AMD Doesn't Regret Spinning off GlobalFoundries

[btarunr]

AMD co-founder Jerry Sanders, in 2009 was famously quoted as stating that "real men have fabs," a jibe probably targeted at the budding fab-less CPU designers of the time. Years later, AMD spun-off its silicon fabrication business, which with a substantial investment of the Abu Dhabi government through its state-owned Advanced Technology Investment Company (ATIC), became GlobalFoundries (or GloFo in some vernacular). This company built strategic partnerships with the right players in the industry, acquisitions such as IBM's fabs, and is now at the forefront of sub-10 nm fab development. It remained one of AMD's biggest foundry partners besides TSMC and Samsung, and is manufacturing its AMD processors at a brand new facility in Upstate New York, USA.

AMD, on the other hand, doesn't regret spinning off GloFo. Speaking at Merrill Lynch Global Technology and Investment Conference, CTO Mark Papermaster said, that going fab-less has helped AMD focus on chip-design without worrying about manufacturing. Production is no longer a bottleneck for AMD, as it can now put out manufacturing contracts to a wider variety of foundry partners. Its chip-designers aren't limited by the constraints of an in-house fab, and can instead ask external fabs to optimize their nodes for their chip-designs, Papermaster said. 14 nm FinFET has added a level of standardization to the foundry industry.



View at TechPowerUp Main Site

 

[Totally]

How I read it, to paraphrase: "Well...before instead of investing in our own fabs, we gave our execs sweet comp packages and as a result our fabs fell behind and started hindering design and product release cycles. Now that we've spun the fabs off it's now someone else's problem. Comp packages are as sweet as they used but now we can claim it's a good thing."

 

[R0H1T]

I bet they regret the purchase of the makers of Radeon, at that price, or having settled with Intel as they may well have gotten 2~5x more.

 

[TheGuruStud]

I bet they regret the purchase of the makers of Radeon, at that price, or having settled with Intel as they may well have gotten 2~5x more.

More a case of get what you can or you'll never get it. Intel could tie up a payout for a decade. They need cash like a payday loan. A just payout would have been over 10x.

Just googled, oh, looky, Intel still hasn't paid the Euro fine. SURPRISE. That'll get drug out till it's reduced to nothing.

Large corporations never have to pay. They own everyone and wrote the laws. They can't even be forced to pay fines to the federal govt (or rather the govt WON'T enforce it).

edited multiple times b/c I can't read

 

[R0H1T]

More a case of get what you can or you'll never get it. Intel could tie up a payout for a decade. They needed the cash like a payday loan. A just payout would have been over 10x.

Yeah that's why I mentioned the purchase of ATI, had they not overspent there who knows what they could've made from the case going to trial.

 

[evernessince]

I bet they regret the purchase of the makers of Radeon, at that price, or having settled with Intel as they may well have gotten 2~5x more.

Well seeing as it was either accept Intel's money or go bankrupt, I don't think they really had much of a choice. A free market without regulations isn't really free as every entity in a laissez faire market will seek to use any exploit they can find. You need only look at modern America and see all the mega corperations. Where there used to be 100s of potato chip companies, there are now less then a handful. Where there used to be many local cable companies, there are now a handful of large ones.

 

[qubit]

Everyone writes it as "GlobalFoundries", but the company writes GLOBALFOUNDRIES everywhere on its website, which I think is pretentious and looks stupid.

I'm glad that AMD made the right strategic move here. I wonder if it would be good for Intel too, or is there some advantage for them keeping it in-house?

 

[PowerPC]

Everyone writes it as "GlobalFoundries", but the company writes GLOBALFOUNDRIES everywhere on its website, which I think is pretentious and looks stupid.

I'm glad that AMD made the right strategic move here. I wonder if it would be good for Intel too, or is there some advantage for them keeping it in-house?

I would bet there isn't. This is why AMD now can outpace them while being a fraction of the size, and delivering cheaper to consumers. If you only have your own fab like Intel you need to work with those people, even if they do subpar work. AMD can just choose another fab tomorrow. Intel has shown time and time again that it is archaic to the core. They are a bloated company that has absolutely no flexibility in this day and age. This is why they have literally done the same thing over and over for the last 10+ years.

 

[bug]

It you look strictly from the point of view of their design division, that is correct.
However, conglomerates exist for a reason: when design stalls, you live on manufacturing and vice-versa.

It's water under the bridge, how they feel about that decision matters little today.

 

[Joss]

A free market without regulations isn't really free as every entity in a laissez faire market will seek to use any exploit they can find. You need only look at modern America and see all the mega corperations

That's a contradiction.
A free, unregulated market doesn't have exploits to be used; that's the nature of "unregulated".
The problem is that our markets, societies, are in fact heavily regulated, monstrously so, hidden behind an appearance of free Capitalism that is no longer. That is what allows global corporations the clout they have.

 

[Totally]

I would bet there isn't. This is why AMD now can outpace them while being a fraction of the size, and delivering cheaper to consumers. If you only have your own fab like Intel you need to work with those people, even if they do subpar work. AMD can just choose another fab tomorrow. Intel has shown time and time again that it is archaic to the core. They are a bloated company that has absolutely no flexibility in this day and age. This is why they have literally done the same thing over and over for the last 10+ years.

I have to disagree, this is only ownership of fabs, it does not preclude them from using others which is why I'm calling bs on the statement. It's like saying just because someone owns a car they aren't allowed to use the train or other transportation services when one is more practical than the other. Owning a fab or not doesn't lock them into specific operations.

 

[bug]

I would bet there isn't. This is why AMD now can outpace them while being a fraction of the size, and delivering cheaper to consumers. If you only have your own fab like Intel you need to work with those people, even if they do subpar work. AMD can just choose another fab tomorrow. Intel has shown time and time again that it is archaic to the core. They are a bloated company that has absolutely no flexibility in this day and age. This is why they have literally done the same thing over and over for the last 10+ years.

If intel's work is subpar, then how come it took AMD 10 years to catch up?
Also, if you don't have your own fab, you have to compete for capacity with other fabless players.
And if intel is so inflexible, how did they manage to branch out into data centers, storage, IoT and even security while the nimble AMD didn't?

 

[sergionography]

Everyone writes it as "GlobalFoundries", but the company writes GLOBALFOUNDRIES everywhere on its website, which I think is pretentious and looks stupid.

I'm glad that AMD made the right strategic move here. I wonder if it would be good for Intel too, or is there some advantage for them keeping it in-house?

Well it depends as There are pros and cons in having anything done inhouse vs "outsourcing". Pros are optimizing the fab process to be a perfect fit for the type of processors you intend to build(low power, high power, clockspeed etc) and having better fine control and quality control over the process which could lead to better overall performance. on the downside/cons it is more expensive and challenging to manage because it adds another big division/department in the company. With intels business model, revenue, and market share, it is doable as it can sustain the fabs. Moving forward however lets for example assume amd picks up considerable market share at the expense of intel then it might be a no brainer to spin off the fabs as intels fabs are not a revenue generator and are rather an operating cost which would thin out their profits; especially at the current rate of how increasingly expensive fab development amd operations has become. Another option they would have ofcourse is to adjust their business model to turn their fabs into a revenue generating division where its run as a normal foundry that contracts customers, which we have been hearing about for years with few contracts here and there, Although with little impact so far and much competition in that segment. This move would also fragment intel and is operationally even more challenging to operate than they are today.

 

[PowerPC]

I have to disagree, this is only ownership of fabs, it does not preclude them from using others which is why I'm calling bs on the statement. It's like saying just because someone owns a car they aren't allowed to use the train or other transportation services when one is more practical than the other. Owning a fab or not doesn't lock them into specific operations.

Even if they could take the bus, would Intel do it? If someone invested 100.000 $ in a new Porsche, how many people like that take the bus? Yea, not many. Even if it's cheaper, better for the environment and often even gets you faster from A to B. They will never see those advantages because they are too invested in their car.

 

[PowerPC]

If intel's work is subpar, then how come it took AMD 10 years to catch up?
Also, if you don't have your own fab, you have to compete for capacity with other fabless players.
And if intel is so inflexible, how did they manage to branch out into data centers, storage, IoT and even security while the nimble AMD didn't?

Well, 10 years ago was a very different time and I never said AMD was perfect either. But it seems like AMD has evolved in a big way in these 10 years while being totally overshadowed by Intel this whole time. They seem to have learned a lot of lessons in those 10 years while let's be honest, Intel has remained stagnant. Yes, Intel had some smaller innovations here and there in the meantime, but so did AMD with their GPUs. However, these smaller innovations isn't what Intel and AMD compete on and not the most important part of Intel's business. It's CPUs.

 

[bug]

Well, 10 years ago was a very different time and I never said AMD was perfect either. But it seems like AMD has evolved in a big way in these 10 years while being totally overshadowed by Intel this whole time. They seem to have learned a lot of lessons in those 10 years while let's be honest, Intel has remained stagnant. Yes, Intel had some smaller innovations here and there in the meantime, but so did AMD with their GPUs. However, these smaller innovations isn't what Intel and AMD compete on and not the most important part of Intel's business. It's CPUs.

Intel didn't stagnate, they simply shifted focus to mobile. Look at where mobile chips are now vs 10 years ago.

 

[I No]

Well, 10 years ago was a very different time and I never said AMD was perfect either. But it seems like AMD has evolved in a big way in these 10 years while being totally overshadowed by Intel this whole time. They seem to have learned a lot of lessons in those 10 years while let's be honest, Intel has remained stagnant. Yes, Intel had some smaller innovations here and there in the meantime, but so did AMD with their GPUs. However, these smaller innovations isn't what Intel and AMD compete on and not the most important part of Intel's business. It's CPUs.

The only reason why Intel remained "stagnant" is that x86 is as old as the sun. There is only so much you can improve on something. Everyone keeps demanding 20-40% jumps gen to gen cuz yeah let's cram more transistors into the silicon until you can't anymore. And since i did mention that x86 is as old as fire I don't see anyone else pushing something else out of the assembly line other than Intel regarding a new architecture for us mere mortals cuz let's face it, AMD has the RnD budget to barely improve on the existing stuff let alone build a whole new arch from scratch.
On the GPU side of things AMD's improving on TeraScale since day one. Not even going to mention that AMD buying ATI was a bone-head move. Spinning off GloFlo falls into the same category (the basic idea was fine) they are also one of the key factors that held AMD back (terrible yields comes to mind) the other one would be the PR team which needed to be taken outside the building and shot on multiple occasions (Bulldozer, Fury, etc.). Just my 2 cents on the whole "Intel is offering small to none performance boosts across gens"

 

[PowerPC]

The only reason why Intel remained "stagnant" is that x86 is as old as the sun. There is only so much you can improve on something. Everyone keeps demanding 20-40% jumps gen to gen cuz yeah let's cram more transistors into the silicon until you can't anymore. And since i did mention that x86 is as old as fire I don't see anyone else pushing something else out of the assembly line other than Intel regarding a new architecture for us mere mortals cuz let's face it, AMD has the RnD budget to barely improve on the existing stuff let alone build a whole new arch from scratch.
On the GPU side of things AMD's improving on TeraScale since day one. Not even going to mention that AMD buying ATI was a bone-head move. Spinning off GloFlo falls into the same category (the basic idea was fine) they are also one of the key factors that held AMD back (terrible yields comes to mind) the other one would be the PR team which needed to be taken outside the building and shot on multiple occasions (Bulldozer, Fury, etc.). Just my 2 cents on the whole "Intel is offering small to none performance boosts across gens"

I'm not demanding 20-40% IPC per core jumps. I'm demanding innovation in the CPU market and specifically I'm referring to current events. As soon as AMD releases more cores, Intel magically pulls it out of their back pocket, as if they weren't sitting on it for years. Just compare Threadripper that is being released in a couple of weeks to what Intel wants to release in Q4 with X299... Don't tell me Intel wasn't purposefully staying stagnant because they had no competition. This has been going on for years! It wasn't even that hard to notice how little Intel cared about innovation in the CPU market and how much they just wanted to keep everything as is.

 

[bug]

I'm not demanding 20-40% IPC per core jumps. I'm demanding innovation in the CPU market and specifically I'm referring to current events. As soon as AMD releases more cores, Intel magically pulls it out of their back pocket, as if they weren't sitting on it for years. Just compare Threadripper that is being released in a couple of weeks to what Intel wants to release in Q4 with X299... Don't tell me Intel wasn't purposefully staying stagnant because they had no competition. This has been going on for years! It wasn't even that hard to notice how little Intel cared about innovation in the CPU market and how much they just wanted to keep everything as is.

It's called common sense. If you can afford to sit tight and your competition won't catch up for 10 years, that's what you'll do. I really don't see your problem.

 

[I No]

I'm not demanding 20-40% IPC per core jumps. I'm demanding innovation in the CPU market and specifically I'm referring to current events. As soon as AMD releases more cores, Intel magically pulls it out of their back pocket, as if they weren't sitting on something like this for years. Just compare Threadripper that is being released in a couple of weeks to what Intel wants to release in Q4 with X299... Don't tell me Intel wasn't purposefully staying stagnant because they had no competition. This has been going on for years!

From a business's point of view would you have done things differently ? Of course the "new" line-up was shelved from a while back. If AMD were in Intel's shoes they would have done the same thing. Usually running businesses into the ground tends to be bad for investors and whatnot. The only reason this keeps happening is that there is no competition and you can't blame the company that's doing it better for that. AMD shot their own feet on several occasions (VEGA @ Computex comes to mind out of the recent mishaps). At the end of the day they are all businesses and they need to make money and at this point AMD needs to play catch-up you do attract more flies with honey rather than vinegar now don't you?

 

[Fx]

That's a contradiction.
A free, unregulated market doesn't have exploits to be used; that's the nature of "unregulated".
The problem is that our markets, societies, are in fact heavily regulated, monstrously so, hidden behind an appearance of free Capitalism that is no longer. That is what allows global corporations the clout they have.

+1

 

[Foobario]

When Intel had a two node lead they were able to charge a premium for their chips that gave them sufficient cash flow to fund the R&D necessary to have their own fans.

That node advantage has slipped to only a one node lead and it seems Intel has been starving their architecture R&D to fund their manufacturing R&D which has allowed AMD to essentially catch up. This is in spite of the fact that AMD is competing with their first iteration of 14nm (really 20nm) while Intel is on their third iteration of a true 14nm node. So architecture can trump manufacturing when there is only a one node disparity, evidently.

Intel has poured billions upon billions of dollars in an attempt to stay ahead in the node races. While the entire semiconductor industry, outside Intel, has been funding the R&D of the foundries and it seems the fabless approach is going to win as it allows emphasis on architecture that has real world advantages.

Had AMD not sold off the fabs they would have gone bankrupt years ago. In less than five years, if Intel doesn't spin off their fabs the cost to keep up with everybody else in manufacturing will put them down a path to insolvency, as well.

 

[TheinsanegamerN]

That's a contradiction.
A free, unregulated market doesn't have exploits to be used; that's the nature of "unregulated".
The problem is that our markets, societies, are in fact heavily regulated, monstrously so, hidden behind an appearance of free Capitalism that is no longer. That is what allows global corporations the clout they have.

That's a contradiction.

A free, unregulated market has no restrictions that can be enforced. Every single exploit we legislate out would be allowed in a truly free market.

price fixing to drive out competitors, buying up all the supply in an area, paying to legislate your opponents out, buying your opponents and shutting them down, price fixing, legislating smaller opponents out of business, ece are all exploits that companies have used in the past, all would be perfectly legal in a 100% free market. Some of these are now illegal, some are still being used.

to claim that a unregulated market would have no exploits shows both a complete misunderstanding of both free market principals and misunderstanding of how economics work. Any truly free market will quickly come under monopolistic control due to the prevalence of shady tactics.

The "free market" of companies competing to provide the best service per dollar can only exist with regulations to prevent abuse. Those regulations wouldnt exist if companies hadnt abused them. That's how those regulations came to exist in the first place.

 

[Vayra86]

I would bet there isn't. This is why AMD now can outpace them while being a fraction of the size, and delivering cheaper to consumers. If you only have your own fab like Intel you need to work with those people, even if they do subpar work. AMD can just choose another fab tomorrow. Intel has shown time and time again that it is archaic to the core. They are a bloated company that has absolutely no flexibility in this day and age. This is why they have literally done the same thing over and over for the last 10+ years.

Doubtful, Intel isn't stupid, and has already moved towards streamlining the business in the past few years. Yes they are committed to their fabs, at the same time they still have a node advantage in some ways, especially in terms of capacity. Also, nothing stops other chip designers to knock on Intel's door and nothing stops Intel to start 'selling' their fab capacity either. Historically, having technology in-house is definitely an advantage in the marketplace, and just the fact that GF and TSMC are pushing out new announcements and show growth, says nothing about long term stability. Not too long ago GF was seen as the ugly baby among fabs, and largely inferior to TSMC. Meanwhile, Intel has always been and still is top of the line.

At the same time, Intel has established a dominance on x86 that isn't easily broken down. Ryzen is a first step, but it is by no means forcing Intel out of the business. Ever since Ryzen got released people are grossly overestimating AMD's comeback. Its a snapshot conclusion, not one based on long term success.

 

[Joss]

to claim that a unregulated market would have no exploits shows both a complete misunderstanding of both free market principals and misunderstanding of how economics work

You're out of your depth, so I leave it here.
I urge you to read a couple of books though, here and here, it'll clarify things for you.