TSMC 90nm production schedules

[Entropy]

Since it's been discussed here a few times:

Taiwan Semiconductor Manufacturing (TSMC) projects to start volume production of 90nm logic IC products for its US-based IDM clients at its 12-inch Fab 12A by the end of this year, said a company source.

The source also projects that 90nm production at Fab 12A will reach over 1,000 wafers per month by year-end and overall quarterly output at Fab 12A should cross 40,000 wafers (8-inch equivalent) by the end of this year.

In addition, TSMC recently assigned J.K. Lin, the former head of Fab 6 who helped to ramp up 0.13-micron capacity at the fab, as the new head of Fab 12A to shorten 0.13-and-below process development time at Fab 12A.

TSMC began volume shipments of 90nm SRAM at an 8-inch fab in the first quarter.

The article is quoted whole from here:
http://www.digitimes.com/NewsShow/Article2.asp?datePublish=2003/07/29&pages=06&seq=34
The schedule is in line with earlier reports.

Entropy

 

[KimB]

Hopefully TSMC will do better with 90nm than they did with 130nm...

 

[andypski]

Why? Was there some problem with 130nm?

 

[SvP]

 

[Dave Baumann]

Its a valid point - who has been having significant issues with TSMC's 130nm process now?

 

[SvP]

I know it's a valid point (assuming you are reffering to my informative post). The smiley is not targeted at andypski's post, but to he fact that in the context of Chalnoth andypski's remark was quite entertaining.

 

[Evildeus]

Its a valid point - who has been having significant issues with TSMC's 130nm process now?

It's not because someone doesn't have any issue now, that he didn't before

 

[Dave Baumann]

ED, the point is that one company has gone out there and successfully besmirched TSMC’s 130nm process – TSMC warned them of the issues and they chose not to pay head to those issues, and then publically moaned about the apparent fabrication issues they are having. Wind on nearly a year and that same company is publicly warning before their investor due to 130nm process issues, while their direct competitor is publicly saying to their investors via conference calls, and to us directly that they have good yields on 130nm. Now, if one does actually have good yields but the other is warning of poor yields (and this is on the mainstream product) does that suggest there is a fab issue or a design issue with that one manufacturer? How many other people using TSMC’s 130 process are currently warning of yield issues?

 

[Evildeus]

Yes DB, i see what you are saying. But for one to say that yield is below what they expected, and the other that they are good, doesn't mean that one yield is better (or way better) than the other one. It depends on what you expected at first

 

[Dave Baumann]

Well, the financials are telling that story. The one that is saying the yeilds are good have rising margins, while the one saying yeilds are bad have falling margins - and they are both paying per wafer...

 

[KimB]

Its a valid point - who has been having significant issues with TSMC's 130nm process now?

Regardless of who has had problems with it, TSMC has been rather slow on getting .13 micron up to speed. They usually aren't far behind Intel and AMD (frequently with an "intermediate" process that is ahead of them), but this time, with .13, they were what, 1.5 years behind?

 

[Evildeus]

Hmm sorry, but the financials respond to the fact that for one the yield is good and for the other it's less than expected, that doesn't mean that the yield is not the same.

Ati was conservative on 0.13, and could have been on yields, contrary to Nv.

Moreover, Ati proposes better parts for almost 1 year now. The stock and financial margins should go up don't you think?

 

[Joe DeFuria]

Moreover, Ati proposes better parts for almost 1 year now. The stock and financial margins should go up don't you think?

OK, so you've established that it's either nVidia to blame for designing inferior products, or it's nVidia to blame for having too high expectations on yield.

Got it.

 

[Evildeus]

Moreover, Ati proposes better parts for almost 1 year now. The stock and financial margins should go up don't you think?

OK, so you've established that it's either nVidia to blame for designing inferior products, or it's nVidia to blame for having too high expectations on yield.

Got it.

Because it can't be? Of course it's them to be blamed to do inferior products or exagereted expectations. It could be them or TSMC, i don't mind, they are the one who makes expectations, and they are the one to be blamed if it's not there. Like Ati makes their own expectations, and they are to be praised if they are obtained or even surpassed

 

[Dave Baumann]

Hmm sorry, but the financials respond to the fact that for one the yield is good and for the other it's less than expected, that doesn't mean that the yield is not the same.

Ati was conservative on 0.13, and could have been on yields, contrary to Nv.

You are still basically governed by the cost of the wafers. At present ATI has a roughtly equal, if not lower, selling price product but operating at high margins.

 

[Evildeus]

And? The market share of Ati is expanding, whereas the Nv's one is at best equal. I don't know what are the contracts between Ati and TSMC, but even if the margin is equal, when your operational cost is a given and you sell more, your financial margin needs to go higher, you can't tell if the way the financial margin of Ati is moving comes from the market share gain or the better yield, or both or anything else.

 

[Dave Baumann]

The operationsal costs on a product in production is the wafers you are paying for, and they likely to be farily fixed. the margins on the products are the best indication to yeilds there is.

 

[Joe DeFuria]

I don't know what are the contracts between Ati and TSMC, but even if the margin is equal, when your operational cost is a given and you sell more, your financial margin needs to go higher, you can't tell if the way the financial margin of Ati is moving comes from the market share gain or the better yield, or both or anything else.

Someone needs to state the "definition" of margins as they are reported in the financial statements. Someone correct me if I'm wrong, but the margin being talked about here is basically revenue vs. per cost of good sold?

If that's the case then market share gain has nothing to do with it. Market share gain will increase your total profits, but does is not a direct reflect on margin at all. (For every unit sold, you have a cost in producing it.)

ATI's margins COULD certainly be a relflection on yield, since essentially, the lower your yield, the higher it "costs" per chip. (As overhead costs are spread over all products sold...)

 

[Evildeus]

Huh?
You have some marginal cost and fixed cost. If you sold more the fixed cost is spread over all the chip sold, and thus increases the margin....

Btw, DT states that many clients are having issues with 0.13 micron.
http://www.digitimes.com/NewsShow/Article2.asp?datePublish=2003/07/30&pages=06&seq=43

 

[RussSchultz]

The operationsal costs on a product in production is the wafers you are paying for, and they likely to be farily fixed. the margins on the products are the best indication to yeilds there is.

Well...I think thats over simplifying.

Assume wafer cost and selling price of final product is equal, the following can affect (gross) margin:
-test time
-packaging cost
-channel expenses (distributor commission, sales commission, etc)
-yield
-die size
-fluctuation in price to close deals, or as part of ongoing business