high mix, lower volume IC devices that carry higher ASPs and margins that fill a void in the ASIC, Analog, Logis, etal sectors. They need to lick their wounds and to partner, humbly with those companies that will need wireless communications type ICs. Way too many DRAM supplers. Once they reaize this and take the appropriate actions, they will be back on the road to recovery faster and in better financial condition.
Along the way, strategic upgrades will be made to allow for an equipment sector recovery and the eventual expansion of this industry. There are enough similarities in the processes to make this possible but there are enough differences to make it necessary to re-allocate equipment and to purchase other equipment to relieve certain capacity limiting restrictions.
Andrew |
