PLEASE SIT DOWN if you are not already, but the long long long telegraphed patent license spat between Intel and AMD over Global Foundries is edging closer to a hot war. Intel sent AMD a letter saying it is breaching the agreement, and AMD is shooting back.
Word of this tiff first leaked out when AMD filed an 8-K form this morning saying that Intel thinks the Global Foundries deal breaches the 2001 patent cross-license agreement between the two companies. It gives AMD 60 days to rectify this breach or its rights under the agreement would be terminated.
This does not mean that Intel's rights would be terminated, however. For that, Intel would need to breach its side of the agreement. That is why AMD sent Intel a 'please get stuffed yourself' letter basically accusing Intel of interfering with AMD's business. This gives Intel 60 days to rectify that problem or its rights get terminated.
According to the agreement, the two sides had to meet, and that was done. The next step is to send letters, which as you can guess, is now done. Mediation follows, and if the two sides don't see eye to eye, then the licenses terminate... sort of.
From there, it will get ugly, and end up with lawyers in court. AMD says that it is in full compliance with the agreement, and Intel obviously disagrees. Given the glacial speed of the courts, this looks to be settled long after the sun dies a heat death somewhere around the year 5 billion.
Ironically, that is after the AMD antitrust trial is set to go forward in early 2010. The next 12 months or so look to be very lucrative for AMD and Intel legal staff, outside counsel, and local pizza joints that deliver late at night.
One interesting side note is what the breaches are. Intel says there are two breaches, one is in the definition of 'subsidiary'. Basically, Intel alleges that AMD does not have enough ownership for Global Foundries to be a real subsidiary, and that means AMD is fabbing too many wafers 'externally'.
That particular issue should be pretty cut and dried, there are legal definitions of such terms. It would shock us greatly if AMD didn't very carefully tap dance around those definitions when it did the Global Foundries deal. We don't see this as going very far, but it makes good headlines.
The second issue is a bit more touchy. The agreement which can be found here has a bunch of things redacted. Intel alleges that AMD is in violation of one of the redacted parts, but neither side can comment on what that bit is, or why AMD is not living up to it. Supposedly. Lets hear it for censorship, three cheers, one, ****, three!
Intel is saying that if AMD says it is in the clear, why not make the document public? On the phone this morning, AMD sort of agreed to do so, provided that Intel makes public the antitrust documents.
When asked about this, Intel didn't seem to mind as long as AMD opened up its side as well. Since there isn't a big company out there that doesn't have snotloads of embarrassing emails stuck on compliance-mandated WORM media, this should be fun to watch.
In the end, we are at step two of a four or five-step process, and that doesn't count appeals, counter-potshots, and general legal manoeuvering. This is unlikely to be settled for years, and will drag on and on and on, with only mild flashes of interest until then.
For now, we have legal posturing, position staking, and general FUDding of the press and markets. With any luck, both sides will agree to open the agreement, and then open the antitrust documents, then things will be fun.
How about it guys? µ
This will be very interesting to read. :)
Bring it on Charlie, keep on digging up the secrets these companies don't want to have coming out!
When Charlied first hinted at the "global-foundries" spin off waaayyy back in late 2007/early 2008, I said the idea was crazy, citing the cross-license agreement. AMD is REQUIRED to build a fixed percentage of its x86 processors at its own Fabs. This was to prevent AMD from out-sourcing and getting a leg-up on Intel. This is precisely why AMD hadn't out-sourced very much in the past. By essentially selling a controlling interest in their Fabs, AMD has essentially violated this basic agreement. I really don't see how AMD can get around that particular problem.
AMD is not required to build a fixed percentage of its x86 processors, it's a fixed out-sourcing percentage, so as long as AMD has xuyqa 50.1% stake in GlobalFoundries theres nothing Intel can say about it.
Intel is just totally scared shitless of what GlobalFoundries and their arabian investors money can potentially do to their monopoly.
Intel isn't likely after AMD's skin. It needs the competition considering any anti-trust issues. It is more likely that they want to make sure that NVidia, or anyone else for that matter, doesn't gain access to any X86 material.
If Intel would want to kill AMD, there would be far easier ways to do so.
AMD has a 34-36% equity stake in Global foundries (which would be what most consider a MINORITY interest). They have an equal VOTING stake - this is where it gets interesting - does an equal voting stake represent a controlling interest? (Or do you need slightly more than 50%?)
What is also confusing, is what analysts and the press have overlooked - AMD has publicly stated (in the last earnings call) that they intend to divest themselves further of Global Foundries over time... which means clearly it will not be a "subsidiary" (even if you can call it one now) as time goes by. That means AMD is banking on removal of the outsourcing restriction when the license is renegotiated in 2010 or will later be in breach of the licensing agreement. That is a dangerous game/gamble to be playing. And frankly I still see a licensing/lawsuit settlement coming.
Regardless of the technicalities, is there any reasonable way to think that a 34% stake is a controlling interest (when there is only 1 other partner with a 66% stake), clearly they are violating the spirit of the agreement. And if the roles were reversed and it was Intel weaseling by on a technicality, I wonder if the reaction of commenters and the press be the same?
At least with Nvidia articles people made funny comments. This article is horribly boring - it only attracts posturing comments from AMtel PR bunnies. How gawd awfully boring.
That and they trip over their own words trying to make a "point" At least with Nvidia stuff I could laugh and watch the flamewars. This stuff is too awfully dry. I think I need to go cut myself with a left over circuit board.... Slowly.
Charlie:
have you read the license agreement, before you opined?
I do believe AMD invented the 64bit Instruction and wasn't the Athlon and FX CPU's the founder of 64bit Processing. If I also remember INTEL took that and called it INTEL64 when clearly it's AMD64 if this gets out of hand why not take away there 64bit Instruction license.
Blue Chip/Designers for X86:
List of former IA32 compatible microprocessor vendors:
* Centaur Technology - originally subsidiary of IDT, later acquired by VIA Technologies, still producing compatible low-end devices for Via
* Chromatic Research - mediaprocessor with x86 instruction set compatibility never completed
* Cyrix - acquired by National Semiconductor, later acquired by VIA Technologies, eventually shut down
* Chips and Technologies - left market after failed 386 compatible chip failed to boot the Windows operating system
* Exponential Technology - BiCMOS compatible microprocessor never completed
* IBM - Cyrix licensee and developer of Blue Lightning 486 line of processors, eventually left compatible chip market
* IDT - original funder of Centaur, later sold off that subsidiary to VIA, no longer in compatible market.
* IIT Corp - 486 compatible project never completed
* Harris Corporation - sold rad-hard versions of 8086 and 80286; product line discontinued.
* MemoryLogix - Multi-threaded CPU core and SOC for PCs never completed
* Metaflow Technologies - project never completed
* Montalvo Systems - Asymmetric multiprocessor never completed
* National Semiconductor - low-end 486 (designed in-house) never widely sold; first acquirer of Cyrix, later keeping only low-end IA32 devices targeted for consumer System-on-a-chips, finally selling them to AMD
* NEC - sold early Intel architecture compatibles such as NEC V20 and NEC V30; product line transitioned to NEC internal architecture
* NexGen - bought by AMD to help develop the successful K6 device
* Rise Technology - after 5 years of working on the slow mP6 chip (released in 1998), the company closed a year later
* Texas Instruments and SGS-Thomson -licensees of Cyrix designs, eventually left compatible chip market
* Transmeta - transitioned to an intellectual property company in 2005.
* United Microelectronics Corporation and Meridian Semiconductor - got out of market after slow 486 compatible missed market window
* ULSI System Technology - never completed x86 SOC; company shut down after one of their employees was convicted for stealing Intel Floating-point x87 design documents
More info over the wiki:
x86-processors for both regular PCs and embedded designs:
* Intel
* AMD
* VIA
* Transmeta (discontinued its x86 line)
* Rise Technology (acquired by SiS)
* IDT (Centaur Technology x86 division acquired by VIA)
* National Semiconductor (sold the x86 PC designs to VIA and later the x86 embedded designs to AMD)
* Cyrix (acquired by National Semiconductor)
* NexGen (acquired by AMD)
* Chips and Technologies (acquired by Intel)
* IBM (discontinued its own x86 line)
* UMC (discontinued its x86 line)
* NEC (discontinued its x86 line)
[edit] x86-processors for embedded designs only:
* Zet IA-32 (Zet is a GPL open source FPGA implementation targeting the Xilinx ML-403)
* ZF Micro (ZFx86 - Cx486DX SoC)
* Nvidia (M6117C - 386SX)
* ALi (x86 products went to Nvidia through the ULi sale)
* SiS (discontinued its Vortex86 line)
* DMP Electronics (Vortex86SX and Vortex86DX, compatible with Intel 486SX and 486DX respectively)
* RDC Semiconducters (R8610 an R8620)
[edit] Manufacturing-only of x86-processors designed by others:
* IBM (manufactures processors for ZF Micro and VIA; discontinued production for NexGen and Transmeta)
* TSMC (manufactures processors for VIA; discontinued production for Transmeta)
* UMC (manufactures processors for Nvidia; discontinued production for Rise, SiS, ALi and ULi)
* Fujitsu (manufactured processors for Transmeta; discontinued x86 production)
* National Semiconductor (manufactured processors for ZF Micro; discontinued x86 production)
[edit] Manufacturing and selling under its own name of x86-processors designed by others:
* IBM (designs by Cyrix; now this line is discontinued)
* SGS-Thomson (designs by Cyrix; discontinued x86 production)
* Texas Instruments (designs by Cyrix; discontinued x86 production)
In my personal opinion if intel wants "just" to protect its IP, they shouldn't go on a all out attack on this.
Of course none of us has access to the fineprint.
from http://www.extremetech.com/article2/0,2845,2343161,00.asp
"AMD owns just a 44.4 percent stake in GlobalFoundries, with the majority owned by the Advanced Technology Investment Co., a fund owned by the government of Abu Dhabi. "
Intel is just talking smack to distract from the fact that world courts are close to convicted Intel for massive violation of anti-trust laws by preventing mobo and system makers from using AMD products. Intel is gonna pay big this time, despite their smoke screen on the cross licensing.
AMD doesn't own 50.1% shares in FoundryCo. I don't remember the figure, but it's less than 50%. Clearly, AMD can't be regarded as the owner of FoundryCo.
@Bobby: Is it wrong that AMD is breaching the license agreement and Intel is trying to prevent them from doing so? If they were going to outsource production, AMD should have thought about it while signing the agreement.
@Dave C: Intel isn't breaching its part of the agreement, so AMD can't take away the x86-64 instructions.
We need less manoeuer and greater stem cell research.
ssj4Goget: “Is it wrong that AMD is breaching the license agreement and Intel is trying to prevent them from doing so?”
Why should you need a “licence” to give Intel some much-deserved competition?
AMD currently has a 34% equity stake in Globalfoundries, ATIC has the remaining 66%... I'll let folks decide who "owns" GF and who is pretty much an investor. The voting rights are 50/50 despite the dramatic difference in equity stakes (I'll let folks decide if this is merely to get around the 50% voting clause in the license agreement or if this is reasonable, given the relative equity stakes).
Licenses are needed because those who do not do the invention or development, should not be able to ride the coattails of those who do. In AMD's case I believe the license is ROYALTY FREE (meaning it doesn't cost them anything), but just because they have access to the x86 IP doesn't mean they can turn around and give it to anyone. Intel is in the same boat here, they can not turnaround and spinoff (ummm, I mean "enter a joint venture with " wink, wink) a company with the AMD IP that they have access in the agreement either. If you just give licenses for the sake of competition then why wouldn't a companies just wait for someone to do all the heavy lifting and freeload when the technology/product is proven?
Finally the real issue is what happens as ATIC continues to investment money into GF? AMD will have to match these investments proportionally to maintain their 34% or they will dip below even the "technical" requirements. AMD has said in the last earnings call that there intention was to not do this and to decrease their stake over time. So even if they are meeting the bare minimum technical requirements of it not being a subsidiary, it is their PUBLICLY stated intent that it will become one!
random thoughts: “Licenses are needed because those who do not do the invention or development, should not be able to ride the coattails of those who do.”
The little flaw in your reasoning is the fact that a patent only applies to the idea, not to the _real_ hard work, which is the implementation and the execution. Ideas are a dime a dozen. So why should a company be allowed to stifle competition just by claiming a patent on an idea?
AMD is outsourcing his production to GlobalFoundries.
So what ?
What's the difference with this : http://tinyurl.com/c9m8ep