[Kitguru] Microsoft might be interested in buying AMD

[NTMBK]

Don't forget that Microsoft also has thousands and thousands of servers in their Azure platform (which is also powering the XBox Live backend). A cheap source of x86 compatible CPUs would be no bad thing for them.

 

[Headfoot]

Anti trust regulations are nowhere near as stringent as people here seem to think. Very, very few deals have ever been barred and even fewer companies have been broken up. The chance is nearly nil.

 

[gorobei]

the only way this would be marginally plausible is if MS is really intent on pushing hololens. there was a weird announcement a while back by amd about getting another console soc contract but it was never specified though everyone assumed it meant the next nintendo console.

given the feedback on the latest hololens portable experience and the weight complaints, a more powerful soc/apu seems to be needed. if ms was serious about taking on facebook/oculus with hololens, they would need some custom soc cued towards portable low power and higher end graphics that wouldnt necessarily be part of amd's future plans and could be satisfied by an off the shelf solution like the 7850+jaguar in the ps4 and xbone.

but given how dire amd's situation is, it seems unlikely ms would need to buy them to get amd to make a really custom soc that they couldnt reuse in their normal markets.

 

[Genx87]

Feel like AMD is in a situation that one of these companies has to be looking at its IP library. Microsoft, Apple, Samsung, ect. These companies can pay a premium and still get AMD really cheap.

 

[Dribble]

I am pretty sure AMD has things all these companies want, but none of them actually want AMD the business. It's more of vultures circling the wounded beast getting ready to pick the bones clean when it finally collapses then anyone actually wanting to save it.

 

[Stuka87]

Looking past any anti-trust stuff. Does MS already have an x86 license? Would Intel be willing to license it to them if not? Buying AMD for the GPU's is one thing, but the XBox uses an 8 core x86 CPU as well. Can't be made without a license which is non-transferable.

 

[SexyK]

Makes no sense on the CPU side. If MS wants to keep growing the Surface business, they need Intel on board with them. AMD doesn't have anything that could come close to powering a slimmer, faster, fanless Surface Pro 4 in the near future.

 

[Phynaz]

Next person that posts an unfounded AMD acquisition rumor should get a ban.

How's this,
According to sources Phynaz is interested in buying AMD.

 

[Headfoot]

Looking past any anti-trust stuff. Does MS already have an x86 license? Would Intel be willing to license it to them if not? Buying AMD for the GPU's is one thing, but the XBox uses an 8 core x86 CPU as well. Can't be made without a license which is non-transferable.

there's tons of ways around this. Reverse triangular merger. Reverse takeover. Plenty of ways to let the surviving company technically be AMD but AMD is really being bought by Microsoft. This is a solved problem.

 

[railven]

Next person that posts an unfounded AMD acquisition rumor should get a ban.

How's this,
According to sources Phynaz is interested in buying AMD.

At current buyout prices, seems very plausible.

 

[Smoblikat]

M$ will now sell CPU's on a pay-per-core basis. This is the perfect sales pitch for AMD CPU's.

 

[railven]

M$ will now sell CPU's on a pay-per-core basis. This is the perfect sales pitch for AMD CPU's.

I'd figure they'd turn it into a subscription service.

 

[frozentundra123456]

I can see MS, and the other rumored buyers wanting some of AMD's products or IP. But since AMD needs money so badly, I would think they could just license a sweet deal to get whatever product or IP the need and get the product without having to take on the risk and debt of owning AMD. I mean, does anyone think Sony/MS didnt get a sweet deal from AMD on the console chips?

 

[therealnickdanger]

According to sources Phynaz is interested in buying AMD.

Yeah, but I heard that Phynaz is only looking to acquire AMD in order to do some patent-trolling. That acquisition wouldn't offer much, innovation-wise.

 

[ShintaiDK]

Don't forget that Microsoft also has thousands and thousands of servers in their Azure platform (which is also powering the XBox Live backend). A cheap source of x86 compatible CPUs would be no bad thing for them.

Even if the CPUs costed 0 it would still be a bad deal vs Intel in the server segment. Running cost would simply destroy the option.

 

[NTMBK]

Even if the CPUs costed 0 it would still be a bad deal vs Intel in the server segment. Running cost would simply destroy the option.

I was assuming that Zen will be more competitive in perf/W. Piledriver would be a disaster, agreed.

 

[NTMBK]

How's this,
According to sources Phynaz is interested in buying AMD.

Fudzilla article in 5, 4, 3, 2...

 

[Phynaz]

^ LOL

 

[destrekor]

Yea, buying ATI was a great move. That is why the market value of the combined companies is 1/3 of what they paid for ATI alone, not even counting for inflation.

If they didn't, AMD would probably be dead and snapped up by the likes of Qualcomm or other partner. So, while it hasn't brought them huge success, it has helped bring them enough success to stay afloat and retain ownership.

If Zen as is good as it has a chance to be, there will be the profits in the CPU line, they'll have the chance to help increase the R&D for the GPU line. Struggling with both product lines is really hurting them right now, but it's probably the only reason they are even producing their own CPUs at this point. ATI bought them more time, and they were right that the APU approach was better than what they were doing.

But the CPUs in the APUs have been utter crap, so they need to get that line fixed up first.
The Discrete GPU business isn't where AMD is really concerned about, it's a nice bonus but they want strong APUs more than anything. So their R&D focus is on the CPUs, and it seems like they've been putting money into Zen as sort of a moonshot, taking their time with it to really get it right. If they hit the mark and the CPU portion of their APU focus is strong and the CPU line makes them more money again, they'll focus on Discrete GPUs so that the tech trickles down to the APU products. New core design for smaller nodes, probably moving on from GCN, and if successful, it'll take a few years to really work out but that'll get them to a stronger point than they have been for a long while. I think if that works out they'll be stronger than AMD or ATI ever was individually, at least, not since the heyday of ATi and AMD's Athlon XP and Athlon 64 era.

 

[ocre]

With the gross mismanagement that has been AMD, perhaps.

But I still can't imagine ATI dropping to the low twenty percent market share.
The acquisition was really really bad for them.

 

[Auric]

Honestly the best move AMD ever made was buying ATI and the best move ATI ever made was buying ArtX. Getting the complete and total LOCK on the millions and millions of consoles sold globally was a gamble that is certainly helping them stay afloat. The margins might be thin, but the revenue keeps them relevant.

Yes, but the conventional wisdom is that they overpaid (compounded by the market peak), and erred in selling the handheld business to Qualcomm just as demand was going to increase exponentially.

 

[boozzer]

if APUs can do high setting 1080p gaming, amd would make truck loads. this is probably 1 or 2 more generations away. end of 2016 or 2017.

that is all amd needs.

 

[therealnickdanger]

Yes, but the conventional wisdom is that they overpaid (compounded by the market peak), and erred in selling the handheld business to Qualcomm just as demand was going to increase exponentially.

Totally agree - that seemed like a misstep.

 

[therealnickdanger]

if APUs can do high setting 1080p gaming, amd would make truck loads. this is probably 1 or 2 more generations away. end of 2016 or 2017. that is all amd needs.

High 1080p gaming in 2017 with games from 2014? I could see that. There are several problems with the idea that this will help AMD :

1. AMD has been offering the best IGP performance for over a decade, until the (much more expensive) Intel Iris Pro arrived, and how exactly did it help them "make truckloads"?
2. If possible, AMD needs to find a way to better utilize the GPU power they have to get competitive general compute performance whilst also bringing down their power/heat profile. I keep hearing about HSA, but haven't seen any day-to-day applications that use it...
3. Right now, the cost savings between a high-end AMD APU and a low-end Intel CPU with dGPU still makes the APU a hard sell for desktop gaming.
4. By 2017, 1080p will be the 720p/786p of this era. We'll be arguing about which IGP does better 4K.

The simple truth is that AMD has to be the budget part until it can beat Intel in pure performance (most important) and perf/W (important to enterprise and ecoterrorists). I think AMD's IGP strength is meaningless in the big picture - the only caveat being real-world HSA performance.

I think Carrizo is a good step in the direction AMD needs to go - assuming it is cheap enough to undercut Intel+dGPU or Iris Pro - and assuming AMD's marketing slides aren't complete fluff. Throw in some HBM or another huge cache like Crystalwell and we could see a shift on mobile. Toss Windows 10/DX12 into the mix and then it could go either way. If games/application are eventually able to effectively use HSA across any CPU/GPU on a given system, then what is an advantage for AMD APU would also be an advantage for Intel+dGPU.

 

[Silverforce11]

1. AMD has been offering the best IGP performance for over a decade, until the (much more expensive) Intel Iris Pro arrived, and how exactly did it help them "make truckloads"?

Because the best iGPU when they are all slow doesn't equate to much. For actual PG gaming, you still needed a dGPU, particularly at 1080p. It hasn't actually changed.

Also, AMD still has the fastest iGPU in an APU. It's in the PS4.

What we need is R390X class with HBM in an iGPU. Feasible? Well, the PS4 is 7870 class with GDDR5.

On 14nm, Zen + GCN2 + HBM2, could easily give 980/390X performance class in an APU. That's when low-end & mid-range dGPU will die off and what's left, are high end options, very expensive.