facebook rss twitter

TSMC CEO talks about firm's 5nm process plans

by Mark Tyson on 19 January 2016, 10:01

Tags: TSMC

Quick Link: HEXUS.net/qacxxw

Add to My Vault: x

While we are still waiting for many key performance components in our computers to graduate from the 28nm process, foundry TSMC has already started talking-up its 5nm plans.

TSMC co-CEO Dr. Mark Liu told attendees at a recent investors meeting that his firm would be ready to roll out its 5nm process technology two years after its 7nm node was launched. The production of 7nm chips is scheduled for H1 2018. However it wasn't made clear if 5nm would be ready for test samples or mass production in that timescale.

Liu provided a little background detail on 5nm development. He told the investors that the node had been in R&D for a year already and would likely use extreme ultraviolet (EUV) lithography. "We've made significant progress with EUV to prepare for its insertion, likely in 5nm," explained the co-CEO.

Coming closer to the present day Dr. Liu indicated that 10nm would be ready for customer tape-outs in Q1 2016 - that's very soon.

In a similar timescale, sometime during this quarter, TSMC's new 16nm FFC node, a low-power and low-cost version of TSMC's 16nm FinFET products, will be ready for volume production. We are hearing more and more about the strong competition offered by Samsung in this industry but TSMC says that it will boost its share of the 14/16nm foundry market over the coming year. TSMC's other co-CEO, CC Wei, said that TSMC should capture 70 per cent of this market segment in 2016, rising from last year's 40 per cent slice of the cake.

One further stream of income was mentioned in the investor meeting. Wei said that TSMC is in track to begin volume production of its integrated fan-out (InFO) wafer-level packaging technology in Q2 2016. He expects "a few very large volume customers," one of which, DigiTimes expects, will be Apple.



HEXUS Forums :: 5 Comments

Login with Forum Account

Don't have an account? Register today!
WOW! a GTX 980 Ti on 5nm Fab will only need Pci-E bus power to run.
Maybe things have changed or what i read was wrong, but doesn't cost per wafer start climbing when going under, something like, 14nm Finfet? If so I'm not sure the extra power savings would be worth the extra costs.
Corky34
Maybe things have changed or what i read was wrong, but doesn't cost per wafer start climbing when going under, something like, 14nm Finfet? If so I'm not sure the extra power savings would be worth the extra costs.

Cost per wafer have been climbing for years. If the cost per wafer goes up 50% but you can cram on twice the transistors then your cost per transistor is still better so on things like GPUs it is well worth changing to the new process.

Problem is, cost per transistor isn't really improving now either. I am expecting some stonking GPUs this summer, but I don't expect bargains.
That's what i must have been thinking of, cost per transistor, i thought historically cost had fallen as fab processes shrunk until we reached 20-28nm, Finfet was a temporary solution to that but it just shifted the size when costs started to climb when going under something like 10nm.
Corky34
That's what i must have been thinking of, cost per transistor, i thought historically cost had fallen as fab processes shrunk until we reached 20-28nm, Finfet was a temporary solution to that but it just shifted the size when costs started to climb when going under something like 10nm.

Finfet is way of getting around leakage currents that otherwise cause power loss on a modern process. It makes costs worse, but is necessary.

The cost problem seems to be largely down to the difficulty in using 193nm ultraviolet light to draw features only 14nm across. Immersion lithography helped for a while, using the refractive index of water to shorten the wavelength of light, that bought about 40% improvement. But now, they are using quadruple pattern lithography https://en.wikipedia.org/wiki/Multiple_patterning and the smaller the features are the more masks have to be used. I think that is the crux of the problem, it used to be that you needed harder to make masks from process to process but now you need more of them.