Its throughput was 1.28 Tb/s and it was impressive enough at the the time that Intel decided they needed their own new interconnect system immediately. Here is an article from 2 years ago. http://news.softpedia.com/news/Intel-Takes-AMD-s-E...
I assume that is the aggregated throughput. Since that fabric supports huge amount of nodes, the per node throughput may be too little to support the CPUs that the customers want. Remember that fabric was designed 7-8 years ago to support Atom N2XX that were shipping at that time. Once Seamicro realized that Atom didn't sell and moved to Xeon (and dump the term "micro server" in that process) , the per-node throughput of that fabric might become such a bottleneck that Xeon can't perform. Has Seamicro ever claimed 10G NIC support or even SATA class storage performance? Have you ever seen any 3rd party benchmark for SM boxes? I have seem none. Anandtech has done 2-3 benchmark in the past few years in this area and they never included SM boxes. Why? Maybe AMD didn't allow any 3rd party to benchmark their boxes. If this is the case, what were they trying to hide?
Neither AMD nor Seamicro has ever disclosed any of their sale figures.
There was nothing left to sell, they only shipped 1 product and it probably didn't sell well. Half the valuation of Seamicro was probably prospective business opportunities, the other half IP. They retain the (diminished) IP and of course the business opportunities and leads either dried up or weren't there to begin with.
This seems to be an unfortunately unsurprising outcome of a somewhat struggling company acquiring outside if their core business. hopefully the patents and talent will be useful still
What a grossly mismanaged company. Would have definitely gone bankrupt by now without the ATI acquisition to prop them up. They've done a good job running that into the ground as well though.
Partially true, others would argue the over-valuation of ATI sandbagged the company with debt and forced it into many of these poor decisions, shifting the company's focus and turning it on to this disastrous path of the "Future is Fusion". As a result they've had to sell their fabs, among other things to stay afloat.
Tbh, both companies would have been better off if the merger never happened. Fusion, hUMA, HSA and all the buzzwords that have sprung from the idea of GPU+CPU, while looking great on paper, have never amounted to anything positive for AMD in the marketplace or on Wall Street.
Because the ATI acquisition didn't prop them up, it put them into the position of being grossly over-leveraged to begin with. Certainly mismanagement, but many would point to the ATI acquisition as the cause, not the savior.
I suspect the problem with fusion/HSA was that they tried to get it going in a market where they have so little control over anything. HSA done right requires lots of small changes to the BIOS, to the OS proper, to the frameworks, to the apps. Doing it in x86 means waiting for everyone else to get on board --- and everyone else asking "Is Intel going to do this? In the same way? Because if not, too much hassle." Basically you can't win.
If they had committed all-in to ARM with HSA at the time, they might have been able to go down a very different path, offering a far superior product to an eco-system that is happy to accept change, an OS vendor (Google) that is happy to create competition for Intel, and app writers that don't whine every time some new API comes along that does things differently from the way they were done back in 1993.
Really good, thought provoking analysis, totally agree. We've even seen recently with the "official" launch of HSA 1.0 so much of it relies on shifting thought on the software side of things. Nvidia faced a similar battle when they introduced CUDA and brought GPGPU to the mainstream but took a grassroots approach that took time. The big difference is Nvidia didn't bet the farm or lose sight of their core competencies in growing their compute business, they piggy-backed all the R&D on their mature and industry-leading consumer GPU business.
AMD on the other hand clearly over-emphasized the value the market placed on the IGP, thinking it was the competitive advantage they needed to beat Intel. In doing so, they lost sight of their core competency which SHOULD have been to maintain market leadership in CPU performance first and foremost.
I agree, they should have licensed tech to each other and restructured separately. I think part of the problem is that they kept trying to "diversify" to stay afloat instead of channeling resources into just a couple of markets. There are two strategies to business, one is integration where you try to get a hold of multiple markets and one is where you focus on a core competency, be good at what you do. AMD grew too fast and was too scattered. It's that simple. They tried to compete on a "system on a chip" level while still investing in things like this server business, but the truth is that they weren't ready to compete with Intel's might and money. Intel was 10x larger and had much more experience. The CEOs thought their research teams could conquer anything and got a big head from the Athlon success. It's a classic tale of simply taking out your wallet too quickly thinking you are going to make more money. People like Hector Ruiz were raking in dough while the company was hurting and mismanaged it. They had "vision" but were poor CEOs in that they tried to diversify too much and didn't focus on core competencies. They saw AMD as big business, but vs Intel they were in fact a small business with a need to focus more on specific projects.
Back when the A64 was the king the top model was expensive (700$), but the popular moderately clocked ones, like the ones you're buying from Intel, were priced OK (300 - 400$).
No, the entry-level Athlon 64s started at $420 MSRP but in reality sold for $450-500 due to low supply/high demand. And the Socket 939 boards were *MUCH* more expensive than previous AMD boards with Via or AMD chipsets, $200+.
Some readers of this space might find it surprising but I was a huge AMD CPU fan during the Athlon Socket A era, but they simply priced out their biggest fans and that made the decision to go Intel very easy once they started dominating AMD with Conroe.
But yeah Intel's dominant virtual monopoly position shows us concerns about pricing are simply unfounded, as these dominant players are still forced to compete against themselves and will need to offer compelling options at existing price points to entice consumers to buy new products.
I agree on the AMD pricing, but intel was also selling their top end above the $700 AMDs Core 2 extremes at $1000.
"But yeah Intel's dominant virtual monopoly position shows us concerns about pricing are simply unfounded, as these dominant players are still forced to compete against themselves and will need to offer compelling options at existing price points to entice consumers to buy new products."
The only reason they haven't had prices skyrocket is because AMD still has another option. If they raised prices like crazy people would buy AMD products because Intel is being stupid. Once AMD is gone there is only one play to get x86-64 and thats intel. Intel can then charge whatever they want and people can't do anything about it.
Yeah again, not going to argue with you on this one if you're not going to bother fact-checking. AMD hit first with their close to $1000 SKUs with their Athlon FX line-up, Intel reacted and Core 2 obviously came after AMD's Athlon FX chips because at that point Intel had regained performance and market leadership with Conroe (Core 2 Duo).
Intel actually announced the p4 extreme edition "gallatin" a week before amd released the athlon fx line in a move many agree was a bid to steal amd's thunder. The fx 51 was released at $733 according to an old anand article and the p4 extreme was released at $999 according to wikipedia.
I'd have taken the FX-51 any day of the week. I have owned just about every chip between the single core Athlon 3000+ to the FX-60/Opty 185 for 939 and I'd still take an Opty 165 over a dual core Pentium D any day of the week.
Intel couldn't compete with AMD's mid ranged CPUs, let alone their flagship CPU. Intel's flagship Pentium 4 was sold for more cost than AMD's matching Athlon - a low end 2.0ghz dual core I believe. You were obviously deluded at the time.
AMD had to charge higher prices. They were still selling a lot less chips than the Pentium 4 was, because of Intel's anti-competitive practices, which AMD sued for. Economies of scale wasn't in AMD's favor at the time, and their chips were a LOT better.
AMD has been keeping Intel's prices low, and the recent gains in overall efficiency and power for CPUs is much less today than it was during previous years. There is less and less reason to upgrade, so to entice buyers the prices have to be lower.
Just add it to a laundry list of poor executive decisions, poor vision, poor management of limited funds. Many questioned this move when it happened, and as usual AMD's anti-Midas touch has turned this questionable acquisition into a steaming pile of poop just a few short years later.
Of course the bigger problem is this casts a huge shadow of doubt over AMD's entire ARM strategy, as their backers and proponents kept insisting AMD would dominate this micro server market with their ARM-based Zen designs. Now, you have to ask why bother with ARM at all if there's not a vertical market in place for AMD's ARM chips.
The IT landscape has changed pretty drastically -- Microservers were a reaction to Virtual Machine overhead that technologies like LXC, Windows Containers, and Docker have obsoleted. Its more efficient now to run many containers on one fat host than yo split them across many tiny physical servers. If rumors are true about Zen's architecture (which essential allows for running many light threads on a core with dedicated resources each, or alternatively fewer heavy threads faster by hanging those resources together) then it will be a good fit for container technologies but an antithetical fit for micro-servers. I'm sure they got some good IP and people, but it was a good bet that didn't age well, and its better to cut it loose than to hold on without a market. As to why they didn't sell it, I suspect that the rest of the world realized what I just wrote, too, and there's not much value left after taking the IP and people out if it.
It wouldn't surprise me if NVIDIA stepped up to the plate. They've been after an x86 license for a long time, and it would knock out their main GPU competitor at the same time.
best case scenario, intel buys their gpu tech, and nvidia buys the x86 tech. nvidia gets the x86 license from intel, intel continues the radeon line, and we have two healthy x86 and gpu vendors again. and the death on intel hd graphics YAY!
Exactly, the only way it could happen is if Intel signed off and allowed the license to transfer, which they mmight consider to avoid anti-trust scrutiny. But it would have to be the right buyer, they wouldn't allow it to transfer to a high market cap company like Samsung or Qualcomm simply because that would challenge their x86 dominance. Something like Via would be a good, safe landing spot for that x86 license.
Actually they don't they lost all previous licenses and agreements when they bought Cyrix from National Semiconductor, and they lost the license Centaur had when they bought Centaur from IDT. It was all settled after court battles and because it was deemed Centaur owns patents and thus invented tech which Intel needs to use they settled and gained an x86-license that way.
AMD is a public company so anybody could buy them but you might get into trouble when you get the controlling majority or if you wish to merge the company. The agreement is obviously not transferable to another business. VIA didn't really purchase businesses that did have x86-licenses just the design companies that worked for National and IDT.
Point is, Via holds an x86 license. At that, though, they've never shown any interest in competing with Intel or AMD. They do their industrial/embedded thing with it, and seem to be happy enough about it.
What I wonder is how that might go if AMD goes under/decides to sell out. Intel holds rights of refusal on the transfer of AMD's x86 license, but what does that look like once government regulators start to look into it? That is, say Samsung steps in, Intel says (without really saying, of course) "no, sorry, you might actually provide competition, no x86 license for you." Would the government say, "too bad, they've got the money and you damn sure don't get to be the only license-holder in the markets you play in"?
Not quite. If a 'change of ownership' happens, AMD will not automatically 'lose' their x86 license, rather, the current licensing arrangement comes up for renegotiation. Many assume that means Intel would refuse to continue the current licensing arrangement, but those people miss the fact that AMD licenses AMD64 instructions to Intel. Yes, that's right Intel users. Your Intel processors are running 64bit AMD microcode, deep down in their little cores. So, basically, if Intel refused to license x86 instructions to AMD, AMD could revoke Intel's license to run AMDx86-64 instructions. Basically, it would be thermonuclear war. So, the common consensus is that another company could buy AMD, and probably renegotiate essentially the same cross-licensing arrangement AMD already has with Intel. In other words, it's not that much of a big problem for a prospective buyer of AMD, especially one with deep pockets like Samsung, Qualcomm or Apple.
You're wrong. The AMD Intel truce is quite specific:
6.3 Change of Control. In the event of a Change of Control of AMD, the definition of AMD Microprocessor as defined in Section 1.5 shall be limited to those devices that fell within Section 1.5 on the date of the Change of Control and shall further be limited to x86 AMD Microprocessors for use in a Personal Computer. “Change of Control” shall mean: (1) any Person or group of Persons (as the term “group” is interpreted pursuant to Rule 13d-5 under the Securities Exchange Act of 1934, as amended) (the “Acquiring Person”) acquires (i) beneficial ownership of capital stock of AMD entitling the holder(s) thereof to more than fifty percent (50%) of the voting power of the then outstanding capital stock of AMD with respect to the election of directors of AMD, or (ii) an interest sufficient to receive more than fifty percent (50%) of the profits or losses of AMD; or (2) AMD enters into a merger, consolidation, reorganization or similar transaction (or series of related transactions) with any Person or group of Persons in which less than fifty percent (50%) of the voting power of the outstanding capital stock of AMD (if it is the surviving entity) or of the Acquiring Person (if it is the surviving entity) with respect to the election of directors following such transaction is held directly or indirectly by Persons who were shareholders of AMD immediately prior to such transaction (or series of transactions); or (3) AMD sells to any Person(s) in one or more related transactions properties or assets representing all or substantially all of the properties and assets of AMD.
Buying seamicro was stupid right from the beginning. They should've known that they couldn't sell Intel processors any longer and that their own processors just weren't good enough. And it also seems that they didn't even tried, because they only released one model. I didn't knew that. What the actual f*ck? I doubt that the IP they bought with it was worth that.
I seem to remember people giving high prasie to AMD for buying this company and how this was going to be a boon for them. I'm not an AMD basher either. The last 9 video cards i have bought are all ATI(ooops sorry) AMD brand. Just another sign of a poorly managed company reaching for something to help it stay afloat. If NV were to try to buy AMD, regulators would be sure to split it up. Hey, we could even see an ATI comeback if that were the case.
It is official, Rory Read is a Piece of S**t. Once AMD decides to kill its ARM attempt he can be upgraded to something worse.
The whole AMD board who bought RR and agreed on his Sea Micro acquisition should basically quit.
3 year dog and pony show neglecting the traditional bread and butter, not able to acheive any meaningful success but catastrophes with their new strategic initiatives, even Dirk atleast had execution to show for his time....
BTW if anybody tries to BS semi custom APU, RR got everything in a silver platter (SNY materialized in 2006 and MS in 2007 time frame) which looks now that any buffoon would have got something right.
He was a huge mistake in leadership. The last bit of his stupidity is now gone. The current leadership is solid. People will be pleased once the new batch of GPGPUs are out. Then even more pleased with Keller's teams work.
Sell when others are greedy and buy when others are fearful. Sounds like everybody's pretty freaking fearful about AMD right now, so it's probably time to load up on shares for cheap. After all, the worst that's likely to happen to this 45 year old tech archon is that it will be acquired for all it's IP for two or three times it's current market cap.
On the other hand, it probably wasn't a good idea for a company that wants to sell processors into the larger server market, to also sell its own servers in competition with its other customers. It's probably better to turn the SeaMicro designs into a reference platform design and implementation unit for working with the OEMs themselves.
I think it was part of this idea to get more and more income from branching out into as many markets as possible. To a point, it's worked, but AMD spending money on SeaMicro at a time when they could've done with fixing Bulldozer's ills obviously didn't go too well. They can still use the IP, however, and cutting off a gangrenous limb to save the body might be a smart idea, but how much money did it actually cost AMD in the end?
It looks to me as if Chipzilla got played for $30- to $40 billion (++plus++) ... and it only cost AMD $400 million over 5 years to storm back into a $20+ billion/yr server market.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
60 Comments
Back to Article
sprockkets - Thursday, April 16, 2015 - link
What, they didn't even bother to sell it?AMD will join the rest of the desktop era into irrelevancy soon enough, if not already.
SaberKOG91 - Thursday, April 16, 2015 - link
IP that was bought is still useful. Freedom Fabric is a wicked fast interconnect system.HighTech4US - Thursday, April 16, 2015 - link
Was that worth $334 million?enzotiger - Friday, April 17, 2015 - link
Really? Do you know its throughput? Do you have any data to back up your claim?haekuh - Friday, April 17, 2015 - link
Its throughput was 1.28 Tb/s and it was impressive enough at the the time that Intel decided they needed their own new interconnect system immediately. Here is an article from 2 years ago.http://news.softpedia.com/news/Intel-Takes-AMD-s-E...
enzotiger - Friday, April 17, 2015 - link
I assume that is the aggregated throughput. Since that fabric supports huge amount of nodes, the per node throughput may be too little to support the CPUs that the customers want. Remember that fabric was designed 7-8 years ago to support Atom N2XX that were shipping at that time. Once Seamicro realized that Atom didn't sell and moved to Xeon (and dump the term "micro server" in that process) , the per-node throughput of that fabric might become such a bottleneck that Xeon can't perform. Has Seamicro ever claimed 10G NIC support or even SATA class storage performance? Have you ever seen any 3rd party benchmark for SM boxes? I have seem none. Anandtech has done 2-3 benchmark in the past few years in this area and they never included SM boxes. Why? Maybe AMD didn't allow any 3rd party to benchmark their boxes. If this is the case, what were they trying to hide?Neither AMD nor Seamicro has ever disclosed any of their sale figures.
chizow - Thursday, April 16, 2015 - link
There was nothing left to sell, they only shipped 1 product and it probably didn't sell well. Half the valuation of Seamicro was probably prospective business opportunities, the other half IP. They retain the (diminished) IP and of course the business opportunities and leads either dried up or weren't there to begin with.stefstef - Wednesday, April 22, 2015 - link
i would not bet on that.Shorty_ - Saturday, March 7, 2020 - link
Just checking in to see how this prediction panned out? :)zangheiv - Monday, February 7, 2022 - link
LOL this genius's comment didnt age too well. Seamicro is where Infinity Fabric came fromextide - Thursday, April 16, 2015 - link
Wow, this semms like an odd move...MrSpadge - Thursday, April 16, 2015 - link
They're burning themselves properly, if they want to arise like Phoenix from the ashes soon.lefty2 - Thursday, April 16, 2015 - link
I wonder if the ARM server business will be the next thing to go. It's a similar sort of unproven product.Taneli - Friday, April 17, 2015 - link
The market potential is small and Intel is effectively destroying it with their Xeon D pricingporphyr - Thursday, April 16, 2015 - link
This seems to be an unfortunately unsurprising outcome of a somewhat struggling company acquiring outside if their core business. hopefully the patents and talent will be useful stillJtaylor1986 - Thursday, April 16, 2015 - link
What a grossly mismanaged company. Would have definitely gone bankrupt by now without the ATI acquisition to prop them up. They've done a good job running that into the ground as well though.Jtaylor1986 - Thursday, April 16, 2015 - link
Entire company is now worth ~1/3 of the actual ATI acquisition value FWIWxype - Friday, April 17, 2015 - link
Wow, that’s sad and puts it all into perspective in a pretty harsh way.chizow - Thursday, April 16, 2015 - link
Partially true, others would argue the over-valuation of ATI sandbagged the company with debt and forced it into many of these poor decisions, shifting the company's focus and turning it on to this disastrous path of the "Future is Fusion". As a result they've had to sell their fabs, among other things to stay afloat.Tbh, both companies would have been better off if the merger never happened. Fusion, hUMA, HSA and all the buzzwords that have sprung from the idea of GPU+CPU, while looking great on paper, have never amounted to anything positive for AMD in the marketplace or on Wall Street.
Jtaylor1986 - Thursday, April 16, 2015 - link
I don't see how buying a company you couldn't afford doesn't fall under the realm of mismanagement.chizow - Thursday, April 16, 2015 - link
Because the ATI acquisition didn't prop them up, it put them into the position of being grossly over-leveraged to begin with. Certainly mismanagement, but many would point to the ATI acquisition as the cause, not the savior.name99 - Friday, April 17, 2015 - link
I suspect the problem with fusion/HSA was that they tried to get it going in a market where they have so little control over anything. HSA done right requires lots of small changes to the BIOS, to the OS proper, to the frameworks, to the apps. Doing it in x86 means waiting for everyone else to get on board --- and everyone else asking "Is Intel going to do this? In the same way? Because if not, too much hassle." Basically you can't win.If they had committed all-in to ARM with HSA at the time, they might have been able to go down a very different path, offering a far superior product to an eco-system that is happy to accept change, an OS vendor (Google) that is happy to create competition for Intel, and app writers that don't whine every time some new API comes along that does things differently from the way they were done back in 1993.
chizow - Friday, April 17, 2015 - link
Really good, thought provoking analysis, totally agree. We've even seen recently with the "official" launch of HSA 1.0 so much of it relies on shifting thought on the software side of things. Nvidia faced a similar battle when they introduced CUDA and brought GPGPU to the mainstream but took a grassroots approach that took time. The big difference is Nvidia didn't bet the farm or lose sight of their core competencies in growing their compute business, they piggy-backed all the R&D on their mature and industry-leading consumer GPU business.AMD on the other hand clearly over-emphasized the value the market placed on the IGP, thinking it was the competitive advantage they needed to beat Intel. In doing so, they lost sight of their core competency which SHOULD have been to maintain market leadership in CPU performance first and foremost.
JonnyDough - Tuesday, April 21, 2015 - link
I agree, they should have licensed tech to each other and restructured separately. I think part of the problem is that they kept trying to "diversify" to stay afloat instead of channeling resources into just a couple of markets. There are two strategies to business, one is integration where you try to get a hold of multiple markets and one is where you focus on a core competency, be good at what you do. AMD grew too fast and was too scattered. It's that simple. They tried to compete on a "system on a chip" level while still investing in things like this server business, but the truth is that they weren't ready to compete with Intel's might and money. Intel was 10x larger and had much more experience. The CEOs thought their research teams could conquer anything and got a big head from the Athlon success. It's a classic tale of simply taking out your wallet too quickly thinking you are going to make more money. People like Hector Ruiz were raking in dough while the company was hurting and mismanaged it. They had "vision" but were poor CEOs in that they tried to diversify too much and didn't focus on core competencies. They saw AMD as big business, but vs Intel they were in fact a small business with a need to focus more on specific projects.YuLeven - Thursday, April 16, 2015 - link
Pathetic. In this pace Intel will continue to roam free, charging all they want for their chips.chizow - Thursday, April 16, 2015 - link
I've paid the same amount for the same relative level of performance since Core 2 Duo btw, never more than $300 for a great performance Intel chip.E6600 $283
Q6600 $299
i7 920 $199
i7 4770K $229
i7 5820K $299
Certainly much less than what AMD wanted when they had a big lead over Intel during the Athlon 64 era.
MrSpadge - Friday, April 17, 2015 - link
Back when the A64 was the king the top model was expensive (700$), but the popular moderately clocked ones, like the ones you're buying from Intel, were priced OK (300 - 400$).chizow - Friday, April 17, 2015 - link
No, the entry-level Athlon 64s started at $420 MSRP but in reality sold for $450-500 due to low supply/high demand. And the Socket 939 boards were *MUCH* more expensive than previous AMD boards with Via or AMD chipsets, $200+.Some readers of this space might find it surprising but I was a huge AMD CPU fan during the Athlon Socket A era, but they simply priced out their biggest fans and that made the decision to go Intel very easy once they started dominating AMD with Conroe.
But yeah Intel's dominant virtual monopoly position shows us concerns about pricing are simply unfounded, as these dominant players are still forced to compete against themselves and will need to offer compelling options at existing price points to entice consumers to buy new products.
Crunchy005 - Friday, April 17, 2015 - link
I agree on the AMD pricing, but intel was also selling their top end above the $700 AMDs Core 2 extremes at $1000."But yeah Intel's dominant virtual monopoly position shows us concerns about pricing are simply unfounded, as these dominant players are still forced to compete against themselves and will need to offer compelling options at existing price points to entice consumers to buy new products."
The only reason they haven't had prices skyrocket is because AMD still has another option. If they raised prices like crazy people would buy AMD products because Intel is being stupid. Once AMD is gone there is only one play to get x86-64 and thats intel. Intel can then charge whatever they want and people can't do anything about it.
chizow - Friday, April 17, 2015 - link
Yeah again, not going to argue with you on this one if you're not going to bother fact-checking. AMD hit first with their close to $1000 SKUs with their Athlon FX line-up, Intel reacted and Core 2 obviously came after AMD's Athlon FX chips because at that point Intel had regained performance and market leadership with Conroe (Core 2 Duo).siberus - Saturday, April 18, 2015 - link
Intel actually announced the p4 extreme edition "gallatin" a week before amd released the athlon fx line in a move many agree was a bid to steal amd's thunder. The fx 51 was released at $733 according to an old anand article and the p4 extreme was released at $999 according to wikipedia.JonnyDough - Tuesday, April 21, 2015 - link
I'd have taken the FX-51 any day of the week. I have owned just about every chip between the single core Athlon 3000+ to the FX-60/Opty 185 for 939 and I'd still take an Opty 165 over a dual core Pentium D any day of the week.JonnyDough - Tuesday, April 21, 2015 - link
Intel couldn't compete with AMD's mid ranged CPUs, let alone their flagship CPU. Intel's flagship Pentium 4 was sold for more cost than AMD's matching Athlon - a low end 2.0ghz dual core I believe. You were obviously deluded at the time.JonnyDough - Tuesday, April 21, 2015 - link
AMD had to charge higher prices. They were still selling a lot less chips than the Pentium 4 was, because of Intel's anti-competitive practices, which AMD sued for. Economies of scale wasn't in AMD's favor at the time, and their chips were a LOT better.AMD has been keeping Intel's prices low, and the recent gains in overall efficiency and power for CPUs is much less today than it was during previous years. There is less and less reason to upgrade, so to entice buyers the prices have to be lower.
chizow - Thursday, April 16, 2015 - link
Just add it to a laundry list of poor executive decisions, poor vision, poor management of limited funds. Many questioned this move when it happened, and as usual AMD's anti-Midas touch has turned this questionable acquisition into a steaming pile of poop just a few short years later.Of course the bigger problem is this casts a huge shadow of doubt over AMD's entire ARM strategy, as their backers and proponents kept insisting AMD would dominate this micro server market with their ARM-based Zen designs. Now, you have to ask why bother with ARM at all if there's not a vertical market in place for AMD's ARM chips.
ravyne - Thursday, April 16, 2015 - link
The IT landscape has changed pretty drastically -- Microservers were a reaction to Virtual Machine overhead that technologies like LXC, Windows Containers, and Docker have obsoleted. Its more efficient now to run many containers on one fat host than yo split them across many tiny physical servers. If rumors are true about Zen's architecture (which essential allows for running many light threads on a core with dedicated resources each, or alternatively fewer heavy threads faster by hanging those resources together) then it will be a good fit for container technologies but an antithetical fit for micro-servers. I'm sure they got some good IP and people, but it was a good bet that didn't age well, and its better to cut it loose than to hold on without a market. As to why they didn't sell it, I suspect that the rest of the world realized what I just wrote, too, and there's not much value left after taking the IP and people out if it.bleh0 - Thursday, April 16, 2015 - link
Someone needs to buy AMD or just purchase their IPs.Gigaplex - Thursday, April 16, 2015 - link
It wouldn't surprise me if NVIDIA stepped up to the plate. They've been after an x86 license for a long time, and it would knock out their main GPU competitor at the same time.TheinsanegamerN - Thursday, April 16, 2015 - link
best case scenario, intel buys their gpu tech, and nvidia buys the x86 tech. nvidia gets the x86 license from intel, intel continues the radeon line, and we have two healthy x86 and gpu vendors again. and the death on intel hd graphics YAY!Krysto - Friday, April 17, 2015 - link
Or even better, Samsung or Qualcomm buys AMD, and actually becomes a strong competitor to Intel.Taneli - Friday, April 17, 2015 - link
Especially Samsung with their competitive fabs and large end user sales would be interesting.nofumble62 - Thursday, April 16, 2015 - link
IP? each patent cost about $20K to file. Can AMD afford it?looncraz - Thursday, April 16, 2015 - link
From my understanding, AMD cannot be bought out, they will lose their x86 license.chizow - Friday, April 17, 2015 - link
Exactly, the only way it could happen is if Intel signed off and allowed the license to transfer, which they mmight consider to avoid anti-trust scrutiny. But it would have to be the right buyer, they wouldn't allow it to transfer to a high market cap company like Samsung or Qualcomm simply because that would challenge their x86 dominance. Something like Via would be a good, safe landing spot for that x86 license.anubis44 - Friday, April 17, 2015 - link
VIA already has Cyrix's old x86 license, which they obtained when they bought them out over a decade ago.Penti - Friday, April 17, 2015 - link
Actually they don't they lost all previous licenses and agreements when they bought Cyrix from National Semiconductor, and they lost the license Centaur had when they bought Centaur from IDT. It was all settled after court battles and because it was deemed Centaur owns patents and thus invented tech which Intel needs to use they settled and gained an x86-license that way.AMD is a public company so anybody could buy them but you might get into trouble when you get the controlling majority or if you wish to merge the company. The agreement is obviously not transferable to another business. VIA didn't really purchase businesses that did have x86-licenses just the design companies that worked for National and IDT.
fluxtatic - Friday, April 24, 2015 - link
Point is, Via holds an x86 license. At that, though, they've never shown any interest in competing with Intel or AMD. They do their industrial/embedded thing with it, and seem to be happy enough about it.What I wonder is how that might go if AMD goes under/decides to sell out. Intel holds rights of refusal on the transfer of AMD's x86 license, but what does that look like once government regulators start to look into it? That is, say Samsung steps in, Intel says (without really saying, of course) "no, sorry, you might actually provide competition, no x86 license for you." Would the government say, "too bad, they've got the money and you damn sure don't get to be the only license-holder in the markets you play in"?
anubis44 - Friday, April 17, 2015 - link
Not quite. If a 'change of ownership' happens, AMD will not automatically 'lose' their x86 license, rather, the current licensing arrangement comes up for renegotiation. Many assume that means Intel would refuse to continue the current licensing arrangement, but those people miss the fact that AMD licenses AMD64 instructions to Intel. Yes, that's right Intel users. Your Intel processors are running 64bit AMD microcode, deep down in their little cores. So, basically, if Intel refused to license x86 instructions to AMD, AMD could revoke Intel's license to run AMDx86-64 instructions. Basically, it would be thermonuclear war. So, the common consensus is that another company could buy AMD, and probably renegotiate essentially the same cross-licensing arrangement AMD already has with Intel. In other words, it's not that much of a big problem for a prospective buyer of AMD, especially one with deep pockets like Samsung, Qualcomm or Apple.andychow - Tuesday, April 28, 2015 - link
You're wrong. The AMD Intel truce is quite specific:6.3 Change of Control. In the event of a Change of Control of AMD, the definition of AMD Microprocessor as defined in Section 1.5 shall be limited to those devices that fell within Section 1.5 on the date of the Change of Control and shall further be limited to x86 AMD Microprocessors for use in a Personal Computer.
“Change of Control” shall mean:
(1) any Person or group of Persons (as the term “group” is interpreted pursuant to Rule 13d-5 under the Securities Exchange Act of 1934, as amended) (the “Acquiring Person”) acquires (i) beneficial ownership of capital stock of AMD entitling the holder(s) thereof to more than fifty percent (50%) of the voting power of the then outstanding capital stock of AMD with respect to the election of directors of AMD, or (ii) an interest sufficient to receive more than fifty percent (50%) of the profits or losses of AMD; or
(2) AMD enters into a merger, consolidation, reorganization or similar transaction (or series of related transactions) with any Person or group of Persons in which less than fifty percent (50%) of the voting power of the outstanding capital stock of AMD (if it is the surviving entity) or of the Acquiring Person (if it is the surviving entity) with respect to the election of directors following such transaction is held directly or indirectly by Persons who were shareholders of AMD immediately prior to such transaction (or series of transactions); or
(3) AMD sells to any Person(s) in one or more related transactions properties or assets representing all or substantially all of the properties and assets of AMD.
Novacius - Thursday, April 16, 2015 - link
Buying seamicro was stupid right from the beginning. They should've known that they couldn't sell Intel processors any longer and that their own processors just weren't good enough. And it also seems that they didn't even tried, because they only released one model. I didn't knew that. What the actual f*ck? I doubt that the IP they bought with it was worth that.JoJoman88 - Thursday, April 16, 2015 - link
I seem to remember people giving high prasie to AMD for buying this company and how this was going to be a boon for them. I'm not an AMD basher either. The last 9 video cards i have bought are all ATI(ooops sorry) AMD brand. Just another sign of a poorly managed company reaching for something to help it stay afloat. If NV were to try to buy AMD, regulators would be sure to split it up. Hey, we could even see an ATI comeback if that were the case.Krysto - Friday, April 17, 2015 - link
The strategy was fine. It's the management/limited funds that have been hurting AMD for the past 5+ years.rocketbuddha - Thursday, April 16, 2015 - link
It is official,Rory Read is a Piece of S**t.
Once AMD decides to kill its ARM attempt he can be upgraded to something worse.
The whole AMD board who bought RR and agreed on his Sea Micro acquisition should basically quit.
3 year dog and pony show neglecting the traditional bread and butter, not able to acheive any meaningful success but catastrophes with their new strategic initiatives, even Dirk atleast had execution to show for his time....
BTW if anybody tries to BS semi custom APU, RR got everything in a silver platter (SNY materialized in 2006 and MS in 2007 time frame) which looks now that any buffoon would have got something right.
Jeesh!
mdriftmeyer - Saturday, April 18, 2015 - link
He was a huge mistake in leadership. The last bit of his stupidity is now gone. The current leadership is solid. People will be pleased once the new batch of GPGPUs are out. Then even more pleased with Keller's teams work.Krysto - Friday, April 17, 2015 - link
I have just one word for AMD: SELL!anubis44 - Friday, April 17, 2015 - link
And I have an expression for you:Sell when others are greedy and buy when others are fearful. Sounds like everybody's pretty freaking fearful about AMD right now, so it's probably time to load up on shares for cheap. After all, the worst that's likely to happen to this 45 year old tech archon is that it will be acquired for all it's IP for two or three times it's current market cap.
mdriftmeyer - Saturday, April 18, 2015 - link
Keep your advice. I am investing $10k next week.psychobriggsy - Friday, April 17, 2015 - link
Sheesh. It's never positive news from AMD.On the other hand, it probably wasn't a good idea for a company that wants to sell processors into the larger server market, to also sell its own servers in competition with its other customers. It's probably better to turn the SeaMicro designs into a reference platform design and implementation unit for working with the OEMs themselves.
silverblue - Friday, April 17, 2015 - link
I think it was part of this idea to get more and more income from branching out into as many markets as possible. To a point, it's worked, but AMD spending money on SeaMicro at a time when they could've done with fixing Bulldozer's ills obviously didn't go too well. They can still use the IP, however, and cutting off a gangrenous limb to save the body might be a smart idea, but how much money did it actually cost AMD in the end?Smell This - Saturday, August 26, 2017 - link
[snicker]It looks to me as if Chipzilla got played for $30- to $40 billion (++plus++) ... and it only cost AMD $400 million over 5 years to storm back into a $20+ billion/yr server market.