SED - Samsung M&A AMD research news

[NTMBK]

If all they want is people they just need to open an office near AMD HQ and start hiring. They'll have no trouble sucking up plenty of talent. Much cheaper and easier then buying a company.

Oh definitely true, and this is precisely what they did with their ARM CPU design team; open an office nearby and start poaching. I was just pointing out that if they also wanted to purchase the company for other reasons (patent portfolio, graphics IP, server ARM CPU design) then they would probably want to keep the design teams.

Not sure that they would want to buy a company so tightly tied to GloFo, though. I can see some sort of nice and cosy semicustom design agreement where AMD designs chips for Samsung, who then fabs them internally, but a full blown takeover/merger seems unlikely.

 

[witeken]

^^ This.

Only reason to buy AMD is if you want their IP and/or products. And lets be honest, its not for their products.

If Samsung thinks it can and wants to compete with Intel in PC and server, which isn't a strange idea considering the cash it can make, then why not buy AMD?

 

[Tuna-Fish]

If all they want is people they just need to open an office near AMD HQ and start hiring. They'll have no trouble sucking up plenty of talent. Much cheaper and easier then buying a company.

Note that they already did this to the team that designed Bobcat. Ever wondered why AMD's plans for a bobcat successor stalled so badly? It was designed by a small team in India and after Bobcat was released (and note that even though it was not that great a success in the market, Bobcat was, for the resources spent on it, a seriously impressive core), the entire design team was poached whole by Samsung.

 

[TurtleCrusher]

A Bobcat core on Samsung's upcoming 14nm process would shake up the mobile market completely. That and dGPU/APU products alike would gain massive marketshare.

AMD has been stopped right in it's tracks due to being on inferior processes. To have the backing of Samsung to inject R&D cash would be almost a slam dunk back to high end/mobile relevancy. Who wouldn't want more competition to bring us better products?

I love seeing these posts. You can really see who is an investor rather than an enthusiast or consumer. It seems like they spend more time talking smack about the competition than using their own products.

 

[frozentundra123456]

If Samsung thinks it can and wants to compete with Intel in PC and server, which isn't a strange idea considering the cash it can make, then why not buy AMD?

Yea, AMD has certainly demonstrated great expertise in the server market lately. Takes real skill to go from 20 or 30 percent of the market to less than 5%.

 

[TurtleCrusher]

Yea, AMD has certainly demonstrated great expertise in the server market lately. Takes real skill to go from 20 or 30 percent of the market to less than 5%.

That's absolutely due to the process disadvantage. You know that.

 

[sandorski]

With enough $R&D AMD could turn out some much better stuff for sure. However, I'm not sure where they would focus that $R&D on, especially if Samsung had a say in the matter. The company itself could be saved and even become very Profitable, but at the same time the Desktop CPU segment may not gain much from it. Hard to say though, but it sure would be nice to have AMD competing Tech v Tech rather than just Price in this segment.

 

[ShintaiDK]

That's absolutely due to the process disadvantage. You know that.

No. The design isnt good enough either. Just look at 32nm SB.

 

[TurtleCrusher]

No. The design isnt good enough either. Just look at 32nm SB.

Do you even know what a 22nm 8-core piledriversteamroller/excavator would've been for the server market? It would've been game-over on the monopoly. It is absolutely a process node advantage. Excavator increased density massively in Carrizo and at 22nm would give Intel a run for their money if they had a comparable node to put it on. This is where Samsung would come in.

Unless you can spout actual facts instead of regurgitating 10 word nonsense posts, you're not fooling anyone.

 

[ShintaiDK]

Do you even know what a 22nm 8-core piledriversteamroller/excavator would've been for the server market? It would've been game-over on the monopoly. It is absolutely a process node advantage. Excavator increased density massively in Carrizo and at 22nm would give Intel a run for their money if they had a comparable node to put it on. This is where Samsung would come in.

Unless you can spout actual facts instead of regurgitating 10 word nonsense posts, you're not fooling anyone.

Why do you think it would be any different than 32nm vs 32nm. Do you think AMD somehow scales better than Intel?

Carrizo only works in mobile, thats how good it is.

The only ones fooling themselves is those seeking some kind of Sugar Uncle for AMD that will save everything for their favourite company without penalty.

 

[Fjodor2001]

I guess AMD could use an R&D budget increase:

Quite impressive what AMD has been able to deliver after all given their R&D budget constraints. Backed with money from Samsung's huge treasure chest it could turn out very nice going forward...

 

[Erenhardt]

(tuned) Mobile apus will soon be a go to cluster parts. So 'only works in mobile' is not a bad feat.

 

[ShintaiDK]

Future products will be even more R&D starved than those uncompetitive products today due to the R&D timeframe. Its a bottomless pit.

 

[ShintaiDK]

(tuned) Mobile apus will soon be a go to cluster parts. So 'only works in mobile' is not a bad feat.

Like the Berlin APU? Oh wait..that one didnt get released because there was no buyers.

 

[Erenhardt]

Like the Berlin APU? Oh wait..that one didnt get released because there was no buyers.

Lets see what Facebook orders. It won't be an apu, right?

 

[TurtleCrusher]

Why do you think it would be any different than 32nm vs 32nm. Do you think AMD somehow scales better than Intel?

Carrizo only works in mobile, thats how good it is.

The only ones fooling themselves is those seeking some kind of Sugar Uncle for AMD that will save everything for their favourite company without penalty.

Because of the evolving BD/PD/SR/EX cores, notably the excavator design. The CPU cores were able to be shrunk by 23% and a 40% drop in power consumption compared to steamroller cores. Let that be on 22nm rather than 28 and the market does a 90 degree shift overnight, maybe further if AMD could force OEMs hand for once.

Don't be surprised to see a few major players try to snatch up AMD.

 

[ShintaiDK]

Lets see what Facebook orders. It won't be an apu, right?

Its still a rumour isnt it. And the rumour specifies its based on ARM (ARMv8-A, same as the X-Gene). Being designed by the semicustom/embdedded division.

 

[artivix]

Unless Intel's lawyers screwed up, I doubt that would matter. A standard change of control provision would cover any form of merger, regardless of who "buys" who, or which entity survives.

When you see the tiny company "buy" the big company, it's usually for a tax reason (sometimes it saves big money!) or maybe a regulatory reason (regulators usually don't have change of control provisions).

The FTC and Intel have a consent agreement that seems in fact to encourage AMD to consider a merger with a larger company and allow greater competition -- see the bold paragraph... Other government entities around the globe will probably have even greater regulatory oversight of transfter licensing terms between AMD and Intel.

Justia Law

Section III.B of the Proposed Consent Order would require Intel to offer to modify the change of control terms in Intel's intellectual property licenses with AMD, NVIDIA, and Via. The Commission is concerned that Intel's past conduct has weakened AMD and Via - Intel's only x86 competitors.

This provision seeks to ensure that these existing competitors can partner with third parties to create a more formidable competitor to Intel.

The existing change of control terms in licensing agreements potentially limit the ability of AMD, NVIDIA, and Via to take part in a merger or joint venture, or to raise capital. The provisions in the Proposed Consent Order are designed to allow AMD, NVIDIA, and Via to enter into a merger or joint venture with a third party, or to otherwise raise capital, without exposing itself to an immediate patent infringement suit by Intel. In the event that AMD, NVIDIA, or Via undergo a change of control, these provisions prohibit Intel from suing for patent infringement for 30 days. Furthermore, Intel must offer a one-year standstill agreement during which the acquiring party and Intel would not sue each other for patent infringement while both parties enter into good faith negotiations over a new license agreement.

The Commission takes seriously Intel's commitment under these provisions in the Proposed Consent Order. The Commission has authority under the Order to evaluate and determine whether Intel in fact engages in good faith negotiations and the Commission will be able to enforce the Proposed Consent Order if Intel does not negotiate in good faith.

In the event the change of control terms are invoked, the Commission will carefully scrutinize Intel's conduct and take action, if appropriate.

http://regulations.justia.com/regulations/fedreg/2010/08/10/2010-19694.html

 

[artivix]

^^ This.

Only reason to buy AMD is if you want their IP and/or products. And lets be honest, its not for their products.

You seem to easily smear companies that developed the X64 consumer and server CPUs and faster graphics memory technologies that brought prices down while boosting performance up for everyone.

Look in the other thread about what Sony achieved with the APU developed with AMD where the sales of PS4s hit a record 20 million units sold to consumers in record time...

The robust home console sales helped Sony's struggling electronics empire to record about $21 billion in revenue during the quarter, with profit coming in at about $1.5 billion.

To not confuse you, while other Sony divisions were losing money, their SCEA console gaming were generating large revenue and profits.

AMD accomplished this without paying Sony to sell millions of their APUs, while in comparison to companies that have some type of reimbursement to OEMs that cost them billions of dollars in losses for a number of years.

 

[Phynaz]

You seem to easily smear companies that developed the X64 consumer and server CPUs and faster graphics memory technologies that brought prices down while boosting performance up for everyone.

Look in the other thread about what Sony achieved with the APU developed with AMD where the sales of PS4s hit a record 20 million units sold to consumers in record time...



To not confuse you, while other Sony divisions were losing money, their SCEA console gaming were generating large revenue and profits.

AMD accomplished this without paying Sony to sell millions of their APUs, while in comparison to companies that have some type of reimbursement to OEMs that cost them billions of dollars in losses for a number of years.

I have no idea how this is relevant to this thread, but let me point out one thing about the consoles;

Last generation IBM supplied the CPU's to every console. This generation IBM is essentially out of the chip business. Consoles aren't going to save AMD.

 

[NTMBK]

I have no idea how this is relevant to this thread, but let me point out one thing about the consoles;

Last generation IBM supplied the CPU's to every console. This generation IBM is essentially out of the chip business. Consoles aren't going to save AMD.

Very true. AMD needs to keep competitive R&D if it wants to win the next generation of consoles. They won this round because they had the right tech at the right price, but if they can't compete with NVidia tech in 5 years' time then they won't win the contract.

Honestly, if AMD wants to win the next round of consoles, it should focus on GPU tech and on deep integration with off-the-shelf ARM cores. A72 cores (or their successor) strapped to HBM memory and a Hawaii size GPU would make for a great console on ~10nm.

 

[Anon_lawyer]

The FTC and Intel have a consent agreement that seems in fact to encourage AMD to consider a merger with a larger company and allow greater competition -- see the bold paragraph... Other government entities around the globe will probably have even greater regulatory oversight of transfter licensing terms between AMD and Intel.

Justia Law

Oh, interesting. That does throw a major wrinkle in to things. I believe this is the final order: https://www.ftc.gov/sites/default/files/documents/cases/2010/08/100804inteldo_0.pdf

It doesn't seem to say that they have to reach a license agreement, just that they have to use good faith efforts to try. With the threat of FTC action if they don't, of course. Negotiating that would be an interesting deal.

 

[III-V]

^^ This.

Only reason to buy AMD is if you want their IP and/or products. And lets be honest, its not for their products.

Right. You can always poach engineering talent by waving dollars in their faces. From one perspective, it may be an even easier route than buying the company, because buying AMD brings its terrible management with it.

Still, AMD has a lot of valuable IP, and their IP alone makes them a sensible purchase.

AMD accomplished this without paying Sony to sell millions of their APUs, while in comparison to companies that have some type of reimbursement to OEMs that cost them billions of dollars in losses for a number of years.

Intel's contra revenue is apparently more difficult to grasp than quantum physics. I've not seen so much ignorance around a single subject, perhaps with the exception of climate change...

 

[mrmt]

Right. You can always poach engineering talent by waving dollars in their faces. From one perspective, it may be an even easier route than buying the company, because buying AMD brings its terrible management with it.

Still, AMD has a lot of valuable IP, and their IP alone makes them a sensible purchase.

Is it? AMD CPUs are nothing to write home about, their GPU IP doesn't fit on the mobile market, where the money is today, and they don't have software support good enough to take on Intel and Nvidia on the HPC market.

AMD GPU IP may fetch some good value on the market, but I don't think it is the game changer you think it is. If it were, given the severe cash crunch the company is facing, AMD BoD would have sold some long ago.

 

[III-V]

Is it? AMD CPUs are nothing to write home about, their GPU IP doesn't fit on the mobile market, where the money is today, and they don't have software support good enough to take on Intel and Nvidia on the HPC market.

AMD GPU IP may fetch some good value on the market, but I don't think it is the game changer you think it is. If it were, given the severe cash crunch the company is facing, AMD BoD would have sold some long ago.

IP doesn't necessarily mean products...