|
Between the lines: Qimonda |
|
|
|
Apr 26, 2007 at 12:43 PM |
The news that Qimonda is to build a new 300mm fab in Singapore is no real surprise as its newest 300mm fab in Richmond, Virginia has been ramping at a good pace since Infineon spun off the division.
Full capacity at Richmond has been touted as being about 40,000wspm in the past, but it could be closer to 50,000!
Anyway, it started ramping from 25,000wspm to 40,000wspm about 9 months ago and so there isn't much headroom left.
However, I am still curious as to why Qimonda didn't carry on its JV agreement with Inotera when Fab 2 was started! The JV had been a success and helped Qimonda share both risk and profits from this extremely cost efficient facility.
Singapore is also a good choice, but why not build in Taiwan?
It was noted in the press release by Kin Wah Loh, President and CEO of Qimonda, that:
"In Singapore, we have found excellent conditions. The overall package of low taxation, incentives and factors such as highly skilled labor and strong infrastructure makes Singapore our place of choice to implement our fully-owned volume production in the Asian market."
Singapore has not been very attractive a location for 300mm fabs in the past. Last year, though, we saw IM Flash announce that a NAND flash fab would be built there, and the conversion of its 200mm fab to 300mm for DRAM production, but nothing had happened in Singapore since UMC and Infineon built a JV 300mm fab in the country in 2001.
It may be that the Singapore Government has now produced better incentive packages than in the past because they are concerned about the incentives offered to chip manufacturers in China!
Interestingly, Qimonda believes that building its own fabs continues to be a viable enterprise!
The reason for my saying this is that Qimonda has led the way in outsourced memory partnerships. Qimonda has actually only built one purpose-made 300mm fab in Dresden; the Richmond fab was originally built as a 200mm facility.
Since that time, Inotera Fab 1 has been Qimonda's most recent facility but the fab capacity isn't all Qimonda's. This tells me that Qimonda was happy being asset-lite and was successful doing this in a commodity market.
However, reading between the lines of the press release may show that there something else being said!
"With the new fab, we put ourselves in the position to fully benefit from our technological expertise, to drive our product roll-out more rapidly and to leverage economies of scale in Asia," said Kin Wah Loh.
In fact, several things are being revealed. Firstly, Qimonda may be saying that it feels its JV partners (Inotera, Winbond & SMIC) are not fully utilizing Qimonda's wafer processes or obtaining the manufacturing efficiencies it feels it can do internally.
Secondly, Kin Wah Loh hits on a sore point in respect to product roll-out. We have noted many times before that SMIC in particular had a torrid time gaining sufficient yields for both Qimonda and Elpida, which meant that planned bit-growth was never achieved at that time. Importantly, this was on past the technology node so the profits from this were slim by the time the ramp happened.
It was actually worse for Elpida who were forced to halt yield work on one node and shift the next at SMIC - otherwise the exercise would have pointless!
The final point here is the ‘economies of scale' aspect to Loh's comments. Although both Inotera fabs are 60,000wspm facilities, only half of the production is Qimonda's. The current trend in memory is to build a minimum capacity of 60,000, a trend that is continuing to move upwards.
It is no surprise then that Qimonda's new fab will be the same size at Inotera's two 300mm fabs, which are both 60,000wspm-plus facilities. But why not build bigger seeing that that is the trend?
Overall, this announcement by Qimonda is quite significant as it demonstrates that the company still believes in owning its own fabs and that Singapore is still able to compete for fabs, despite the free cash in China!
|
|
|
Blogs 
