AMD Likely Skipping 10nm For 7nm, Restructures GlobalFoundries Agreement

[Guest]

AMD restructured its Wafer Supply Agreement with GlobalFoundries, which indicates that, unlike Intel, it will likely skip 10nm in favor of 7nm.

AMD Likely Skipping 10nm For 7nm, Restructures GlobalFoundries Agreement : Read more

 

[M_Sketch]

Sounds good for amd, can't wait for any type of CPU competition, especially amd, haven't seen anything new for a while.

 

[Guest]

It is all just talk, they need to release something realy.

 

[RomeoReject]

I recognize this is a computer site, not a business one, but isn't offering 20% of your company to someone you're contracting work to somewhat odd? Global Foundries is basically the employee here, yet AMD is saying "But you can also buy 20% of the value, which is almost assuredly tied in to your own work."

 

[Jeffrey_32]

GF already owns 18% of amd. To take advantage of the new shares they will have to sell of existing shares first. Not really and "employee" relationship.

 

[viewtyjoe]

The stock transfer deal makes more sense when you consider that GlobalFoundries used to be AMD's semiconductor fab business, and it's a way for AMD and GF to signal they are returning to a closer relationship than they have had previously. It also incentivizes GF doing well for AMD, because if Mubadala's shares increase in value due to AMD's stock rising, it is effectively free value to them.

 

[grsychckn]

Sorry for the previous comment. @Romeoreject: AMD is not offering 20% of their stock, rather <10% (75m shares of ~795m). Add to it that this is a warrant to exercise 75m options at the execution price of $5.98 AND that they cannot exceed 20% ownership of AMD on the whole. To me, the price of the options set at $5.98 means that AMD and GloFo believe the target price to be above $5.98 or GloFo won't exercise the options. Also, I've read unofficially that GloFo already owns something like 17% of AMD anyway so they could only exercise ~20m options and would have to immediately sell for profit in order to both stay below 20% ownership and have the ability to exercise more of these options.

 

[salgado18]

I recognize this is a computer site, not a business one, but isn't offering 20% of your company to someone you're contracting work to somewhat odd? Global Foundries is basically the employee here, yet AMD is saying "But you can also buy 20% of the value, which is almost assuredly tied in to your own work."

I believe they did this deal to get a lower cash proposal. After all, they will really need this competitive edge in the near future.

 

[jasonelmore]

Yes it indeed does sound good for AMD, but Global Foundries has got to get their thermal problems under control.. 14nm should not be running hotter (or using more power) than 16nm TSMC, but it is.

 

[mavikt]

I'll give them an A for effort; Having something 7nm ready to compete when Intel releases Cannonlake (as per the editors speculation)

 

[Martell1977]

This should be very interesting. AMD has historically always been behind Intel from node to node by varying degrees. This might give them a small jump on a smaller node, plus Zen is being teased as a HUGE improvement over AMD's last generation.

I really hope this lights a fire under Intel, they get complacent with no real competition.

 

[antilycus]

seems to be a good business decision. GF will want to produce a good product for AMD to leverage the market and gain back market share / demand. It's win / win for everyone but Intel.

 

[bit_user]

The agreement does allow AMD to source wafers from other entities, but it also ties AMD into wafer purchase targets, which means that AMD has an obligation to order a fixed set of wafers every year or face levies. AMD is also required to pay an offset fee for any wafers it procures from other fabs.

Ouch. Tying yourself to such a (historically) under-performing supplier... I hope there's an escape clause for when GF is late on their process enhancements and/or they miss power targets.

 

[Karadjgne]

Wonder who'll be the first to reach 1nm. I'd not be all that surprised if AMD is successful at 7nm, if Intel doesn't follow suit and aim the Gens after Cannon lake for a 5nm. Would actually be a good sized bump after this piddling 5-10% bumps since Sandy-Bridge.

 

[joytech22]

Great to see AMD finally putting the pieces together.

I know it sounds weird, but I feel like Intel has been waiting for this moment and avoided a monopoly position on purpose.

Without AMD, we lose one of only three x86 licensed manufacturers (And VIA has no high performance offering) and would lose Nvidia's only real GPU competitor.

AMD is also good guy software, new and open API's that all teams can benefit from - Nvidia/Intel have kept those doors shut.

 

[redgarl]

If AMD can release a broadwell cpu for 350$, guess who will get back in the race.

 

[srmojuze]

I recognize this is a computer site, not a business one, but isn't offering 20% of your company to someone you're contracting work to somewhat odd? Global Foundries is basically the employee here, yet AMD is saying "But you can also buy 20% of the value, which is almost assuredly tied in to your own work."

I think AMD will go down in history as "the tech company that could". Every move seems strange and counter-intuitive. From purchasing ATI then IMO mangling that, creating great chips for quite a while then (to me) messing everything up with this Global Foundries stuff, and selling off Imageon (Adreno) just at the time when it would become one of the biggest players in a multiple-billion-person market!

 

[none12345]

"Wonder who'll be the first to reach 1nm."

Its unlikely that we will ever see 1nm. You are at 1 or 2 atoms wide at 1nm for silicon. Of course there will probably be something called a *1 nm process*, but it will be even a more made up number then it already is at that point.

 

[Axonn]

For the sake of competition, I sincerely hope AMD manages to reach 7nm at least in the same time with Intel, but I think that's wishful thinking on our part. Yes, Intel is spending on 10nm, but that doesn't mean they are not spending on 7nm either. Don't forget how much money Intel has. Most probably, Intel will reach 7nm at least 6 months ahead of everybody while in the same time absolutely destroying AMD's 14 nm primitive tech with their 10nm one. Not to mention that AMD's 14 nm tech is about 18 months behind Intel. Like somebody from Intel said: the fact that they're also on 14nm doesn't mean we're at the same level. Intel's 14nm is highly refined. I may sound like an Intel fanboy here and trust me, I'm not. This is an engineer speaking.

 

[Sam Bittermann]

Sounds like Glo Fo found another way to screw over AMD long term again...they could barely get off of 28nm.

 

[Beholder88]

Since AMD has been able to hold on long enough, it only makes sense that they end up on the same process node as Intel. Getting down to the 7nm level is hard, and who knows how much further down they will be able to go without some new technology being invented/perfected. It's going to come down to who has the better architecture design and who can make the best improvements and bring them to market faster. It's going to be a competition between the design and development departments at AMD and Intel, moreso than what it has been before.

 

[alidan]

Wonder who'll be the first to reach 1nm. I'd not be all that surprised if AMD is successful at 7nm, if Intel doesn't follow suit and aim the Gens after Cannon lake for a 5nm. Would actually be a good sized bump after this piddling 5-10% bumps since Sandy-Bridge.

If i remember right 6nm is a hard limit, as beyond that point and its a massive quantum tunneling issue, and even then, its apparently a massive issue even before then, it just gets to great to overcome passed 7nm.

with that said, this is on silicon, what is likely to happen after silicon is a move to something bigger but takes higher clocks better, i think the front runner use to be graphene that could get 50ghz on air at 350nm, im not looking into this passed a cursory search, but i remember this being a thing a long time ago, though its still all in research phase as far as im aware.

 

[Daekar3]

Glad to see some competition for Intel. My computers have been Intel/Nvidia for a while, would be nice to see a shakeup. I wonder what process node the next-gen consoles will be on? Those are surely still going to be AMD.

 

[bit_user]

I wonder what process node the next-gen consoles will be on? Those are surely still going to be AMD.

I don't know about PS4 Neo, but anything after that will probably be 14 nm or 16 nm. Nothing smaller, for sure.

Console APUs are big and cost-constrained. That means they'll never be on a bleeding-edge process node.