News GlobalFoundries Expands Production Capacity With a $4 Billion Fab in Singapore

[Admin]

GlobalFoundries' new fab in Singapore to produce 450,000 300-mm wafers a year.

GlobalFoundries Expands Production Capacity With a $4 Billion Fab in Singapore : Read more

 

[InvalidError]

TSMC GloFo having to invest 4G$ in what likely is a new 14nm-class fab shows that demand for chips made on older processes is still alive and well despite popular belief that chip manufacturers who aren't investing in bleeding-edge processes are doomed. Support chips made on 22-1200nm processes may not be making headlines anymore but they are still very much needed today.

 

[mamasan2000]

GloFo, not TSMC

 

[InvalidError]

GloFo, not TSMC

Oops

Well, that makes an even stronger case about the market for ~14nm processes being alive and well.

 

[AgentLozen]

I haven't heard anything about Global Foundries in a long time. I'm glad that they're alive and well. Thriving even, it seems.

They were making 12nm chips for AMD a few years ago then they announced that they weren't planning to move to 7nm. Does anyone know if they're still making 12nm chips?

 

[InvalidError]

Does anyone know if they're still making 12nm chips?

A 7nm fab costs ~10G$ to build, so I think we can say with pretty high confidence that 4G$ must be 14nm-class.

The fab is earmarked primarily for analog, RF and power management stuff which doesn't need bleeding-edge process tech: analog stuff requires bulky relatively high current components and traces, RF requires relatively large planar passive components along with possibly significant power and power management stuff requires large transistors, all stuff that does not benefit much from process shrinks except for tighter tolerances, higher yields.

Using a 14nm-class process to make parts that would likely be perfectly fine on 45+nm does have the benefit of opening the option to make stuff down to ~14nm should the need arise.

As I have written in previous threads about GloFo's "imminent" demise, everyone will need someone to make interposers and substrates of all sorts when all major chips go with some version of 3D-stacking and 12-22nm will also be perfectly fine for that. It may not be as glorious as making the bleeding-edge parts of future chips but it should keep GloFo afloat for the next 10+ years.

 

[hotaru.hino]

As I have written in previous threads about GloFo's "imminent" demise, everyone will need someone to make interposers and substrates of all sorts when all major chips go with some version of 3D-stacking and 12-22nm will also be perfectly fine for that. It may not be as glorious as making the bleeding-edge parts of future chips but it should keep GloFo afloat for the next 10+ years.

And even then, there's still a ton of demand and use for 8-bit, 16-bit, and stupid simple 32-bit micros. When those things are literally a buck or two, it doesn't really make much sense to spend 10 figures on something that needs an astronomical amount of volume to recoup the cost.