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

[mrmt]

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.

No, it is lack of money in Globalfoundries balance sheet for the project, *and* Mubadala telling them that they won't give cash or warrant debt for this. This is just one example that perfect explains why people that think Mubadala will going to spend money willy-nilly with Globalfoundries (and AMD) are wrong.

While Mubadala is owned by the government of Abu Dhabi, they aren't funded entirely by the government. They also have bonds and debt, some earmarked to specific ventures. Those debts and loans have covenants, as usual practice on the market. So while they can take a more long term approach to investment, it is still an investment, it cannot be a charity venture.

BTW, nobody here is saying that GLF will go kaput, this is a red herring of yours.

 

[mrmt]

Isnt it what you call sunk cost?

Sunk cost =! Throw good money after bad money.

 

[Piroko]

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.

It's not even six years ago that Nokia completely dwarfed Apple in mobile volume and mobile R&D.

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

True with the money. A bit sceptical about that article though, no sources, no WSA reference and, well, a former inq writer

 

[ShintaiDK]

It's not even six years ago that Nokia completely dwarfed Apple in mobile volume and mobile R&D.

True with the money. A bit sceptical about that article though, no sources, no WSA reference and, well, a former inq writer

Nokia and Apple works with different perimeters. It cost pennies to design a phone compared. Apples entire R&D budget is 3.4 billion$ for example. And there is no factory costs.

Is this article better?
http://www.ft.com/intl/cms/s/0/d15bd0aa-84a9-11e1-b6f5-00144feab49a.html#axzz2QRSWTJaG

 

[Imouto]

BTW, nobody here is saying that GLF will go kaput, this is a red herring of yours.

Oh, cmon.

 

[zebrax2]

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.

Good for the investors before the spinoff i guess but if the details on the deal was as secretive as what I'm lead to believe how about those who invested after the spinoff?

 

[krumme]

Perhaps Mubadala is striving to be sole investor for a penny?

 

[Ajay]

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

Well, maybe you're the guy to ask this one question: Is the Malta fab running production @ 28nm yet.

 

[mrmt]

Perhaps Mubadala is striving to be sole investor for a penny?

The only thing worth the money for Mubadala is the R&D engine, and the further AMD situation gets worse, the further the R&D engine is dilapidated. If they were serious about an acquisition, they would have done so rather sooner than later.

 

[Piroko]

Nokia and Apple works with different perimeters. It cost pennies to design a phone compared. Apples entire R&D budget is 3.4 billion$ for example. And there is no factory costs.

Your original statement was that it's all about R&D and volume and I did specifically compare the same core business. Neither R&D nor volume are a guarantee for success.

Is this article better?
http://www.ft.com/intl/cms/s/0/d15bd0aa-84a9-11e1-b6f5-00144feab49a.html#axzz2QRSWTJaG

Still a year old, but does at least seem to be the original source.

There is also this, from August last year: http://electronics.wesrch.com/wequest-EL1K1ZS-turnaround-at-globalfoundries-with-ajit-manocha
It's a marketing blurb, but some good details in there.

edit:
Think I've answered Ajays' question with my link as well

 

[krumme]

The only thing worth the money for Mubadala is the R&D engine, and the further AMD situation gets worse, the further the R&D engine is dilapidated. If they were serious about an acquisition, they would have done so rather sooner than later.

Hmm perhaps. Or the R&D engine is not dilapidated the way we think, or perhaps Mubada does not care. Anyway we bend it, it looks like bad business management. Just plain bad.

At least the funny - move german technicians to sahara - project stopped.

(this is not in any way meant as a critique of fx. Marokko, or tribes of inner Somilia, as they at least dont have high cost, to burden the building of a foundry)

 

[ShintaiDK]

Your original statement was that it's all about R&D and volume and I did specifically compare the same core business. Neither R&D nor volume are a guarantee for success.

So you picked a random segment to fit your hopes in another?

 

[Idontcare]

Your original statement was that it's all about R&D and volume and I did specifically compare the same core business. Neither R&D nor volume are a guarantee for success.

Given the context of the discussion - the monetary expenses associated with developing new process nodes as well as building state of the art production facilities - I would have thought the specific nature of the R&D in that business segment was self-explanatory.

The barrier-to-entry R&D costs associated with developing a consumer product like a smartphone are no where near the barrier-to-entry costs of developing a leading edge process node on a competitive schedule.

As to "guarantee for success", you don't always get what you pay for, true, but you also don't get what you didn't pay for either.

If you don't pay the billions necessary to develop leading edge process nodes then one isn't suddenly going to be dropped into your lap, reality doesn't work that way.

But of course you can spend billions and develop crap products, incompetence works that way.

 

[Idontcare]

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.

GloFo is not close to going bankrupt, but they are on the verge of becoming irrelevant to fabless IC design houses that lean heavily on having access to the leading edge process node as a means of competing within their own business segments.

For example Qualcomm isn't going to put up the timeline shenanigans of GloFo, it would put them out of business as their competitors would eat their lunch while Qualcomm is stalled at the starting gate waiting for GloFo to get their node off the ground.

This is the fate the other foundries have suffered in the past. It happened to IBM with Xilinx and the SiLK debacle. It happened with SMIC when they couldn't get 90nm off the ground, and it happened with both Chartered and UMC when their 45nm nodes fell a year behind TSMC's 40nm timeline.

There are worse fates than bankruptcy, you could be perpetually black-balled in the industry as being non-credible. That is what happened to IBM's foundry aspirations after the SiLK situation (I know from first hand experience).

At least with bankruptcy you can emerge from Chapter 11 and hang a sign on your door that says "under new management" in a way that matters psychologically to your potential customers. GloFo won't get that opportunity, but they don't have a path at this time to remaining relevant at the leading edge.

They've pretty much ceded that space to TSMC and Samsung as far as anyone in the industry can determine.

Which is fine too, the problem isn't in finding your place, rather the problem is in refusing to acknowledge your place so you keep floundering about and getting nowhere in the meantime. GloFo will get there too, eventually.

In the meantime there is a reason, a very basic one at that, why TSMC has a virtual monopoly on 28nm and is looking very likely to repeat that with 20nm. It doesn't have anything to do with hyperbole or rhetoric, nor whimsical catch phrases and cliches. If you've ever been in the business of foundry work, and not just in the business of reading and writing about foundry work, then everything that is going on is completely to be expected and there are no surprises going on here.

 

[Piroko]

So you picked a random segment to fit your hopes in another?

No, I'm trying to get people to realize that R&D spending will tell you jack shit about the position (and future) of a company in its respective markets. The same - although to a lesser extent - is true for volume. The market doesn't care if you spend ten times more than your competition if your product fails to deliver. And spending even more in R&D does not protect you from doing wrong choices.

Given the context of the discussion - the monetary expenses associated with developing new process nodes as well as building state of the art production facilities - I would have thought the specific nature of the R&D in that business segment was self-explanatory.

If you don't pay the billions necessary to develop leading edge process nodes then one isn't suddenly going to be dropped into your lap, reality doesn't work that way.

I have a bit of a problem believing that R&D for a current leading edge process, let's say 20nm, actually costs TSMC 'billions of dollars'. For one, their total R&D has just passed 1.x b $ in 2011. And the huge increases of R&D happening with a lot of companies lately reek of accounting tricks. Additionally, a lot of R&D is done for them by tool manufacturers. But you can hurt and delay your competition by several quarters with preferred customer contracts.

 

[mrmt]

I have a bit of a problem believing that R&D for a current leading edge process, let's say 20nm, actually costs TSMC 'billions of dollars'. For one, their total R&D has just passed 1.x b $ in 2011.

But it *is* billions of dollars.

TSMC is spending 1.4 billion dollars in R&D alone in 2012, up from 1.1 billion in 2010. But this number doesn't tell the whole story. As node takes 4-5 years in the R&D pipeline, it's safe to say that the next future node will cost at least 3x the amount of money of the current R&D expenditure. 20nm Should cost TSMC some 3-3.5 billion, 16nm will cost some 4 billion.

But R&D alone doesn't mean anything, the retooling of the factory is also important, and here comes the Rock's law. "The cost of a semiconductor chip fabrication plant doubles every four years". Meaning your CAPEX double every four years. This is what crushed AMD in 2009 and this is what keep SMIC, UMC, IBM and others from competing with TSMC, Intel and Samsung for the bleeding edge node. Even if they had the R&D, they cannot afford the fabs.

So in order to research 20nm *and* build a single factory with significant capacity you need some 12 billion dollars, spread around 4-5 years. TSMC current CAPEX is 9 billion dollars, or more than the entire GLF investment alone.

And the huge increases of R&D happening with a lot of companies lately reek of accounting tricks.

Do you mind describing those accounting tricks? I have a professional interest in them.

R&D wouldn't be my first choice if I were to make an *cof* accounting trick *cof*, because the rules bounding the line are too rigid and it tends to be highly scrutinized.

 

[mrmt]

Hmm perhaps. Or the R&D engine is not dilapidated the way we think, or perhaps Mubada does not care. Anyway we bend it, it looks like bad business management. Just plain bad.

Of course it is dilapidated. Do you remember the last time AMD delivered a road map? Do you remember the last AMD chip that didn't have any kind of hiccup? Why would AMD use synthesis tools if they are more inefficient than hand-design if they weren't interested in using fewer resources for their projects? What we see here is an R&D engine that is resource starved, and everything I'm describing here starts well before 2007, when they were in the bleeding edge. In 2013, on top of the added complexities future MPUs, we have even *less* people working on the designs, and definetly out of the bleeding edge.

By looking the situation from other side, we can reach the same conclusion. Rory Read didn't look for more funds to feed a cash-starved but strong R&D engine, he changed the direction of the company in order to generate products more in line with their current (and future) R&D capabilities. Kabini, ARM, Embedded, what those markets have in common is that they are less R&D intensive than a competition against Intel.

So, to wait this R&D engine to further deteriorate would be really bad business. Even worse than to acquire it now.

 

[SocketF]

At least with bankruptcy you can emerge from Chapter 11 and hang a sign on your door that says "under new management" in a way that matters psychologically to your potential customers. GloFo won't get that opportunity, but they don't have a path at this time to remaining relevant at the leading edge.

Yes, if they will ramp 20nm in the same way as 28nm than they'll really can give up the foundry business.

I just remembered the FD-SOI story, there will be a presentation about it soon, hold by a GF guy:

SoC Differentiation using FDSOI – A Manufacturing Partner’s Perspective
Subramani Kengeri, VP, Advanced Technology Architecture, GLOBALFOUNDRIES

http://www.soiconsortium.org/workshops/hsinchu/52c9e7312d249eda36071933123639a4/index.php

Seems GF will really use FD-SOI. If that works well, it could be a way out of the dilemma. If FD-SOI has more or less the same performance characteristics as 20nm bulk, they could position that against TSMC's 20nm.
AMD could be happy too, because they could port Kabini to it and thus fulfill their contracts with GF.

In short: The lateness of GF could be evend out a bit by the FD-SOI advantages.

TSMC could work as 1st choice for AMD, GF as a 2nd choice with slightly improved architectures. Like tick-tock, just between Fabs.

The advantage for AMD would be that they could plan a bit more ahead. FD-SOI processes will always be ready *after* the bulk versions. So if there are any problem with the bulk process, AMD would have more time to react.

The only problem is the state of FD-SOI ... we probably wont hear any interesting news at the presentation mentioned above, but lets wait and see ...

 

[mrmt]

Seems GF will really use FD-SOI. If that works well, it could be a way out of the dilemma. If FD-SOI has more or less the same performance characteristics as 20nm bulk, they could position that against TSMC's 20nm. ...

Why would you want 20nm performance with 28nm costs when you have 20nm performance with 20nm costs in TSMC?

TSMC could work as 1st choice for AMD, GF as a 2nd choice with slightly improved architectures. Like tick-tock, just between Fabs.

Not gonna happen. AMD is done with TSMC as soon as they can. The exclusivity waiver is no more.

 

[itsmydamnation]

Of course it is dilapidated. Do you remember the last time AMD delivered a road map? Do you remember the last AMD chip that didn't have any kind of hiccup?

bobcat infact it was 6 months early. Trinity was fine also its delay was llano inventory based. Vishera was fine as well. Jaguar also seems to be fine, Steam roller looks like it has slipped, the question is how much.

Why would AMD use synthesis tools if they are more inefficient than hand-design if they weren't interested in using fewer resources for their projects?

its not more inefficient, that's why they are using it, thats why GPU manufactures that design far bigger yet repeated ASIC's have been using them for many generations.

What we see here is an R&D engine that is resource starved, and everything I'm describing here starts well before 2007, when they were in the bleeding edge. In 2013, on top of the added complexities future MPUs, we have even *less* people working on the designs, and definetly out of the bleeding edge.

you haven't actually described anything correctly...... s*** in s*** out. AMD's problems even during core2 era had nothing to do with R&D or money but direction. They had no plan and it showed.

By looking the situation from other side, we can reach the same conclusion. Rory Read didn't look for more funds to feed a cash-starved but strong R&D engine, he changed the direction of the company in order to generate products more in line with their current (and future) R&D capabilities. Kabini, ARM, Embedded, what those markets have in common is that they are less R&D intensive than a competition against Intel.

you can only do so much after the horse has bolted.

 

[SocketF]

Why would you want 20nm performance with 28nm costs when you have 20nm performance with 20nm costs in TSMC?

Because 20nm costs are not much cheaper due to double pattering?
Unfortunately I don't have numbers how much that costs. Would be interesting to have. But it seems not only the production process is affected, you also have to adjust the design process for it.
Thus porting from 28nm TSMC to 28nm FD-SOI@GF instead to 20nm TSMC could save additional costs.

Not gonna happen. AMD is done with TSMC as soon as they can. The exclusivity waiver is no more.

Well the current WSA is due until Q1/2014. I guess AMD will sit together with GF again and discuss the situation. The outcome is totally open in my opinion. Or do you have a good guess?

 

[ShintaiDK]

AMDs WSA with GloFo is locked until 2024.

Its a lose/lose for AMD. GloFo got them by the balls. And GloFo is the foundry joke of the semiconductor industry.

 

[Idontcare]

I have a bit of a problem believing that R&D for a current leading edge process, let's say 20nm, actually costs TSMC 'billions of dollars'. For one, their total R&D has just passed 1.x b $ in 2011. And the huge increases of R&D happening with a lot of companies lately reek of accounting tricks. Additionally, a lot of R&D is done for them by tool manufacturers. But you can hurt and delay your competition by several quarters with preferred customer contracts.

I don't think you understand how a new node is developed, which is probably why you can't believe the numbers.

My piece of the process development pie alone, just one little process development engineer, was roughly $8m/year for the fab work. Then there are the ex-fab costs, for example I had a $2m NSF federal grant that required dollar-for-dollar matching from TI over just 2yrs.

And I was just one engineer; I worked with literally hundreds of other engineers who all had to be equally resourced to get their job done. And even then, getting the job "done" was a four year journey for each node, longer still if you got into the path-finding research (like the NSF grant).

You mention tool vendors; I stand with my mouth open in aghast if you think tool vendors develop production worthy processes with their tools. Not only that, do you really think the tool vendors are charities who develop processes (and R&D expense on their books) and do not amortize that expense back into the sales price of the tools?

You don't get anything for nothing, if you think you are getting something for nothing then you are getting the wool pulled over your eyes. That is true in every industry, not just semiconductors.

There is a reason so many companies choose to be fabless. And the trend is decidedly uni-directional.

 

[mrmt]

Because 20nm costs are not much cheaper due to double pattering?
Unfortunately I don't have numbers how much that costs. Would be interesting to have.

Given that nobody in the 1st tier chose to go GLF 28nm FDSOI over TSMC 20nm, quite the opposite, everybody is lining up for 20nm, I guess we have an empirical evidence of what those numbers would show.

Well the current WSA is due until Q1/2014. I guess AMD will sit together with GF again and discuss the situation. The outcome is totally open in my opinion. Or do you have a good guess?

WSA is valid through 2024. What is valid through Q114 is the *size* of the purchase commitment.

I don't have a good guess. If AMD is to move back to the older purchase commitment but has only 2013 volumes, they are bound to pay some over 500 million in take-or-pay charges every year.

 

[SocketF]

Given that nobody in the 1st tier chose to go GLF 28nm FDSOI over TSMC 20nm, quite the opposite, everybody is lining up for 20nm, I guess we have an empirical evidence of what those numbers would show.

Well yes, but nobody of these company has a WSA with GF, i.e. is forced to buy from GF. Even if GF would be less expensive, everybody will go to TSMC, given GF's 28nm history. AMD would probably do the same .. if they could ...

WSA is valid through 2024. What is valid through Q114 is the *size* of the purchase commitment.

Yes sure, forgot to write "3rd amendment", thanks ;-)