AMD splits in two...

I'm with silent_guy here. How could The Foundry Company possibly compete with TSMC, which just has so many many competitive advantages? Scale, expertise, and relatively low labor costs being soem of these advantages.

labor costs in semiconductor fabrication are a VERY minor portion of the end wafer cost. And while TSMC does have scale, all their scale is at the far end of the trailing edge. the vast majority of TSMCs manufacturing is on the trailing edge at >65nM currently.

With just its current 2 fabs, AMD has a significant volume advantage at <65nM. The foundry company will likely not be focusing on the same parts of the market as TSMC.
 
Location? There's a large amount of third party services around TSMC. Testing, packaging etc. Not sure how important that really is though.

The third party services service basically the entire semi industry not just TSMC. From the tester, dicing, packaging viewpoint, one wafer is the same as another.
 
The third party services service basically the entire semi industry not just TSMC. From the tester, dicing, packaging viewpoint, one wafer is the same as another.
I think the point was that there are a pile of 3rd parties that support TSMC locally. You can get a complete finished product from an ODM where the chip is never more than a truck drive from one stop to the next in Taiwan. Or if you want a zillion choices, take a short hop over to the mainland.

How many 3rd party services are there in Dresden or Albany? Sure AMD has the complete test/dice/package flow for exactly what their chips need, but how flexable is it?

I guess that's the main question for this split: how flexible will the Foundry end up being, and how will they get to that point. In-house fabs are know for having extremely strict design requirements and lots of little quirks, which they can get away with since there's joint development on the chip layout and the process. Are they going to make a large investment to make the fabs and flows more accessible to customers, when it doesn't help AMDs designs at all? Or are they going to try to get customers with all the restrictions in place?
 
They're the de-facto standard for fabless companies. If you ever need a piece of silicon IP, standard cells, IO pads, Ethernet/USB/PCIe/... transceivers, fuse boxes, ram compilers, PLL's etc, you're going to find them first (and, in many cases, only) for TSMC processes.

Their only customers are external ones and they have years of experience in keeping them happy. Company dedicated fabs are often used to one particular tool flow which makes then less flexible.

They have massive economy of scale and should be able to get the best pricing wrt new equipment. They are able to partially fund new fabs with older fabs that are already written off (6 according to their website.) AMD has never had that luxury and TFC won't be able to do so any time soon. It won't be easy to compete on price with TSMC.

They're a safe bet. They're highly profitable and aren't going anywhere. It's to be expected that AMD will have priority when new state-of-the-art processes are coming on-line and that other customers will have to wait in line.

Location? There's a large amount of third party services around TSMC. Testing, packaging etc. Not sure how important that really is though.
While that analysis is likely accurate for the short term, long term I believe TFC will provide competition. The investors need scale to get a return on their investment.
 
labor costs in semiconductor fabrication are a VERY minor portion of the end wafer cost. And while TSMC does have scale, all their scale is at the far end of the trailing edge. the vast majority of TSMCs manufacturing is on the trailing edge at >65nM currently.

With just its current 2 fabs, AMD has a significant volume advantage at <65nM. The foundry company will likely not be focusing on the same parts of the market as TSMC.

Some people seem to think that labor costs are not all that relevant to capital intensive industries. For those that think that I owuld just look to any number of capital intensive industries that have moved to China. There is a reason.

It will be interetsting to see how many AMD employees will be with the design company and The Foundry. I don't know of any business where labor costs are not material. Even if it 25% of operating expenses, if your labor costs are half of what the next guy is paying, that's a significant margin advantage.

It might not seem all that relevant whne there is a company like Intel out there, but those margins are protected by monopoly money.
 
Some people seem to think that labor costs are not all that relevant to capital intensive industries. For those that think that I owuld just look to any number of capital intensive industries that have moved to China. There is a reason.

name one CAPITAL intensive industry that has moved to china.

It will be interetsting to see how many AMD employees will be with the design company and The Foundry. I don't know of any business where labor costs are not material. Even if it 25% of operating expenses, if your labor costs are half of what the next guy is paying, that's a significant margin advantage.

In Fabs, people are a very MINOR cost. Things like water are bigger issues than labor. The number of employees actually involved with the fabs themselves are pretty small for a modern day fab.
 
name one CAPITAL intensive industry that has moved to china.



In Fabs, people are a very MINOR cost. Things like water are bigger issues than labor. The number of employees actually involved with the fabs themselves are pretty small for a modern day fab.

Could you be more ridiculous?

TSMC has 20,000+ employees. It does not matter if they are in bunny suits or pulling teeth. They are part of TSMC. My guess is considering how efficient TSMC is compared to its competitors that most of the employees are productive and necessary to the business in some way. The vast majority are in relatively low cost regions.

What is the cost of these 20,000 employees. If total salaries and benefits are 100k/yr that is $2 billion dollars. Even if is $50k/yr or somewhere in between that number is not insignificant.

As for capital intensive industries that have gone to China you must be joking. What hasn't, including semiconductor manufacturing? Much manufacturing is capital intensive. Just because it doesn't cost $3 billion to make a plastic toy factory doesn't mean it's not capital intensive relative to the industry it's in. Textiles, electronics, plastics, chemicals, LCDs, tires, I mean the list goes on forever...
 
What is the cost of these 20,000 employees. If total salaries and benefits are 100k/yr that is $2 billion dollars. Even if is $50k/yr or somewhere in between that number is not insignificant.

I don't know for taiwan, but at least here in brazil, where you have you can find an upper middle class can live with the same luxury as a average middle class in US, a really good salary means 30k/yr, because the life costs are absurdly much lower in comparison.
 
From the NYTimes follow up story:
http://www.nytimes.com/2008/10/08/technology/business-computing/08chip.html said:
The Advanced Technology cash infusion will allow the Foundry to upgrade one of the Dresden plants so that it can be used to make cutting-edge processors for a wide variety of companies within a year. A top priority will be to adapt the plant to bid for A.M.D.’s graphics chips business. Taiwan Semiconductor Manufacturing, one of the leading contract chip factories, makes those products today.

Down the road, Foundry hopes to leverage a technology alliance with I.B.M. to woo even more customers. I.B.M. has formed semiconductor partnerships with Toshiba, Samsung, Sony and others — all of whom could turn to Foundry for extra capacity.

So it looks like they will be trying to make their fabs generic. Wonder when we can expect an AMD fab'd ATI chip. I doubt R800. R900 maybe? Past that?
 
Could you be more ridiculous?

TSMC has 20,000+ employees. It does not matter if they are in bunny suits or pulling teeth. They are part of TSMC. My guess is considering how efficient TSMC is compared to its competitors that most of the employees are productive and necessary to the business in some way. The vast majority are in relatively low cost regions.

now answer how many of those employees are in the services side vs the manufacturing side?

As for capital intensive industries that have gone to China you must be joking. What hasn't, including semiconductor manufacturing? Much manufacturing is capital intensive. Just because it doesn't cost $3 billion to make a plastic toy factory doesn't mean it's not capital intensive relative to the industry it's in. Textiles, electronics, plastics, chemicals, LCDs, tires, I mean the list goes on forever...

most manufacturing IS NOT capital intensive, it takes money to make the factory but labor is still a significant portion of the cost.

LCDs, at least not real display LCDs (vast majority of the factories are in Korea, Japan, and australia/singapore), have not moved to china, all the others you've listed are much more labor intensive than capital intensive. In fact most economists would flatly state they were purely labor intensive. Plastics, chemicals, tires, PCBs are all LONG term (20+ year) investments, while fabs are at best 8-10 year investments and a single fitting is more like 4.
 
Back
Top