EETimes: TSMC starts FinFETs in 2013, tries EUV at 10 nm

[Ajay]

The fact that GLF isn't advertising 20nm but going straight to 14XM, and AMD silence about 20nm SKUs are consistently coherent with this 1.16x screw up.

Yes, that is bizarre. I have no idea where GFL currently is. Last I heard, Dresden was working on 20nm and the expansion at Malta was supposed to get that process (so 28 nm & 20 nm @ both Dresden and Malta). Anybody know if Malta has even ramped up 28nm?

@IDC:
Thanks.

TO the density discussion, I could only find this from GF:
http://www.globalfoundries.com/technology/14XM-FAQ.aspx

So we have one claim with 200% from the manufacturer and one with 116% from the competitor. I guess the truth is in between, probably around 160%

Because it's a pet peeve of mine, 2x == a 100% improvement (since you already have x, you add 100% of x to double the quantity). No offense intended.

That page is OFN. And that's obvious since 20nm didn't start shipping in 2H12. Honestly, it seems like GFL has their heads buried in the sand.

 

[Khato]

Other troubling point about just making a 'gate level density' claim is that it depends upon multiple factors. If you go by the industry norm of minimum SRAM size instead then TSMC only gets a ~1.56x scaling from 28nm to 20nm (0.127um^2 for 28nm and 0.081um^2 for 20nm.) My guess is that TSMC is playing games in order to get a 1.9x scaling for themselves and 1.16x for GLF.

 

[Ajay]

Other troubling point about just making a 'gate level density' claim is that it depends upon multiple factors. If you go by the industry norm of minimum SRAM size instead then TSMC only gets a ~1.56x scaling from 28nm to 20nm (0.127um^2 for 28nm and 0.081um^2 for 20nm.) My guess is that TSMC is playing games in order to get a 1.9x scaling for themselves and 1.16x for GLF.

Ah, not like that trick hasn't been done before, in just about every industry

I remember when some marketing guys wanted me to come up with better comparison numbers between the company I was working for an their competitor. I said that it was a standardized test and that there was no way to manipulate the numbers. They pressed me and I said if they wanted to make up numbers that was on them, but I wasn't going to. So, they fudged the numbers - I've never had the vaguest interesting market since.

 

[mrmt]

The one thing I have never understood about the AMD/GloFo exclusivity contract is, at what point does GloFo failing to deliver on promises excuse AMD from the contractual obligation?

The agreement wasn't fully disclosed (typical AMD), but we already got two exceptions involving GLF failures to deliver:

- When GLF got subpar yields in the 32nm introduction there was a provision that AMD would pay only for the good dies and pay for the bad dies only if GLF reached a specific threshold after 6 months in production. GLF got it, and AMD was on the hook for another 400 million.

- When GLF didn't get their 28nm production online AMD still had to pay 700 million to manufacture Kabini at TSMC.

Given the two examples, what I see here:

- AMD has no say on the WSA regardless of what GLF delivers, much less when.

- AMD must pick something from a range of choices determined by GLF regardless of the rest of the market.

- The only way that AMD might get the upper hand is if GLF *commits* itself to a given target, and if GLF fails AMD isn't entitled to freedom, but to compensation.

So in practice, there is no "out" clause of the WSA except under the most extreme situations.

It must be a very good contract because they haven't gotten to tear it up yet. I see that contract as a major hurdle that is severely hindering AMD's efforts at a comeback.

Or a very cost liability, one that AMD cannot afford to pay if it wants out.

 

[krumme]

Any guesses to what happened on GF since it slowed down?

It looks like it happened already with 32nm?

Btw: thanx for all the good post. Especially idc.

 

[krumme]

Samsung sells loads of phones and can always use both new and old nodes. Powerhungry or lean variants. Obvously it gives them a huge advantage. Then the can use the small players for the leftovers.

But right now everyone and his brother is giving money to tsmc, and the consolidation is accelerating. It does not strike me as the usual Samsung strategy to keep a business going witout getting shortterm profit. I dont see they can continue this run. Or what is the purpose for them then?

This looks like commodity market to me unlike oled. Why use money on something that is well good enough for angry birds.

 

[Imouto]

Samsung is also and by a wide margin the top NAND seller.

 

[mrmt]

This looks like commodity market to me unlike oled. Why use money on something that is well good enough for angry birds.

Samsung poached an entire Opteron team from AMD. They are *not* aiming for something good enough for Angry Birds.

 

[cytg111]

High volume 20nm in 2014, but FinFET in 2013?

What are these guys smoking?

Hey, even pipe dreams make it from time to time .. look at charlie sheen, its doable

 

[zebrax2]

I'm still scratching my head why there is still no cases filled at least by the AMD investors to the masterminds behind the GF-AMD deal. AMD right now is almost like a sacrificial lamb for GF

 

[Homeles]

I'm still scratching my head why there is still no cases filled at least by the AMD investors to the masterminds behind the GF-AMD deal. AMD right now is almost like a sacrificial lamb for GF

Yeah, except AMD is a sick, malnourished lamb. The economic gods are not pleased with the offering.

AMD will not keep GloFo afloat. GloFo doesn't really have much reason for existence other than AMD, and once GloFo's ludicrous wafer agreements finally suck the life out of AMD, they'll have nothing else to keep the con going. No one beats TSMC at what has become their own game.

 

[Imouto]

Yeah, except AMD is a sick, malnourished lamb. The economic gods are not pleased with the offering.

AMD will not keep GloFo afloat. GloFo doesn't really have much reason for existence other than AMD, and once GloFo's ludicrous wafer agreements finally suck the life out of AMD, they'll have nothing else to keep the con going. No one beats TSMC at what has become their own game.

Really dunno why ppl keep saying this. GloFo (31%) grew more than TSMC (18%) in 2012 while AMD was cutting down its orders. Its grow can easily overcome losing AMD.

Not to talk about Samsung almost doubling its foundry sales for two years in a row.

 

[mrmt]

I'm still scratching my head why there is still no cases filled at least by the AMD investors to the masterminds behind the GF-AMD deal. AMD right now is almost like a sacrificial lamb for GF

Because AMD was essentially bankrupt at the time of the spin off. Were not for the operation, AMD would be unable to pay debt and would have to file for bankruptcy rather sooner than later. As bad as it was, and there is no question that the WSA was a very bad deal for AMD, it was the agreement that kept the engine running.

GloFo doesn't really have much reason for existence other than AMD

Not really. Since GLF acquired Chartered this is no longer true. By the beginning of 2012 AMD was only 20% of the revenues, with the last ammendments in the WSA they should stabilize around 15-18%. So even if AMD goes kaput tomorrow, there's still plenty of volume to GLF to work with. The main problem is that the bulk of this volume is n-2, 65nm.

AMD utility for GLF is to be a poster child of the foundry in a bleeding edge process, but given the latest screw ups, AMD is becoming a poster child of GLF incompetence in implementing bleeding edge nodes.

 

[ShintaiDK]

Really dunno why ppl keep saying this. GloFo (31%) grew more than TSMC (18%) in 2012 while AMD was cutting down its orders. Its grow can easily overcome losing AMD.

Not to talk about Samsung almost doubling its foundry sales for two years in a row.

Its easy to post that when you are tiny. TSMC is still 3½ times bigger than GloFo. And TSMC sits on 28nm. TSMC grew 2½ more than GloFo in relative terms.

 

[Imouto]

Growth in what? One of TSMCs gigafabs produce more than the entire GloFo.

And what does size has to do with growth?

You give way too much credit to companies on the top. All your posts picture gloom and doom for every other company and that's just not true. Intel, Nvidia, TSMC, it doesn't matter which one, competitors are dead meat since they're not on the top.

 

[ShintaiDK]

And what does size has to do with growth?

You give way too much credit to companies on the top. All your posts picture gloom and doom for every other company and that's just not true. Intel, Nvidia, TSMC, it doesn't matter which one, competitors are dead meat since they're not on the top.

Its all about volume and R&D. Its all nice if you wish to root for the little one. But reality also comes to play.

TSMC makes money, GloFo is just another AMD.
http://news.techeye.net/chips/globalfoundries-will-lose-cash-for-years

 

[Imouto]

Its all about volume and R&D. Its all nice if you wish to root for the little one. But reality also comes to play.

TSMC makes money, GloFo is just another AMD.
http://news.techeye.net/chips/globalfoundries-will-lose-cash-for-years

Tell to those guys at ATIC. They seem to have insane amounts of money to burn and by the way it looks they won't stop throwing money into GloFo.

 

[ShintaiDK]

Tell to those guys at ATIC. They seem to have insane amounts of money to burn and by the way it looks they won't stop throwing money into GloFo.

People dont keep throwing good money after bad. Nomatter how rich they are.

And if that was the case, why didnt they spend more? Then they wouldnt have GloFo to be a second rank foundry.

The rationale about "insane amounts of money" simply dont hold water.

 

[AtenRa]

People dont keep throwing good money after bad. Nomatter how rich they are.

And if that was the case, why didnt they spend more? Then they wouldnt have GloFo to be a second rank foundry.

The rationale about "insane amounts of money" simply dont hold water.

GlobalFoundries Fab 8 in Malta going from $6 Billion invested to $8 Billion: adding 1,000 jobs

Globalfoundries Expected to Apply for Permission to Build Fab 8 Module 2.
(Globalfoundries to Start Making Preparations for Assumed 450mm Fab Shortly)

They heavily investing money in their fabs around the globe, they have the money to spend you know, they lose one billion from GloFo, they make 10B from oil.

No need to panic, they have oil reserves for more than 10 years

 

[Imouto]

Those guys are investing in tech and renewable energy all around the world. They also threw $8 billions into the Abu Dhabi facility which is expected to run under renewable energy too.

 

[ShintaiDK]

GlobalFoundries Fab 8 in Malta going from $6 Billion invested to $8 Billion: adding 1,000 jobs

Globalfoundries Expected to Apply for Permission to Build Fab 8 Module 2.
(Globalfoundries to Start Making Preparations for Assumed 450mm Fab Shortly)

They heavily investing money in their fabs around the globe, they have the money to spend you know, they lose one billion from GloFo, they make 10B from oil.

No need to panic, they have oil reserves for more than 10 years

Oh yes, the faithful hopes in the oil. How far have GloFo come until now? The company is a disaster. And for NY, how much is subsidized? 25%?

So GloFo invest..1.5-2 billion$? TSMC invest 9+ billion$.

 

[mrmt]

Tell to those guys at ATIC. They seem to have insane amounts of money to burn and by the way it looks they won't stop throwing money into GloFo.

The fact that Mubadala (ATIC was acquired some time ago, bankrupt when the bond market in the UAE went bust) has very deep pockets doesn't mean that they are willing to sink all this money in GLF.

Yes, they are investing heavily, yes, they are willing to lose a lot of money and wait for a lot of time to have profits, but this is far from a blank check. People assume that just because the UAE wants to diversify its economy it is willing to burn all their cash reserves without any kind of criteria, and you have to look no further than the oil industry to see that they are keen to make money.

With plenty of profitable business opportunities out there, why would Mubadala burn all their cash with two unprofitable companies? (AMD and Globalfoundries)

 

[mrmt]

Those guys are investing in tech and renewable energy all around the world. They also threw $8 billions into the Abu Dhabi facility which is expected to run under renewable energy too.

Ops, sorry, the Abu Dhabi facility was canned. Not enough money for it.

 

[Imouto]

Ops, sorry, the Abu Dhabi facility was canned. Not enough money for it.

I think it's more a "sort your shit out before going forward" than lack of money. GloFo is obviously in trouble but it's not even close to go kaput.

 

[krumme]

With plenty of profitable business opportunities out there, why would Mubadala burn all their cash with two unprofitable companies? (AMD and Globalfoundries)

Isnt it what you call sunk cost?

I dont know what Mubadala wants other than fairy tale storries, and their strategy is quite unknown to me. And i seriously doubt anyone inside AMD or GF knows it either.

But anyway this burning of money kind of reminds me why Intel wants to compete with TSMC, Samsung and everyone and his brother, to sell mobile cpu.