TSMC Confirms Plans to Start Low-Volume 20nm Manufacturing Next Year.

TSMC Expects to Produce 20nm Chips Only in Low Volumes

[07/22/2012 01:11 PM]
by Anton Shilov

Taiwan Semiconductor Manufacturing Company reiterated plans to start making chips using 20nm technology, but admitted that it would be a low-volume production. The head of TSMC believes that the world's largest contract semiconductor manufacturing will start to make chips using 20nm technology in volume only in 2014.

"We will start some production of 20nm next year, but the small scale, very low, what we would call a risk type of production basically. The 2014 will be a ramp year for 20nm SoCs," said Morris Chang, chief executive officer of TSMC.

TSMC plans to offer only one version of 20nm technology that fits to different applications, from ultra-mobile chips that power smartphones to ultra high-performance that are used on graphics cards.

Nowadays Taiwan Semiconductor Manufacturing Company offers four versions of 28nm process technology: 28LP (poly/SiON) for low-power cost-efficient chips, 28HPL (HKMG) for low-power applications, 28HP (HKMG) for high-performance chip designs and 28HPM (HKMG) that combines elements of high-performance and low-power process technologies and is mostly designed for microprocessors for tablets, superphones and notebooks. With 20nm, TSMC will not give any choice and there will be only one option.

TSMC also confirmed plans to use FinFET technology with its 16nm-generation of manufacturing processes in 2015.

"We believe that the 16nm FinFET will ramp in, perhaps the second half of 2015," said Mr. Chang.

Tags: TSMC, 28nm, 20nm, Semiconductor

Discussion

Comments currently: 10
Discussion started: 07/22/12 07:29:58 PM
Latest comment: 07/28/12 09:25:31 AM
Expand all threads | Collapse all threads

[1-4]

show the post

5 [Posted by: beenthere Rating 394.92

| Date: 07/22/12 07:29:58 PM]

Reply

just go up with 16nm and forget about 20nm. get ahead of the game. lead and control or would you be follower and let other pull you by the nose.

0 [Posted by: idonotknow Rating 26.34

| Date: 07/23/12 01:52:43 PM]

Reply

Hi,
Dose any one know why they don't jump from let say 28nm directly to 16 why always make it half.
and why not go from 300mm wafer to 600mm and not 450mm.
Is this a biz game or impossible technologically?

By the way, Intel's CPU is to expensive just because no body can compete. and they take over GPU too, bad for other Company like Nvidia. I don't like Intel's domination.

2 [Posted by: JanGozi Rating 5.78

| Date: 07/24/12 01:07:11 AM]

Reply

- collapse thread

show the post

4 [Posted by: BestJinjo Rating 571.38

| Date: 07/24/12 07:37:52 AM]

Reply

Thanks,
I thought if they research 2 step ahead would be cheaper. but maybe it is not possible.
I know ARM CPU is smaller and gives more performance par watt compere to Intel CPU. and they are moving fast. I think it all depend on how open minded you are in relation to innovation and technology.
And I know they (the big Company) are not always right and honest.
I have worked in Nokia as software engineer and I know some about it.
that is why some of them goes down like Nokia.

0 [Posted by: JanGozi Rating 5.78

| Date: 07/28/12 09:25:31 AM]

Reply

Not only the process but also the tools and so on, have to be researched first.
U cannot use tech that isn't there yet, or jump over several borders.
Going straight to 600mm wafers it's the same, it wouldn't make sense yet.
Why not produce in 4nm, why not go for a 500 hp car that only needs 3 litres?
Why not NOW make this and that and so on, you need to research things before you can use them in mass production ;-)

0 [Posted by: Rollora Rating 15.17

| Date: 07/24/12 10:40:33 AM]

Reply

again 300mm to 450mm size wafers is a little more than doubling. alse you get problems whit bending (thicker wafers) and speed. so goong from 300 to 600mm would make the problems even bigger , higher cost etc.

0 [Posted by: massau Rating 74.47