[WCCFT]Samsung may aquire AMD

[5150Joker]

Not for purely AIB dGPU market, in fact, it probably won't be AIB if they are selling consumer Radeons, it'll be Samsung built & labeled, premium marketing.

But the GPU HPC sector is fast growing and you bet Samsung wants a slice of that, along with x86 Servers. They already sell heaps of Samsung PCs, notebooks, servers, just with Intel solutions on the inside. AMD has shown its GPUs are equally capable in HPC as well as consumer Radeons, two birds, one-stone.

I read an interview of their late CEO a long time ago, his vision and goal is whatever market they choose to enter, they dominate it, the entire supply chain, everything made by Samsung to maximize quality control & profitability. This is why they for so long, have stuck with Exynos SOCs when its worse than Qualcomm, but they can't keep doing that, as soon as those other SOCs move to 14/16nm ff at TSMC, Exynos would be a joke in comparison.

The brilliant part of Samsung's business strategy is controlling the entire chain, very few companies have that luxury. It's why Apple, despite the lawsuits, is still beholden to them for chip development. If they did acquire AMD, I'm sure the high end of the market (servers) would be a big priority for them since that's where a lot of the dollars are concentrated. A little fact a lot of people don't know is that Samsung also makes cars so if NVIDIA is entering "deep learning" with GPUs, there's no reason AMD can't do the same and what do you know, Samsung could stick those Radeon branded "deep learning" GPUs in their own cars.

 

[lavaheadache]

I had my girlfriend read the article (she's S. Korean) and she said the gist of the article is that Samsung may just poach AMD engineers rather than an outright M&A since it would be cheaper.

Also if for some odd reason Samsung did purchase AMD along with the IP, what do you guys think the chances are that Samsung would continue the AIB market? I'd put it almost at zero. If they want AMD, it's for the lucrative high volume markets like SoC for tablets/phones/notebooks and possibly the x86 for the server market (if they can somehow retain the x86 patent). I don't think they'd care at all about the shrinking AIB market.

I think it's time I come up with some quick cash and start buying some AMD and NVIDIA stock (they just got downgraded recently).

All I got out of that was that your GF is south Korean... mmmm. =)

 

[DarkKnightDude]

wccftech.com

Yeah, I'll believe it when I see it posted somewhere else.

 

[5150Joker]

All I got out of that was that your GF is south Korean... mmmm. =)

lol that part is nice.

 

[boozzer]

hmm, so if this is true = no more high end gaming gpus from amd = nvidia monopoly = 1000$ crap nv gpus = the end of my gaming hobby. that would suck, but I think I am ok with the money I will be saving.

 

[3DVagabond]

hmm, so if this is true = no more high end gaming gpus from amd = nvidia monopoly = 1000$ crap nv gpus = the end of my gaming hobby. that would suck, but I think I am ok with the money I will be saving.

Yeah, I would have to believe that there isn't enough money in dGPU for Samsung to even care.

 

[Tsavo]

Truth is, Samsung is buying AMD for their SSD and memory business.

You heard it here, first!

 

[Silverforce11]

Yeah, I would have to believe that there isn't enough money in dGPU for Samsung to even care.

No, but combined HPC + dGPU?

 

[Edgy]

Hmm... Not sure what Korean speaking person was reading but I just read the article and the main theme is that there has been (in the past) and ongoing talk within the industry inside Korea that Samsung purchasing AMD would bring much needed expertise and value for Samsung.

The article also mentions that in the past, at least 1 Samsung executive have been known to discuss the topic that at some point, purchasing of AMD would eventually become a serious consideration within Samsung even though there have not yet been any official discussion regarding this topic among the top level Samsung executives so far.

Toward the end of the article, it does mention as a counterpoint that outright purchasing of AMD may bring too much risk and rather poaching key engineering talent from AMD would bring much reward without the risk. The article cites that Samsung has in the past hired at least 2 known talent from AMD (Michael Goddard and someone name Pat Petteler?).

Anyways - it's just article based on rumors and what ifs within Korea industry nothing solid.

 

[HurleyBird]

Yeah, I would have to believe that there isn't enough money in dGPU for Samsung to even care.

I wouldn't be so sure. Nvidia certainly does all right, No reason Samsung couldn't also. If they did continue with dGPUs we might only see first party products though, cutting the likes of Sapphire and HIS out of the loop.

 

[3DVagabond]

I wouldn't be so sure. Nvidia certainly does all right, No reason Samsung couldn't also. If they did continue with dGPUs we might only see first party products though, cutting the likes of Sapphire and HIS out of the loop.

~$4.5b total for everything, not just dGPU, compared to over $300b.

 

[Riceninja]

a samsung backed AMD with samsung's fabs, r&d budget, marketing power and efficient management/vertical integration will dominate the gpu/apu market within 3 years and compete toe to toe with intel in cpu. samsung hasn't entered an industry without becoming #1 or a very close #2 in short order.

if there's anyone sweating bullets at this rumor its nvidia and their supporters.

 

[ShintaiDK]

Assuming there is any truth to this, not that AMD havent been sold multiple times before.

People should ask themselves what Samsung needs AMD for and what parts Samsung is not interested in at all.

The only thing AMD got to sell is its IP. Now imagine what would happen to AMD, if Samsung was to buy it for that IP.

Because one thing is certain, Samsung isnt in the business of corporate welfare to keep an uncompetitive company alive by spending countless billions down a black hole. Nor are they going to invest in something with a ROI without timeframe.

If Samsung is to buy AMD with debt, contracts and excess workers. They want that money back. And that ROI would be a real long tooth even in the best case.

 

[ShintaiDK]

Because the source is actually Korean news rather than the typical fud from western tech sites made up for page hits.

http://economy.hankooki.com/lpage/industry/201503/e20150317173417120250.htm

Really, how long do you think Samsung can hold off against putting Qualcomm in all its devices if they do not improve their GPU tech? They are lucky this time due to being first to 14nm ff for the S6, its making that device look competitive against typical Snapdragon high-end, which are still on 20nm.

Its still a tabloid news site. They also said a north korean missile was found in alaska etc. Plus just look at their news.

 

[Silverforce11]

If Samsung is to buy AMD with debt, contracts and excess workers. They want that money back. And that ROI would be a real long tooth even in the best case.

They have the foundry, they're ready with 14nm ff!

GF is using Samsung 14nm ff tech, which undoubtedly AMD would have already made designs (and advanced stages) for that node with their CPU, APU and GPU. Not far fetched to assume a quick transition to production at Samsung foundries.

The ROI would be almost immediate.

If they can actually acquire x86/x64 (this is the tricky part), they can start their crusade into Servers, HPC and if the next-gen AMD CPU/APU isn't garbage, they would have their tech in Samsung branded desktops & notebooks instead of Intel.

Edit: I think whether x86/x86 is transferable in a take over or not will be the main deciding factor whether Samsung would be interested in throwing away some chump change for a major opportunity in a new market.

 

[5150Joker]

Intel decides the x86 transfer and the chances of that patent being licensed to a foreign conglomerate and approved by the US govt are probably as good as me winning the lotto. It's probably why Samsung or any other company hasn't bothered taking over AMD. Some of their IP and engineering talent may be worth it, but not the company as a whole.

 

[ShintaiDK]

They have the foundry, they're ready with 14nm ff!

GF is using Samsung 14nm ff tech, which undoubtedly AMD would have already made designs (and advanced stages) for that node with their CPU, APU and GPU. Not far fetched to assume a quick transition to production at Samsung foundries.

The ROI would be almost immediate.

If they can actually acquire x86/x64 (this is the tricky part), they can start their crusade into Servers, HPC and if the next-gen AMD CPU/APU isn't garbage, they would have their tech in Samsung branded desktops & notebooks instead of Intel.

Edit: I think whether x86/x86 is transferable in a take over or not will be the main deciding factor whether Samsung would be interested in throwing away some chump change for a major opportunity in a new market.

So you say Samsung more or less gets 5B$+ back right away. From a company that cant make money today and is severely R&D starved for future products? Plus they are willing to compete head on, not only with Intel but also nVidia? Samsung isnt that dumb.

How big is Samsungs 14nm capacity? Whats the wafer cost? And what about AMDs contracts, they dont dissapear either by a buyout.

A buyout would rather end in a massacre. Because the only thing worthwhile will be the IP. But the IP isnt worth that much.

 

[Silverforce11]

So you say Samsung more or less gets 5B$+ back right away. From a company that cant make money today and is severely R&D starved for future products? Plus they are willing to compete head on, not only with Intel but also nVidia? Samsung isnt that dumb.

How big is Samsungs 14nm capacity? Whats the wafer cost?

Before Samsung moved into the panel display market, imagine the same naysayers telling them "You want to risk billions setting up a manufacturing chain, to compete against LG/Sony/Panasonic?!"..

Likewise for the Memory & SSD market!

Then for the SOC market, competing against TSMC & GF!

Surely your memory must not be so short.

 

[ShintaiDK]

Before Samsung moved into the panel display market, imagine the same naysayers telling them "You want to risk billions setting up a manufacturing chain, to compete against LG/Sony/Panasonic?!"..

Likewise for the Memory & SSD market!

Then for the SOC market, competing against TSMC & GF!

Surely your memory must not be so short.

Those are low entry R&D segments. Plus for the panels, Samsung have already put it into a seperate company. Meaning a future sale is underway. And what happend to Samsungs PC division?

Again, how is Samsung going to get all those billions back, plus the profit they expect on a regular basis? AMD is pretty much gone in the PC segment. So is it the low margin semicustom that will make the day?

And what about that 14nm? How is the cost and wafer volume?

 

[5150Joker]

Samsung is also losing business in smartphones now. The Chinese brands are chipping away at them and it won't be long before they catch up. It would be suicidal of Samsung to try to compete against Intel if by some miracle they ever did get an x86 license -- IBM learned that the hard way. Samsung isn't stupid, they will stick to ARM SoC and AMD's only real value to them is graphics IP which they will either buy, license or just poach engineers and create a version that's similar but enough to pass USPTO.

 

[Silverforce11]

Samsung is also losing business in smartphones now. The Chinese brands are chipping away at them and it won't be long before they catch up.

A leader with foresight will know that without diversification, they will shrink against the massive competition in mobiles.

Servers and HPC, especially GPU-accelerated HPC is a growing market, ripe for the picking for someone with the foundries at the cutting edge.

Some here think its such a massive risk buying AMD, no, its chump change. They setup massive production for panels, TV, at vastly greater cost to compete in a market that at the time, was completely dominant with giants like Sony, LG & Panasonic. When they started into the FAB business, it was a major investment, also against giants in the field. Why did they started making ram & storage modules? Because they didn't want to pay others for it each time they sold a device, it would mean less profit potential as well as being beholden to others.

Likewise for their foundry, didn't want to pay TSMC for SOCs. Now they make their own SOCs and also make for others.

Each time they sell a PC, notebook, ultrabook, server, they are making money for Intel. Why should they? They have their own fabs ready on the leading tech node.

 

[Erenhardt]

Because they didn't want to pay others for it each time they sold a device, it would mean less profit potential as well as being beholden to others.

Likewise for their foundry, didn't want to pay TSMC for SOCs. Now they make their own SOCs and also make for others.

Each time they sell a PC, notebook, ultrabook, server, they are making money for Intel. Why should they? They have their own fabs ready on the leading tech node.

Why pay amd licensing fees if you can buy it whole for much less in the not-so-long-run.

 

[5150Joker]

Silverforce, you're repeating history from a time with different players and economies of scale. When Samsung first started selling TVs, they were crappy B&W ones sold to China. Now their TV dominance is again being challenged by upstarts like Vizio and Chinese brands that are selling TVs with more features at lower prices so their position is hardly cemented in any category. As for HPC's, it's a drop in the bucket right now and NVIDIA is dominating with Tesla and even Intel with it's new Phi is having a tough time catching up.

 

[Keysplayr]

Called it a few years ago. Samsung SoCs are rubbish when it comes to GPU tech, they are using generic MALI cores from ARM which is lacking, at the same node, they have horrible perf and perf/w. They needed a full node advantage as seen in Galaxy 6 to match rivals Qualcomm (Adreno IP from ATI!) and IM's PowerVR (from the old days when 3d accelerators started!).

It's the same problem Intel faces when they try to compete in the mobile scene. Even with their node advantage its not competitive versus ARM + Adreno/PowerVR or even Mali.

Samsung also has a strong presence on desktops & notebooks and they want to control what goes inside their products, basically the entire chain if possible. Reliance on Intel/Qualcomm is losing them potential profit with each product sold. Likewise they see a strong growth in HPC/Servers.

This is a win for consumers, because AMD has the tech, IP & engineers, but they lack good management & marketing as well as reliance on TSMC. Samsung's foundries making Radeons & APUs? Aw yeah!

Samsung SoCs are not rubbish and their performance is far from horrible.
http://bgr.com/2015/03/02/galaxy-s6-vs-htc-one-m9-benchmark/

 

[itsmydamnation]

no one seems to be considering K12, maybe they dont care about x86? there is lots of other stuff AMD offers that would take significant time to develop and implement, like an entire NUMA architecture and interconnects and server grade platforms.


I dont think they will by AMD, but i haven't seen a good reason in this thread from the naysayers as to why not.