Abu Dhabi Fund Said to Be Exploring Sale of Globalfoundries

[Lonyo]

So they gave up on the crappy foundry that is the joke of the industry. Not surprised.

I remember when some people thought this would be nothing but a sugar daddy with endless money.

Yeah and they are also "giving up" on a load of other investments.
This isn't due to GF performance but due to oil price performance. No point trying to spin it as anything else.

Plus GF is a capital intensive business that needs cash input to grow, which is the opposite of what Abu Dhabi can afford. It is basically the worst fitting business to what they have currently, which is a cash problem.
They aren't getting in money from oil revenues due to the low oil prices. They aren't getting profits from GF and it needs capital expenditure to develop, which they can't afford. Therefore even if it was performing better they would probably get rid of it because it's the opposite of what they need right now. Right now they need businesses which are generating free cash flows to prop up the government spending.

 

[VirtualLarry]

Is it possible that GF will go out of business, if they can't find a buyer, before they get a chance to mass-produce Zen for volume?

 

[DrMrLordX]

Who uses 22 SOI again?

IBM, for POWER8. So 22nm SOI actually has a market presence, unlike the non-existent 28nm SOI and 20nm processes of GF.

Yeah and they are also "giving up" on a load of other investments.
This isn't due to GF performance but due to oil price performance. No point trying to spin it as anything else.

Plus GF is a capital intensive business that needs cash input to grow, which is the opposite of what Abu Dhabi can afford. It is basically the worst fitting business to what they have currently, which is a cash problem.
They aren't getting in money from oil revenues due to the low oil prices. They aren't getting profits from GF and it needs capital expenditure to develop, which they can't afford. Therefore even if it was performing better they would probably get rid of it because it's the opposite of what they need right now. Right now they need businesses which are generating free cash flows to prop up the government spending.

I was thinking the same thing. It would be interesting to see what else Mubadala has put on the market.

Is it possible that GF will go out of business, if they can't find a buyer, before they get a chance to mass-produce Zen for volume?

Not likely. AMD has that upcoming $1 billion contract (NX?) and you know GF will be fabbing all those chips. They have too much in the pipe to just close up shop.

 

[dark zero]

Not likely. AMD has that upcoming $1 billion contract (NX?) and you know GF will be fabbing all those chips. They have too much in the pipe to just close up shop.

In the worst case, AMD would be taking care of the whole fab!.... that would be crazy to watch....

If AMD did an excavator FX on 22 nm SOI....only to gain time and maintain market share....

 

[jpiniero]

Not likely. AMD has that upcoming $1 billion contract (NX?) and you know GF will be fabbing all those chips. They have too much in the pipe to just close up shop.

I'd have to think that the NX will be fabbed at TSMC. I'm sure that without the WSA AMD would run as far away from GloFo as possible.

 

[ShintaiDK]

Yeah and they are also "giving up" on a load of other investments.
This isn't due to GF performance but due to oil price performance. No point trying to spin it as anything else.

Plus GF is a capital intensive business that needs cash input to grow, which is the opposite of what Abu Dhabi can afford. It is basically the worst fitting business to what they have currently, which is a cash problem.
They aren't getting in money from oil revenues due to the low oil prices. They aren't getting profits from GF and it needs capital expenditure to develop, which they can't afford. Therefore even if it was performing better they would probably get rid of it because it's the opposite of what they need right now. Right now they need businesses which are generating free cash flows to prop up the government spending.

If GloFo could pull its weight they would keep it. Because then it would be one of the investments that would keep Abu Dhabi running in the future. Because oil wont rebound.

They sell it because it cant.

 

[krumme]

Who want to build their future console business up 8 years from now on the foundation of abu dhabi and whatever the political mood and fluctuating oilprices and the mubadala ownership ?
Its not transparent business environment.
Add the wsa on top of it. And its the definition of not transparent.
Thats something you likely want to avoid as its objective risk you can better control by using eg ss or tsmc.
Risk = money.

 

[DrMrLordX]

I'd have to think that the NX will be fabbed at TSMC. I'm sure that without the WSA AMD would run as far away from GloFo as possible.

Does TSMC have the capacity for that? And where were the PS4/Xbone APUs fabbed?

 

[Arachnotronic]

Does TSMC have the capacity for that? And where were the PS4/Xbone APUs fabbed?

Chip manufacturing decisions are often made years in advance, giving the foundries time to put capacity in place to satisfy expected demand.

If AMD has won the next gen console contracts and plans to do the chips at TSMC, TSMC probably already knows and will make sure the capacity is there.

 

[dark zero]

Does TSMC have the capacity for that? And where were the PS4/Xbone APUs fabbed?

TSMC is extremely well known for their hiatuses....

 

[jpiniero]

Does TSMC have the capacity for that?

The NX is using 28 nm, so capacity shouldn't be a problem.

 

[mrmt]

Is it possible that GF will go out of business, if they can't find a buyer, before they get a chance to mass-produce Zen for volume?

Zen should not be impacted by any Globalfoundries sale because almost all if not every money allocated to 14nm is already under contract, and AMD allocation is way small even for Globalfondries standards.

But whatever comes after 14nm is under threat now.

 

[mrmt]

Yeah and they are also "giving up" on a load of other investments.
This isn't due to GF performance but due to oil price performance. No point trying to spin it as anything else.

If anyone is spinning something it's you, this is what we have as facts:

The subpar foundry is turning out to be a subpar business and is being jettisoned by the mother investment fund. It would still be a subpar business regardless of the oil price, with the difference that if oil prices were higher Mubadala could give a couple of punches more on this bloody knife to see if it would yield a different result, but with the current scenario Globalfoundries will be either sold or at least have its investment budget cut to the bone.

 

[DrMrLordX]

The NX is using 28 nm, so capacity shouldn't be a problem.

If it's 28nm, then GF would be the go-to company for that. Their 28nm process is actually quite mature. I'm not sure why they would even bother with a different foundry, since they could use those chips to help deal with the wafers they get as a part of the WSA.

That would also be consistent with what AMD has done most-recently with the PS4 and Xbone APUs. Initially, they had the chips manufactured at TSMC. Apparently they switched foundries to GF for those APUs about two years ago.

Assuming the NX deal is already set in stone, one would assume they've gone full-bore for GF. TSMC would only be considered seriously if GF just vanished in a puff of smoke, forcing them to deal with a secondary supplier.

Had they planned to produce the NX chip on a smaller process, then I could see them considering TSMC, especially if they negotiated all the contracts several months ago when TSMC's 16nm/16nm+ nodes were proven while GF's 14nm LPP was not.

 

[mrmt]

That should also mean that Mubadala will dump their 20% stake on AMD.

 

[NTMBK]

Worst case, GloFo's owner will end all R&D and just keep running the existing 14nm-and-above process nodes, making some revenue on that long tail. There's a difference between exiting an industry gracefully, and immediately ceasing all production.

 

[MiddleOfTheRoad]

Worst case, GloFo's owner will end all R&D and just keep running the existing 14nm-and-above process nodes, making some revenue on that long tail. There's a difference between exiting an industry gracefully, and immediately ceasing all production.

That would be my guess, too. I don't see GloFlo completely shutting down. But, I can see them ceasing to move forward with any additional investment in future nodes. If they milk the existing fabs with no new investment -- they would likely be pretty profitable as long as they can find enough buyers for the old tech.

 

[ShintaiDK]

Old nodes are always in demand due to initial cost. Off the hip guess, I would say ~half of all TSMCs production is still 40nm or higher. Maybe even 65nm and higher.

 

[Arachnotronic]

Old nodes are always in demand due to initial cost. Off the hip guess, I would say ~half of all TSMCs production is still 40nm or higher. Maybe even 65nm and higher.

Not a bad guess at all...

 

[mrmt]

Worst case, GloFo's owner will end all R&D and just keep running the existing 14nm-and-above process nodes, making some revenue on that long tail. There's a difference between exiting an industry gracefully, and immediately ceasing all production.

Why would you go for the cheapskate copy when you can get the real deal with Samsung, or get an even better node with TSMC?

 

[MiddleOfTheRoad]

Why would you go for the cheapskate copy when you can get the real deal with Samsung, or get an even better node with TSMC?

Well, trying not to be Captain Obvious.... But cheapskate copies generally sell at a cheapskate price. Just saying.

 

[Arachnotronic]

Well, trying not to be Captain Obvious.... But cheapskate copies generally sell at a cheapskate price. Just saying.

How's that strategy working out for AMD?

 

[mrmt]

Well, trying not to be Captain Obvious.... But cheapskate copies generally sell at a cheapskate price. Just saying.

Foundries are usually a multi-year relationship, and going with a cheapskate copy of a dead-end foundry isn't the most wise answer to a problem if you are just a bit forward thinking.

I'm not of the opinion that Globalfoundries will fold either, I think their most probable destiny is to become a second-tier foundry, competing against SMIC and UMC for whatever lagging edge customers they can get. IBM R&D teams should not be able to compete with TSMC or Intel, but they certainly can with UMC, plus the CAPEX profile would be much softer than on the bleeding edge and they already have plenty of capacity installed. All they need to do is stop chasing the bleeding edge and go for the lagging edge full steam.

The only issue I see for this is that IBM probably tied them to develop bleeding edge nodes for their POWER processors, but they can just free up IBM to manufacture the chips wherever they want.

 

[NTMBK]

Why would you go for the cheapskate copy when you can get the real deal with Samsung, or get an even better node with TSMC?

If you're tied into a multi-year WSA?

 

[MiddleOfTheRoad]

How's that strategy working out for AMD?

Judging from the recent boost on their stock price, apparently not too badly.

I also suspect they grabbed some additional market share in dedicated GPU's this quarter -- the pricing on Radeons was definitely more aggressive during Black Friday / Cyber Monday than for Nvidia's offerings.