Friday morning dawned bright and fair as a collection of journalists stumbled out of a Dresden city center hotel ready to be bussed to GlobalFoundries' Fab 1 complex on the outskirts of the city. Down one narrow street could be spied the famous Frauenkirche (Church of our Lady).
Frauenkirche from Salzgasse, Dresden
This beautiful church was built in the 18th century but was destroyed in the firebombing of Dresden during World War II. It's reconstruction, finally completed in 2005 was done as a symbol of reconciliation.
After a 20 minute drive we arrived at GlobalFoundries Fab 1 complex
Approach to GlobalFoundries' Fab 1 complex.
Ground was broken for the site in 1996 and the first 200-mm wafer manufacturing cleanroom inaugurated in 1999. That first cleanroom was later emptied of equipment and re-equiped with 300-mm chipmaking equipment. The fab now only processes 300-mm diameter wafers.
So could it be retrofitted for 450-mm wafers one day? That would certainly prevent the site becoming a "sunset wafer fab" and prevent all the energy and excitement moving across to GlobalFoundries' Fab 8 in New York state.
And as Rutger Wijburg said in the second half of 2012 more than 50 percent of the output of Fab 1 in Dresdden, which currently stands at about 50,000 wafer starts per month, will not be for AMD.
So that means more than 25,000 wafers per month heading towards new foundry customers of GlobalFoundries.
I find it somewhat comical how so many assume that GF doesn't already have volume customers contracted and lined-up for years to come... And also that AMD is the only major company-customer simply because they are the only (obvious) one publicly known. Are we to believe that they are that horribly unprepared?
Do folks truly believe that the *producing* fab in New York is simply idle and waiting for volume customers... when AMD doesn't even have a single product project in that fab? Hmm...
The skilled force can come from somewhere else jeremybirch, there are more foreigners living in UAE than locals :-) I agree that setting a fab there does not make much sense on technical grounds but ATIC owns GF, so that's a good enough reason :-) I do not see why would they would otherwise pump billions of dollars into a distant business that is not even breaking even.
GF is in the position of having to do something extraordinary to avoid becoming a distant 3rd rate player to the likes of TSMC and Samsung in the years to come. Nothing described here fits that bill, whereas TSMC and samsung continue to invest billions every year in new capacity construction and retrofit/upgrade. And they don't seem to take 4 years to get a fab up and running like GF is doing in new york. So how can GF compete and grow their business, and what volume customers (aside from AMD) would want to go to GF with those obvious limitations?
Apart from loads of oil cash to invest and a liberal supply of sand, not sure why a fab in Abu Dhabi makes any sense at all. There is precious little local skilled labour force, and an awful lot of dust to cope with!
Regarding Abu Dhabi, noithing substantial.
Rutger Wijburg, general manager at Fab 1, told me that the vision to build a wafer fab in Abu Dhabi is still in place but the timetable has been pushed back.
The original timetable called for ground-breaking in 2012 and chips out in 2014.
I suspect groundbreaking time may become more likely when Fab 8 in Malta, Saratoga County, New York is in volume production.
But the ATIC sovereign wealth fund has been burning money at a tremendous rate. DO they have the appetite for another $5 billion to $10 billion commitment in a location where almost all the workforce would have to be parachuted in.
And it would not be great news for Dresden as it would probably reduce the chances of 20-nm upgrades at that fab complex. We will see.
This is completely false. In fact, demand is such that TSMC cannot come close to filling AMDs orders alone (let alone yield well at the node), and they will continue to be a key customer at the GF site. Regardless GF has numerous other extremely promising contracts to fulfill for several years to come, hence the EXPANSION projects in both Germany and New York.
Typical GF slander-trolling post above.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.