AMD, Nvidia Reportedly Tripped Up On Process Shrinks

itwbennett writes: In the fierce battle between CPU and GPU vendors, it's not just about speeds and feeds but also about process shrinks. Both Nvidia and AMD have had their move to 16nm and 20nm designs, respectively, hampered by the limited capacity of both nodes at manufacturer TSMC, according to the enthusiast site WCCFTech.com. While AMD's CPUs are produced by GlobalFoundaries, its GPUs are made at TSMC, as are Nvidia's chips. The problem is that TSMC only has so much capacity and Apple and Samsung have sucked up all that capacity. The only other manufacturer with 14nm capacity is Intel and there's no way Intel will sell them some capacity.

Comments

I got an idea

[Anonymous Coward]

Build your own fab

Re:I got an idea

[Anonymous Coward]

ain't nobody got time for that!

Re:I got an idea

[Anonymous Coward]

Buy them out instead... It worked for CSG. :P

Re:I got an idea

[Anonymous Coward]

Still have to remember that AMD is a fabless company.

Re:I got an idea

[Luckyo]

And get owned by intel's illegal monopolistic deals with OEMs?

Been there, done that. Signed: AMD.

Re:I got an idea

[alexander_686]

Maybe yes, maybe no.

Fabs are incredibly capital intensive. If you build one you sort of need to run it at full hilt to make money off of them. And then you are on the tread mill of always having to build another one to have the latest and greatest.

I am not saying they should not build it, just that it is a completely different beast then design and marketing of their chips. You can't do it half way. FYI, almost everybody agrees that the reason why the x386 dominates today is not because Intel had the best or most efficient designs for microchips but that it had the best run and most efficient fabs. Intel has publicly said it want to be the "McDonalds" on the CPU world – standardized and cheap.

Are AMD and Navida ready to bit off such a big chuck? I doubt it.

Re:I got an idea

[ArhcAngel]

Dear OEM X,
We are intrigued by your new technology Y. We have begun production of our own version we copied straight from your designs and need you to sign the attached licensing agreement giving us complete autonomy to use your patents as we see fit. We very much enjoy our relationship and would hate to see OEM X lose its ability to purchase our chips at base rate and be forced to seek them from 3rd party suppliers.

Sincerely, Intel

Re:I got an idea

[Anonymous Coward]

Fab cost for 16nm - 20nm process: $10 billion. There's a reason so few players have fabs for small feature sizes.

Re:I got an idea

[erice]

Build your own fab

While not a bad idea, it doesn't solve the problem. When you have your own fab, you are pretty much obligated to use it. Even when it is late, low on capacity, or a full node behind. You can reduce this risk by throwing a lot of money at R&D and spare capacity. However, this is more than a little bit expensive. That is why AMD doesn't have a captive fab anymore. They can't afford it.

TSMC is in the business of making chips. They don't make money if they can't make chips. I haven't heard that Apple or Samsung have an unexpected block buster products at 14nm. That means that either TSMC grossly underestimated demand at 14/16nm/20nm (not likely) or they are having manufacturing problems that are slowing production. If it were just TSMC screwing up, you would bet UMC other fabs would exploiting this opportunity to steal business. Since this isn't happening, it is good bet that a hypothetical AMD or Nvidia fab would have the same production trouble.

Re:I got an idea

[itzly]

It's both things. The x86 is a good design, and the intel fabs are good.

Re:I got an idea

[Anonymous Coward]

Build your own fab

AMD did exactly that. Intel responded by raising their prices and offering "discounts" to vendors in exchange for refusing to stock any AMD products, ensuring that AMD's production capacity was wasted because customers found it incredibly difficult to find anyone willing sell them (despite having a large performance, price and temperature advantage over Intel at the time). AMD took Intel to court over this and won, but Intel managed to drag the case out so ridiculously long that AMD weren't able to keep their foundries. Their R&D took a big hit too.

Re:I got an idea

[alexander_686]

I read an interview of John Sculley, ex Apple CEO, 10 years after he left Apple. (There is a link to it somewhere here on Slashdot. ) He said that one of his great mistakes was to choose Motorola's PowerPC RISC chip over Intel's x386 design. All of Apple engineers pulled for the Motorola chip, pointing to its better architecture. And it was true, for a given bit of silicon Motorola was better than Intel.

Except that Intel's fabs were so much more efficient than Motorola's, they were able to deliver more powerful chips at a cheaper price even with poorer design. Which allowed them to make fat profits, which they plowed back into newer fab plants, which let them sprint past Motorola.

Re:I got an idea

[Anonymous Coward]

What u talkin about Willis? They are fab-ulous!

Re:I got an idea

[Anonymous Coward]

Do you have a few billion $$ laying around? And a couple of years, hundreds of acres of spare land?

Re:I got an idea

[Anonymous Coward]

That's long-term thinking that doesn't serve the share holders and we will have NONE of that around here!!

Re:I got an idea

[unixisc]

Maybe yes, maybe no.

Fabs are incredibly capital intensive. If you build one you sort of need to run it at full hilt to make money off of them. And then you are on the tread mill of always having to build another one to have the latest and greatest.

I am not saying they should not build it, just that it is a completely different beast then design and marketing of their chips. You can't do it half way. FYI, almost everybody agrees that the reason why the x386 dominates today is not because Intel had the best or most efficient designs for microchips but that it had the best run and most efficient fabs. Intel has publicly said it want to be the "McDonalds" on the CPU world – standardized and cheap.

Are AMD and Navida ready to bit off such a big chuck? I doubt it.

Fully agree w/ this - it costs a ton of cash to just build the fab, then quite some time before they are in full production, and only after they've run for a few years and been depreciated that their products are profitable. By then, it's usually time to shrink, but by now, at ~15nm, there ain't too much scope for that.

But in addition to that, the costs of running a fab are recurring - even if there is no demand for a product, the fab has to be kept running, thereby piling on inventory. Which is a good reason why a lot of companies either started of as, or later became, fabless. The only one who is profitable is Intel, as a result of being a full 2 generations ahead of the competition. Reason? Because they make volumes of products that they can sell in the market profitably, thereby enabling themselves to put in a whole ton of cash into newer fabs, better equipment and everything else geared towards process improvements.

Re:I got an idea

[unixisc]

Wasn't it Gil Amelio who made that comment? But at the time, Intel was at par w/ PowerPC, and slower than other CPUs, like PA-RISC and Alpha. What made Intel catch up w/ RISC CPUs was Microsoft using the NT kernel to replace the Windows 9x kernel, and as a result, having an SMP capable mainstream OS. So now Intel could toss more cores into it to catch up w/ competing silicon, and because Intel's processes too were so much ahead of everyone else, those Core CPUs were still cheaper than PA-RISCs, Alpha's and SPARCs.

The other problem that Apple had was that b'cos their platform strategy was still in shambles - it was long before OS-X was up and running, they didn't have much demand in the market, and as a result, didn't have the volumes that might have helped Motorola/Freescale be competent w/ Intel.

Re:I got an idea

[Anonymous Coward]

It had a lot less to do with the fabs than with the fact that neither Motorola nor IBM were interested in a full-on performance competition with Intel or AMD. IBM considered the PowerPC as a gateway drug to their larger POWER architecture mainframes, and didn't really want to push it much past the 620. Motorola just wanted to make high-speed communication processors, so once it was fast enough for gigabit switches they stopped caring and stuck with the 4xx series. Apple was spending all its time cajoling them to make better processors, and it just wasn't happening, despite the fact that the PPC beat the x86 on several metrics because it wasn't saddled with a complex CISC-to-microcode converter that took 1/4 of the core's transistors. Meanwhile Intel and AMD were charging on ahead with higher speeds and, later, multicore chips of their own.

And IMHO Apple was never in a technical position to transition from 68k to x86. It would have been a lot harder to get the 68k emulator up to speed because the x86 architecture was vastly different from 68k, more than PPC was. And then there's the whole issue of little-endian vs. big-endian, which was far easier to manage once Apple had transitioned to Mac OS X and XCode than before when the primary developer tools were made by third parties such as Metrowerks. Prior to Mac OS X, Apple's developer tools were expensive, huge, and very slow. And this is ignoring the reality that, if they moved to x86 and commodity chipsets, it would have been a lot easier for a cloning market to grow and strangle Apple's sales, like Power Computing almost did.

So as much as Sculley may have wanted to move to x86, it likely would have resulted in developers abandoning the platform and the end of Apple. The technology just wasn't there.

Re:I got an idea

[simm_s]

Yes, what people do not understand is that Intel is primarily manufacturing company. Intel builds the fabs and R&D/demand fills them.

Re:I got an idea

[simm_s]

AMD failed because R&D could not deliver. In 2000 AMD totally owned the server space and was making huge inroads into the client PC space. Intel pivoted and beat them with better designs.

Re:I got an idea

[UnknownSoldier]

AMD: They are fab-u-less!

FTFY. :-)

Re:I got an idea

[Anonymous Coward]

Try reading how Intel paid Dell billions in cash to not use AMD chips. Mike Dell actually wanted to go with AMD, but Intel's massive cash infusions proved too delicious to ignore.

Re:I got an idea

[Trongy]

AMD did not own the sever space. With the introduction of AMD64, they had superior CPUs. However, thanks to intel's illegal monopolistic deals with OEMs they didn't get the sales and profits to plough back into R&D and sustain their technical lead.

Re:I got an idea

[cheesybagel]

There's some sort of analogous law for fab price similar to Moore's law. I think its like each new process fab, which you have to build every ~18 months, costs like twice as much as the previous generation.

Re:I got an idea

[cheesybagel]

Motorola also sold a lot less chips so they could not invest nearly as much on fab technology as Intel did. You can thank that to Apple's exorbitant prices back then.

If Motorola chips were that expensive they wouldn't sell as much as they did for things like automotive and washing machines and crap like that.

Re:I got an idea

[cheesybagel]

The article seems kinda bogus. Samsung uses their own fabs. AFAIK the capacity at TSMC is being used by Qualcomm (which does sell some Krait chips to Samsung) and Apple. Yes AMD (ex-ATI) and NVIDIA seem to have been left in the cold with this. It might push some of them to Globalfoundries if they can play their cards right. AMD at least has people with experience with their production process.

Re:I got an idea

[Anonymous Coward]

This man speaks the truth.

Re:I got an idea

[alexander_686]

Sculley said it was his mistake - which is not saying that Amelio didn't say it as well. Here is the old Slashdot link.

http://apple.slashdot.org/stor...

Re:I got an idea

[Anonymous Coward]

Gateway too. At the time AMD was the better value proposition and used DDR RAM. Intel chips were more expensive, about the same performance and use Rambus (RD) RAM. For some reason (which we know now) Gateway dropped the better value AMD chips and went with Intel instead, in a time when they were trying to be very competitive with the other manufacturers.

Re:I got an idea

[Kjella]

Well that too but there's a long feedback loop from profits to production capacity, building a new process facility typically takes 3-4 years. AMD held the lead from mid 1999 (launch of Athlon) to early 2006 (launch of Intel Core). First they had to make some money on that achievement, this still being the dotcom days the stock flew through the roof until the dotcom bubble burst in early 2000. Everything tech-related came crashing down, you could buy AMD stock in 2002 for less than in 1999 despite their excellent products at the time. AMD probably had a helluva time finding funding to expand.

So if you don't get serious ramp-up capital until 2002, well you're not going to get it online until 2005-2006 and if you look at the stock charts then January 2006 is when they peaked. All looked bright, AMD still had the best product and finally the production capacity to knock Intel out of the driver's seat. Then Intel Core gave them a real kick to the nuts. They just couldn't sell as much as they had planned for, the financial burden from the fabs was dragging the whole company down. That's the trouble with betting too much on growth, when you fail you fail hard. But then everyone had been shouting build more, take down Intel. Hindsight is easy.

Re:I got an idea

[jrumney]

By the time they got to multi-core, Intel had moved to a RISC architecture internally anyway.

Re:I got an idea

[TheRaven64]

No it isn't. There's a reason that AMD spun off their fab business. Fabs are insanely expensive and AMD doesn't have the volumes needed to justify a cutting edge fab by themselves. Having a fab running at anything below full production capacity is just throwing money away, but no one wants to buy capacity on an AMD fab because they know that when yields are low AMD chips will get priority. They'll buy capacity in a GlobalFoundaries or TMSC fab, because then they're just one customer in many.

The real question is why Intel won't fab the chips. Their commercial advantage has always been in production techniques, not in chip design. Throughout the lifetime of x86, other people have been able to compete with Intel designs, in spite of being one or two generations behind in process tech.

Re:I got an idea

[Anonymous Coward]

A fab would cost them almost as much as their annual revenue. It would potentially bankrupt them to gamble on such a risk.

Re:I got an idea

[Agripa]

Like the other commenters said, AMD never had the lead in monopolistic behavior. The billions Intel spend to keep Dell and Gateway and others from using AMD processors was cheaper then the development and manufacturing costs to do the same.

The only thing AMD managed to accomplish was sinking Itanic as a replacement for X86.

Re:I got an idea

[Agripa]

During the same time period I tried to avoid Motorola's products because they had poor availability. Anything they produced which did not have another source was on allocation.

Re:I got an idea

[Agripa]

Motorola did this with other lower profile products as well. Their availability was always questionable.

Re:I got an idea

[Agripa]

The real question is why Intel won't fab the chips.

Intel does fab other company's chips now but they are hardly going to do so for companies which directly compete with them.

Consumers tripped up on exclusive deals

[Anonymous Coward]

Who cares about the designs when my AMD GPU doesn't run software properly because Nvidia has a special deal with that software company? And visa-versa? Stop worrying about design and fix the real problem...your current products are unreliable because of greed.

Re:Consumers tripped up on exclusive deals

[Pubstar]

Someone doesn't know about Nvidia GameWorks, apparently.

Samsung?

[Anonymous Coward]

Don't they have their own FAB?

Re:Samsung?

[Anonymous Coward]

Yes they do.

Business must be really good if they still need to get help from TSMC to fill their orders...

Re:Samsung?

[alexander_686]

Actually, they may not. It boils down to what you define as "Samsung". "Samsung" is a lose collection of companies with crossholdings. IIRC, the fab plants are in a different company than the company that does the cell phones. Besides, just because you can make one bit of the phone (CPU) does not mean are the best at making other bits of the phone. I think they have farmed out the cell network bits out.

Is that why...

[moondo]

So, this is why we can't have nice things?

Re:Is that why...

[RLiegh]

Yes. Exactly.

bean counters ruin another company

[banbeans]

Who didn't see not having their own fabs was going to bite them in the rear?
Only a bunch of bean counters would not have seen this coming.

Re:bean counters ruin another company

[TWX]

It's a lot simpler to sit down at a CAD workstation to do layout design than it is to manufacture that design.

Trouble is, at the rate we're going, the manufacturers won't need us for very much longer, especially if economies in other countries that don't respect IP become strong enough that those manufacturers can sell unlicenced production there, even if they can't sell it here. I'm thinking of that happy little picture of the world with a circle over portions of South and East Asia and the large islands in the Indian Ocean where half of the population of the planet lives.

We don't need to manufacture everything domestically, but we need to be capable of manufacturing everything domestically, even if we outsource a lot of it. Being capable is not simply a technical matter, it's also an economic one in that if it becomes too expensive to start up production if a foreign source is lost to us, then we effectively can't produce the thing anymore.

Re:bean counters ruin another company

[gstoddart]

Pretty much this.

The people who own that fabrication capacity are selling it to the highest bidder. Or the one who invested the money to build it.

If you don't control your own production, don't be surprised when you have to wait in line.

But don't act all shocked that the 3rd party who actually builds your products can't guarantee your stuff gets built.

This gets a big "duh".

Re:bean counters ruin another company

[Anonymous Coward]

Haven't they both existed without fabs for most of 20 years?

Re:bean counters ruin another company

[DarkOx]

How are they bitten exactly. Neither one of them can get chips. There is therefore no real competitive disadvantage. They only way I see them getting bitten is if Intel decides they really want a slice of the high end GPU market; ups their game on the design side AND allocates their 14nm facilities to GPUs in favor of cranking out more CPUs.

Re:bean counters ruin another company

[gstoddart]

How are they bitten exactly. Neither one of them can get chips.

Well, then it's a good thing there's only two of them.

A third company who owned some fabrication facilities would completely screw things up.

Re:bean counters ruin another company

[Anonymous Coward]

Saw it coming the moment AMD purchased ATI outright.

Re:bean counters ruin another company

[Bengie]

AMD sold off their fab some time around 2009.

Re:bean counters ruin another company

[Anonymous Coward]

Even this is still cheaper unless the lost revenue exceeds the cost of building and operating your own fab, but it's true that placing the control of your core business in the hands of others is really ceding control of your business.

Re:bean counters ruin another company

[Damarkus13]

I guess you missed the part where Intel had already made the leap to a 14nm process.

Re:bean counters ruin another company

[Anonymous Coward]

Haven't they both existed without fabs for most of 20 years?

AMD no, they used to fab their own stuff till the manufacturing part of the company spun off. ATI graphics GPUs (not the IGP) are produced at TSMC.

Re:bean counters ruin another company

[OverlordQ]

Intel has been doing their own fab for a while. While AMD (well GlobalFoundries) does their CPUs, they dont do their GPUs.

Re:bean counters ruin another company

[fuzzyfuzzyfungus]

Does anyone know how good, bad, or awful Intel's GPUs are these days? I know that the performance of the GPUs they bother to sell is poor, that much is obvious; but GPU components parallelize nicely: pretty much all market segmentation for AMD, NVIDIA, and Intel (for a given generation of design) is a matter of how many of that manufacturer's GPU-sub-element you get. The most expensive one gets as many as the largest die will hold, the next most expensive gets the same die but with 25% or whatever disabled, and so on.

Are Intel's relatively tepid performance numbers just because their GPUs are relatively small(and have access only to a slice of DDR3 stolen from main memory, not fast dedicated GDDR5), and thus likely solvable in short order if they were to build larger ones; or are their designs themselves relatively tepid, and even allocating similarly generous die space and RAM would still be pretty pitiful?

Re:bean counters ruin another company

[itzly]

It does you little good to keep a fab, if you're not able to compete with intel on the process shrink rate.

Re:bean counters ruin another company

[DarkOx]

Well, then it's a good thing there's only two of them.

I think the bean counters might deserve some credit though. Don't you think somebody asked "what happens to our business if things go wrong at TSMC?" and the answer was well our chief competition also uses them so there will probably be little impact to our market share, they won't be able to supply anyone with chips either.

Re:bean counters ruin another company

[Anonymous Coward]

It does you little good to keep a fab, if you're not able to compete with intel on the process shrink rate.

No one competes with Intel on process shrink rate. So their margin's are lower than Intel's because their cost per gate is higher.

GF and TSMC must be quite concerned about Intel's custom foundry business.

Re:bean counters ruin another company

[ArmoredDragon]

We don't need to manufacture everything domestically, but we need to be capable of manufacturing everything domestically

That is already the present reality, but things are often manufactured overseas for a few reasons:

- Chinese companies have the capability of rapidly adjusting manufacturing processes as a result of last minute design changes. While technically US companies have this ability, most companies just won't do it (in some cases labor unions are the biggest hindrance because they only permit their members to do one job and one job only, and instead of re-allocating existing labor, they're forced to hire new people, which just isn't economical or practical.)

- Even though it is possible to find the required skill set in the US, often the workers you do find aren't as good at a particular task as some people who live overseas and do that kind of work all the time. For example, how many Americans do you know that are good at operating the machinery used for making textiles? Chances are, they're harder to find than in China, but if you really wanted to get it done here, you could, just you'll pay more, it'll take longer, and the craftsmanship probably won't be as good.

Sometimes for really small jobs though, it does make sense to hire local talent instead. For example, it's probably cheaper to have a special kind of shirt made in quantities of 10,000 in China, but if you need only 100 of them, then it's probably cheaper to have it done here instead.

At any rate, the US is still the top place to go for very high end semiconductor fabrication. Presently the worlds most advanced semiconductor fabrication plant is located in Chandler, Arizona. Though any products that come out of it will probably end up being stamped "Made in Malaysia" or some similar location, mainly because the final package is assembled there rather than here, which mainly has to do with the scenarios described above.

Re:bean counters ruin another company

[bws111]

Yep. That is why no other companies use 'suppliers'. Oh, wait. EVERY company is dependant on its suppliers. Does Intel operate its own mines? Does it create all of the tools, etc in its fabs? No? The stupid bean counters at Intel!

Re:bean counters ruin another company

[ArmoredDragon]

Does intel even make a discreet GPU? I thought they were all baked into their newer CPUs.

Re:bean counters ruin another company

[Anonymous Coward]

Who didn't see not having their own fabs was going to bite them in the rear? Only a bunch of bean counters would not have seen this coming.

Are you arguing that all companies should be completely vertically integrated, producing all components that go into their product themselves?

Re:bean counters ruin another company

[LessThanObvious]

It may have something to do with ATI legacy manufacturing as much as AMD CPU legacy. If ATI didn't have their own fab maybe it wasn't doable for AMD to outlay cash to create one for GPU, not that legacy ATI tooling would be up to task for modern GPU fab anyway.

Re:bean counters ruin another company

[Anonymous Coward]

They actually do create all their tools in-house. The only reason they do not directly operate their own mines is because Intel takes pride in their "conflict-free minerals" and they can only do that so long you can't show a direct link - by not directly operating them :)

Re:bean counters ruin another company

[gstoddart]

You attribute far more long-term thinking to the idiots who run companies than I do.

And are you really saying that if I really wanted to destroy large US companies all I'd have to do is to bomb factories in China? Or sow dissent? Or cut off their electrical supply?

Because that sounds like something out of a Bond film.

Hell, build your own fabrication plant and then bomb the ones who build for your competitor; instant monopoly. :-P

Re:bean counters ruin another company

[knightghost]

I really don't see that. Manufacturing in the USA typically runs Lean and often Cell based with process changes made in minutes. The people also tend to have a wider range of skills and experience. The states with unions pretty much don't do any more manufacturing.

Outsourcing only works from 2 fundamentals - ignoring IP (theft) and currency manipulation.

Re:bean counters ruin another company

[fuzzyfuzzyfungus]

They don't offer any GPUs not packaged with CPUs; but I'm working on the assumption that the same basic GPU block could be turned into a discrete GPU if Intel thought that it was worth doing.

Re:bean counters ruin another company

[bug1]

I'm thinking of that happy little picture of the world with a circle over portions of South and East Asia and the large islands in the Indian Ocean where half of the population of the planet lives.

Madagascar ?

I dont see the connection, maybe i need to research geographic sterotypes...

Re:bean counters ruin another company

[LWATCDR]

Simple gamers do not buy new cards until you can get new cards.
No one buys new cards until there are new cards to buy...

Re:bean counters ruin another company

[houstonbofh]

Outsourcing only works from 2 fundamentals - ignoring IP (theft) and currency manipulation.

You left out government regulation, taxes, and pollution controls.

Re:bean counters ruin another company

[tlhIngan]

Who didn't see not having their own fabs was going to bite them in the rear?
Only a bunch of bean counters would not have seen this coming.

Fabs are expensive. As in, REALLY expensive. They cost $1B to build, and easily $10+B to equip (or more). So you're already in the hole many billions of dollars and you haven't produced anything yet.

Oh, and there's a clock ticking away, because you're going to need to spend another $10+B to buy ALL NEW EQUIPMENT to handle the upcoming node shrink.

The only way to do that is to have a fab running continuously churning out product, and to find uses for the older node equipment (which now has to be discounted some because people want to use the new shiny equipment).

Fabs like TSMC however have been major innovators because it allowed fabless companies like NVidia, Broadcom, Qualcomm, and hundreds more to exist without having to make major capital investments in fabs but get the technology out there. Imagine you want to do a chip, and now you need to convince your investors that to make it, you need to spend $15+B. Versus a few million.

Re:bean counters ruin another company

[hendrips]

Considering the astonishing rate that AMD was losing money on its fabs, and the fact that upgrading a single fab to a new process node would cost more than AMD's entire market capitalization, I'm going to have to side with the bean counters here.

AMD coundn't even keep its production facilities running. How could they possibly have kept up with TSMC - the world's premier foundry operator?

Re:bean counters ruin another company

[bws111]

So Applied Materials is lying when it says:

Applied Materials, Inc. has been recognized as one of 18 companies receiving Intel Corporation's Preferred Quality Supplier (PQS) award for their performance in 2013. Applied Materials demonstrated industry-leading commitment across all critical focus areas on which Intel suppliers are measured: quality, cost, availability, technology, customer service, labor and ethics systems, and environmental sustainability. Applied Materials is recognized for their significant contributions providing Intel with wafer fab capital equipment, mask capital equipment, fab automation software, and services, deemed essential to Intel's success.

Because that kind of make is sound like Applied Materials equipment is 'essential to Intel's success'.

Re:bean counters ruin another company

[SydShamino]

Also, exploiting cultural differences. You won't get many Americans willing to live and eat in a dorm attached to the manufacturing plant, 800 miles from family, so they really have no distractions but work. (And of course you mention regulation, but having those workers for 12 hour shifts, 6 days a week, for 50 weeks straight can't hurt the bottom line - just the workers.)

Re:bean counters ruin another company

[SydShamino]

If NVidia build a fab, it would be great for them if/when their fab had cutting-edge processes and NVidia chips were the most profitable thing to run. But... what happens when their fab is out of date? NVidia chip designers would likely be forced to design for the NVidia fab anyway, and their hardware would fall behind. Or... what happens when their fab is updated? If they are one of the few on a new process, assuming they aren't sued by Intel for patent violations, should the NVidia fab lose out on potential revenue by building NVidia chips instead of more profitable Apple or Samsung chips?

Basically, if you aren't big enough and so far ahead of everyone else to keep all your own equipment running (i.e. Intel), nor are you able to contract in work (i.e. TSMC, GlobalFoundries), you aren't going to be successful with your own fab any more. You're either holding back your design shop, or holding back your fab, and either way you're not making as much profit as your competitors, which means you can't invest as much into your design shop or your fab.

Re:bean counters ruin another company

[Guspaz]

ATI never had their own fab. They were always fabless.

Re:bean counters ruin another company

[Anonymous Coward]

uninformed opinion here but:

Intels GPUs are integrated to the CPU, so any extra space allocated to more cores in the GPU takes away space available to the CPU.

Not to mention any thermal budget for the GPU shares the CPU's thermal budget.

Their highest end integrated GPUs aren't bad, but I just cant see them truly competing with equivalent generation dedicated units, and it probably comes down to thermal budget, if nothing else.

Re:bean counters ruin another company

[Guspaz]

Intel's integrated GPUs are pretty decent these days. At this point, they tend to outperform lower end discrete nVidia/AMD GPUs of a similar generation. Intel's Iris Pro 5200 GPU, for example, performs slightly faster than an nVidia GeForce 640. Each generation has seen Intel's chips creep slightly higher up the nVidia/AMD product line: it's not long since they creeped past the x30 tier of nVidia GPUs.

Re: bean counters ruin another company

[Anonymous Coward]

They tried to build one a few years ago but canceled it because it wasn't competitive.

Re:bean counters ruin another company

[dinfinity]

Don't be silly. It's not as if GlobalFoundries is putting AMD at the back of the line.
They just haven't been able to keep up with TSMC technologically (GF couldn't promise to deliver 20nm or 16nm when TSMC could), and they probably wouldn't have been able to if they were still a part of AMD.

The only big difference is that it would have been a pretty hard sell to have your GPUs be produced by a third (technologically more advanced) party instead of in your own (struggling) fabs.

Re:bean counters ruin another company

[blackomegax]

The Iris Pro is a competent gaming chip, but it only matches AMD's iGPU at best. And yes, DDR3 is a severe limitation. As are Intel GPU drivers. (garbage) But as silicon goes, they are actually good GPU tech and plow through OpenCL.

Re:bean counters ruin another company

[edxwelch]

Ha ha. You really think owning a fab is advantagous?
AMD did have it's own fab and the massive capital expenditure to keep that going nearly drove them to bankruptcy. Neither AMD nor Nvidia have the volumn to justify owning their own fab.
The fabless model works just fine - you will note that Apple and Qualcomm, the companies getting first tabs on 14/16nm, are both fabless companies.

Re:bean counters ruin another company

[ArmoredDragon]

I'll let somebody more experienced on it do the talking:

http://theweek.com/article/ind...

(And no, I'm not a fan of him, just I don't see any reason why he'd be wrong on this.)

Re:bean counters ruin another company

[edxwelch]

Ha ha. You really think owning a fab is advantagous?
AMD did have it's own fab and the massive capital expenditure to keep that going nearly drove them to bankruptcy. Neither AMD nor Nvidia have the volume to justify owning their own fab.
The fabless model works just fine - you will note that Apple and Qualcomm, the companies getting first tabs on 14/16nm, are both fabless companies.

Re:bean counters ruin another company

[sjames]

Point one can be addressed in negotiations. I'll bet a pay bump on days where a reconfiguration is required would get the job done. As for point 2, if there were jobs here calling for the skill, we would soon have workers with that skill available.

Re:bean counters ruin another company

[Anonymous Coward]

There are other companies out there. In terms of large ones, Matrox comes to mind. They are well-known for embedded graphics and driving many monitors. They seem to concentrate more on 2D performance than the 3D stuff, though.

I also would not discount Intel. Intel has effectively pushed the minimum entry level card to higher specifications, especially on laptops. A good graphics card from 3 years back can be outperformed by a modern CPU with integrated graphics.

With a price-performance comparison, I am not sure if consumers as a whole will be willing to shell out premium dollars without getting a high performance delta.

Re:bean counters ruin another company

[TehZorroness]

I really don't see that. Manufacturing in the USA typically runs Lean and often Cell based with process changes made in minutes. The people also tend to have a wider range of skills and experience. The states with unions pretty much don't do any more manufacturing.

I don't know if all shops work this way, but I work as a CNC lathe toolsetter and programmer in a non-union shop in New Jersey and what you say is pretty accurate based on my experience. Most of our operators will typically run a cell of 2 to 4 machines. As long as there are more contracts than we have machines (which is always, or we'd be out of business) we are constantly breaking down setups and retooling our machines for the next production run. We produce hundereds of different parts for some of our clients, and print revisions happen somewhat regularly. When we get a revised print, it usually only means we have to change a couple numbers in the g-code to define the new toolpath and tool the machine exactly how we did before the revision. While it may present an oppertunity to re-negotiate, It is hardly something we are willing to loose a contract over.

Now, I'm sure making microprocessors is quite a bit more sophisticated, but I can imagine that the biggest difference between different model CPUs built on the same process would be the code controlling the machine (automagically generated by CAM software from a CAD model), and they can probably be switched without making significant physical changes to the machine itself. If I remember correctly, it can easily take a couple years of calibration before a fab can produce anything reliable consistantly. I imagine once those machines are set up, they probably spend most of their time worrying about an earthquake happening on the other side of the world, not about loading a new program onto it.

- Chinese companies have the capability of rapidly adjusting manufacturing processes as a result of last minute design changes. While technically US companies have this ability, most companies just won't do it (in some cases labor unions are the biggest hindrance because they only permit their members to do one job and one job only, and instead of re-allocating existing labor, they're forced to hire new people, which just isn't economical or practical.)

- Even though it is possible to find the required skill set in the US, often the workers you do find aren't as good at a particular task as some people who live overseas and do that kind of work all the time. For example, how many Americans do you know that are good at operating the machinery used for making textiles? Chances are, they're harder to find than in China, but if you really wanted to get it done here, you could, just you'll pay more, it'll take longer, and the craftsmanship probably won't be as good.

I'd never trust any of this crap. All of these captains of industry complain about how few skilled workers we have here in the US - be it tech, manufacturing, whatever. All they really want is to pay as little for the labor as possible, and as long as it costs less in China, they will keep making up excuses and funding think-tanks to support their point of view. If they were trying to have their products manufactured with the best craftsmanship possible, by the most skilled workers they could employ, they wouldn't be outsourcing them to the cheapest place they could find. I'm not going to be that guy that says we do everything best in the US, but doing things best is not why we are outsourcing to China either.

Re:bean counters ruin another company

[Anonymous Coward]

The Ivy Bridge 3D accelerator in my Thinkpad X230 is surprisingly capable; It's able to play lots of older games at least as well as my desktop nVidia GeForce 7950GT AGP.

I've run WH40k: Dawn of War and Space Marine, Jade Empire, Star Point Gemini 2, Back to the Future, Antichamber, Neverwinter Nights 1 and 2, Deus Ex 3 and Dragon Age Origins and they're all very playable.

This is at native res (1366x768), although usually with detail settings dropped down some.

I do wish Lenovo would do a version of this laptop with an AMD APU to see how they compare, but I know on my old laptop (X121e) the Intel variant was rated a lot faster than the APU-equipped AMD version despite the APU's 3D capabilities being a lot better on paper. (To be fair, the CPU part of the brazos APU was massively underpowered compared to the i3's CPU part)

Re: bean counters ruin another company

[EETech1]

Or... They worked so well some(TLA)body got them all, while Intel tweaked it into knights corner / knights landing / xeon phi, where they now use it to dominate the top 500.

Re:bean counters ruin another company

[Anonymous Coward]

Not that there's anything wrong with that.

Re:bean counters ruin another company

[Fwipp]

You know that bombing factories is, like, illegal, right? And that most people have, you know, ethics?

Re:bean counters ruin another company

[Solandri]

If you don't control your own production and aren't willing to pay more for it than anyone else, don't be surprised when you have to wait in line.

FTFY. nVidia and AMD could easily jump ahead of Apple and Samsung in TSMC's queue - they just have to offer to pay more than Apple and Samsung are willing to pay. All that's happening is that they've made a conscious decision that it's more cost-effective to pay less and have their 16nm/20nm products come out later and deal with unhappy customers, than to pay more to make their customers happier.

Re:bean counters ruin another company

[Guspaz]

Agreed, nothing wrong with that. But it was highly disappointing when they got bought out by a foreign company.

Re: bean counters ruin another company

[ihavnoid]

Except that in this case they will be quickly out of business if they actually tried to build a fab, which is a 10 billion dollar investment which will go obsolete in 2~3 years.

The fundamental problem is that there isn't enough demand for discrete GPUs, so they are getting a hard time hitting the economy of scale. What I see is that mobile chips are getting the latest and greatest process technologies these days - I believe the market size of mobile processors is order of magnitudes larger than discrete GPUs, so good luck outbidding them.

Re:bean counters ruin another company

[UnknownSoldier]

This still applies. /Olbg. Intel Santa GPU
http://www.dvhardware.net/news...

Through it should be said ...

"Debunking the 100X GPU vs. CPU Myth: An Evaluation of Throughput Computing on CPU and GPU"
http://sbel.wisc.edu/Courses/M...

Re:bean counters ruin another company

[unixisc]

The fabless model works fine provided 2 things happen:

1. The customer (AMD/Nvidia/Apple/Qualcomm) are accurate in forecasting what they will need, and have a consistent demand at the fab, not varying horrendously, as many smaller players do

2. The customer is not dwarfed by other bigger customers during times of allocation. When market demand is high and semiconductor prices go up, then fabs usually pick their top tier customers, have them allocate which products need to run uninterrupted, and do that. Smaller customers fall by the wayside.

Chances are neither Apple nor Nvidia would face that predicament. I'm not so sure about AMD.

Re:bean counters ruin another company

[gstoddart]

LOL, yes on both, and I wasn't advocating it ... I was comparing it to a Bond Movie because it sounds so damned crazy.

Re:bean counters ruin another company

[edxwelch]

AMD have an agreement with globalfoundries, so it may not be as bad as you think

Re:bean counters ruin another company

[Anonymous Coward]

AC since this obviously is a small world.

My father was part of a team that bought a fab. Not a new one, but one that had been cutting-edge when it was built a few years before. Clearly, this is the kind of purchase that goes all the way to the top of the company. Even for the largest companies, the CEO and CFO personally will be involved in such investments. Building a new cutting-edge fab is only going to be more expensive and stressful.

And at that point, you will put a price on risk. The bean counters will have estimated the upsides, the downsides, the estimated profit/loss, and yes they will have seen that the expected profits of having a fab are still positive. But the downside is not just losing money. If that fab fails, the company fails, for all but the largest players (i.e. Samsung and Intel). Even TSMC and GF will collapse if they'd lose one cutting-edge fab.

Re:bean counters ruin another company

[Agripa]

Point one can be addressed in negotiations.

Which is just another cost.

if there were jobs here calling for the skill, we would soon have workers with that skill available.

If there were jobs here calling for that skill at an appropriate wage and if the infrastructure existed to support that skill.

Re:bean counters ruin another company

[Agripa]

They only way I see them getting bitten is if Intel decides they really want a slice of the high end GPU market; ups their game on the design side AND allocates their 14nm facilities to GPUs in favor of cranking out more CPUs.

Intel does not need to worry about the high end GPU market which is discrete; increasing integration will eventually allow integrated GPUs to catch up to and outperform discrete GPUs at which point Intel will only be competing with AMD assuming that AMD still exists. Unless x86 is displaced by ARM, nVidia will not be a factor because they lack an x86 CPU so they have been diversifying away from the x86 GPU market.

Re:bean counters ruin another company

[sjames]

Which is just another cost.

So doing something extra costs extra? You don't say!

Didn't AMD used to have fabs?

[Anonymous Coward]

I might be misremembering history, but I thought at one point AMD did have fabs.

Re:Didn't AMD used to have fabs?

[Fwipp]

Yes, they spun them off into Global Foundries.

Re:Didn't AMD used to have fabs?

[Anonymous Coward]

...then GF spun out of AMD's control and couldn't deliver a competitive process anymore. Then AMD went full retard and rushed to release an APU. That lead to a product with terrible IPC and integrated graphics that were still far too slow to be considered revolutionary. Then AMD became even more broke than it already was. Then AMD failed to compete by delivering low power chips because they have no fab to do R&D. A series of completely unpredictable events.

Re:Didn't AMD used to have fabs?

[Anonymous Coward]

In 2011, AMD's laptop APUs were revolutionary. You could not buy that kind of performance and low power anywhere else.

In 2014, AMD's laptop APUs are still better than the alternatives, but you can not buy that kind of performance and low power anywhere except for HP at a significantly higher price than the 2011 version.

I wonder why?

AMD sold fabs in 2008 right?

[Anonymous Coward]

http://www.forbes.com/2008/10/10/cio-amd-fabs-tech-cio-cx_es_1013amd.html

capitals ?

[swell]

" AMD, Nvidia Reportedly Tripped Up On Process Shrinks "

Is there some good reason to start every word with Caps?

Perhaps it would be even more helpful to title stories this way:

" AMD, NVIDIA REPORTEDLY TRIPPED UP ON PROCESS SHRINKS "

  - Or maybe you should just use caps where needed.

Re:capitals ?

[Anonymous Coward]

Uh, yeah. It is standard practice to capitalize every word of a headline. Since forever. OK, maybe 'on' should not be capitalized, but that is the only one.

Re:capitals ?

[Anonymous Coward]

>Is there some good reason to start every word with Caps?

Title case.

Re:capitals ?

[bigdady92]

Because AMD is an abbreviation of Advanced Micro Devices (AMD) therefore it's proper English to capitalize all the letters.

You have a low enough user account number to know that.

Re:capitals ?

[Anonymous Coward]

Lol, try again

Re:capitals ?

[swell]

Have a look at Google news where headlines from many newspapers and journals are reproduced. You won't see many publishers capitalizing every single word.

"Proper English"? Your journalism text is from a different century. My copy of The Associated Press Stylebook (2005), says this: "*capitalization* In general, avoid unnecessary capitals. Use a capital letter only if you can justify it by one of the principles listed here." - I cannot find an exception for headlines.

Re:capitals ?

[SydShamino]

AP style is to only capitalize the first word and proper nouns

https://www.apstylebook.com/?d...

search for "News media" and read the questions that follow

Re:capitals ?

[Anonymous Coward]

The Associated Press Stylebook applies to the Associated Press. Every publication has a house style.
If you've been around here for any length of time, you know that Slashdot's house style includes dupes, slashvertisements, and traditional title case for article headlines.

Re:capitals ?

[Anonymous Coward]

" AMD, Nvidia Reportedly Tripped Up On Process Shrinks "

Is there some good reason to start every word with Caps?

Perhaps it would be even more helpful to title stories this way:

" AMD, NVIDIA REPORTEDLY TRIPPED UP ON PROCESS SHRINKS "

- Or maybe you should just use caps where needed.

Caps, yah, lotsa caps thr
wht I dnt get is wht's teh n3d fr all thse vowls

Re:capitals ?

[Anonymous Coward]

Both forms are acceptable, all of you need to get over yourselves. It is NOT important.

http://en.wikipedia.org/wiki/Letter_case

In English-language publications, varying conventions are used for capitalising words in publication titles and headlines, including chapter and section headings. The rules differ substantially between individual house styles.

The convention followed by many British publishers (including scientific publishers, like Nature, magazines, like The Economist and New Scientist, and newspapers, like The Guardian and The Times) and U.S. newspapers is to use sentence-style capitalisation in headlines, where capitalisation follows the same rules that apply for sentences. This convention is usually called sentence case. It may also be applied to publication titles, especially in bibliographic references and library catalogues. Examples of global publishers whose English-language house styles prescribe sentence-case titles and headings include the International Organization for Standardization.

As regards publication titles it is, however, a common typographic practice among both British[6] and U.S. publishers to capitalise significant words (and in the United States, this is often applied to headings, too). This family of typographic conventions is usually called title case. For example, R. M. Ritter's Oxford Manual of Style (2002) suggests capitalising "the first word and all nouns, pronouns, adjectives, verbs and adverbs, but generally not articles, conjunctions and short prepositions".[7] This is an old form of emphasis, similar to the more modern practice of using a larger or boldface font for titles.

Production

[fxsoap]

Interesting that they both use the same supplier for their critical component and are competing products....I'm guessing neither have enough money to build their own production labs like Apple did with that special glass they use. Imagine how amazing they would be as a joint company.

Re:Production

[slew]

Interesting that they both use the same supplier for their critical component and are competing products....I'm guessing neither have enough money to build their own production labs like Apple did with that special glass they use.

Apple didn't spend their own money on production of their "special" glass (it is purchased gorilla glass 4 from Dow Corning)...
On that whole GT advanced technologies sapphilre disaster, they attempted to purchase their own production labs (and lease them back to GT for production), but apparently that ain't gonna happen now...

Imagine how amazing they would be as a joint company.

Actually Nvidia, Apple, AMD, Qualcomm and Xilinx are major customers of TSMC's advanced processes and all are subject to the whims of the supply and demand for wafers at TSMC. It's just that Apple is a bigger and newer customer and generally customers in that position get preferential treatment to win their business. FWIW, at a capital cost of about $2B, I don't think Nvidia and AMD are going to get back into the fab biz any time soon. Then their is the whole poly silicon suppliers and wafer suppliers and Applied Materials steppers, etc, etc... Companies are all highly independent on specialized suppliers. This is not unsual in this business (or any other modern enterprise). Very few companies are totally vertically integrated these days...

Re:Production

[hendrips]

Hah - $2B would be an optimistic number just for upgrading an existing fab. A brand new fab would be somewhere in the $5B to $10B range, judging by what Intel and TSMC are spending.

Re:Production

[alexander_686]

Apple didn't spend their own money on production of their "special" glass (it is purchased gorilla glass 4 from Dow Corning)...
On that whole GT advanced technologies sapphilre disaster, they attempted to purchase their own production labs (and lease them back to GT for production), but apparently that ain't gonna happen now...

That is not true. Apple has invested billions in Corning, Samsung, Foxconn, etc. You have only heard about GT because it blew up.

It is not uncommon for a customer to help finance a vendor's factor. Why should a company (let's day Corning) build a huge, risky, cutting edge factory? It is a huge risk. Why should a company (let's say Apple) design a product around cutting edge technology that might not be available in mass quantities?

So Apple pitches in cash to finance the new factory and in return they get preferred pricing / exclusive rights of the output for 6 to 12 months.

As an example, Tesla's "Gigafactory" is actually a Panasonic factory.

Re:Production

[slew]

That is not true. Apple has invested billions in Corning, Samsung, Foxconn, etc. You have only heard about GT because it blew up.

Although I only have access to public information, AFAIK, it isn't Apple that invested in Corning, actually it is Samsung that invested in Corning (~7.4% stake which makes Samsung the largest single shareholder of Corning).

Although in the past, Apple had made an investment in Samsung Semiconductor to ensure flat panel technology availability, I believe they no longer use Samsung flatpanels nor hold that investment.

For their semiconductors, their relationship appears to be a foundry deal only. Samsung has made investments themselves on fab capacity based on being an Apple supplier. Of course as industry practice, Apple likely makes component pre-payments based on forecast demand on a discount basis (which gives Samsung funds to finance it), but that's not the same as an equity investment.

On the other hand, Apple apparently has invested a few billion in Sharp as their new flat panel supplier. There is also a rumored multi-billion dollar Foxconn investment (for expanding iphone production capacity), but I can find no public evidence of a Corning investment.

That's not to say that Apple and Corning don't collaborate very closely on gorilla glass production, but there doesn't appear to be this billion dollar investment in Corning that you are referring to.

Re:Production

[alexander_686]

Of course as industry practice, Apple likely makes component pre-payments based on forecast demand on a discount basis (which gives Samsung funds to finance it), but that's not the same as an equity investment.

When I was talking about "investments" I was actually referring to this type of arrangement. Not all investments have to be equity. Some can be simple short term loans to finance inventory and production. That being said, some of the arrangements can be huge, complex, and multi-year and can get pretty near the line of equity. But if not the marriage of equity at the very least it is a very deep partnership of cohabitation. See GT as an example. Of course, the accountants make sure it never crosses the line into equity – that tends to open a can of worms - and often it must be publicly disclosed.

As for Corning I was a bit lazy when I said that Apple invested billions in Corning – I don't know the exact amount or level of involvement. I do know that Apple did help Corning with cash in order to ramp up production in exchange for a limited time exclusive on the glass.

Actually...about Intel...

[fuzzyfuzzyfungus]

I certainly wouldn't expect to see it happen(well, maybe with a very low probability); but it wouldn't surprise me if someone at Intel Legal has written up an "AMD/Foundry Contract Opinion.doc" and squirreled it away somewhere.

Given that AMD isn't terribly threatening anymore, we aren't in the Netburst vs. A64 beatdown era now, Intel is probably saved a fair amount of unpleasant antitrust inquiry(US and abroad) by AMD at comparatively limited cost in product margins or lost design wins. If it came to it, selling them foundry services would probably be preferred to letting them die.

Re:Actually...about Intel...

[bill_mcgonigle]

it wouldn't surprise me if someone at Intel Legal has written up an "AMD/Foundry Contract Opinion.doc" and squirreled it away somewhere.

Or they can take advantage of the situation to acquire nVidia's market position in GPU computing and patent arsenal (to annihilate the trolls - probably not gunning for AMD). Intel wants a viable AMD, to keep DoJ off its back. nVidia is just a market competitor with a manufacturing problem.

Re:Actually...about Intel...

[Anonymous Coward]

I doubt Intel would help AMD, maybe Nvidia. Look up history, AMD only exists as a company to do fab work for Intel designs originally. If Intel had enough fab ability in the 80s, AMD wouldn't exist today. This is the textbook definition of irony.

Re:Actually...about Intel...

[edxwelch]

No, that's not it.
Intel said that they will provide foundry services for any company willing to pay the price, no exceptions.
The reason Intel is a non-option for AMD and Nvidia is, obviously, the price is too high.
Intel are accustomed to enjoying a 65% margin and that's the way they want to keep it. Brian Krzanich publicly stated this at the Citi 2014 Global Technology Conference in so many words.

No...

[YuppieScum]

NVidia has always been fabless, but AMD owned its own fabrication plants until a few years ago... when they were spun off into a separate company called Global Foundries.

Goodbye AMD

[Anonymous Coward]

Your shortsighted business practice of selling off your foundry has just killed you. It's gonna suck having Intel as the monopoly for the X86 CPU market.

Re:Goodbye AMD

[serviscope_minor]

Your shortsighted business practices of getting fucked over by Intel's almost completely unpunished illegal monopoly abuse has just killed you.

FTFY.

Re:Goodbye AMD

[arbiter1]

o please, intel has superior cpu, straight out. AMD hasn't had a decent cpu since athlon xp/64. Blaming intel for AMD failure to innovate their cpu is straight up retarded. Intel has the market share they do cause they have made cpu's people want, not slow power hungry piles of silicon.

Re:Goodbye AMD

[Carewolf]

o please, intel has superior cpu, straight out. AMD hasn't had a decent cpu since athlon xp/64. Blaming intel for AMD failure to innovate their cpu is straight up retarded. Intel has the market share they do cause they have made cpu's people want, not slow power hungry piles of silicon.

Go back to the time period we are talking about. The fabs were sold off in 2008 which was when AMD was still a top dog performance wise, but losing money due to Intel illegal deals with OEMs.

Re:Goodbye AMD

[towermac]

AMD not having a superior cpu since Athlon is the direct result of Intel's illegal monopoly abuse.

Didn't this just happen? Or maybe you are very young...

Re:Goodbye AMD

[unixisc]

How? Since AMD came out w/ a 64-bit x86 that was endorsed by Microsoft, they had an advantage over Intel, and did a cross-licensing of their own 64-bit vs Intel's 32-bit instruction sets. Intel hardly had any advantage over AMD there.

The advantage Intel had over EVERYONE - AMD, HP, DEC, IBM, Sun, MIPS, you name it - was fabs!!! They are 2 generations ahead of AMD, and were at least 1 generation ahead of everybody else. In case of HP, they used to fab their PA-RISC, before Itanic was out. AMD, being 2 generations behind on process shrinks, couldn't be expected to keep up w/ Intel, who could simply toss more cores at any problem, since every OS out there now is either NT-based or UNIX-based (meaning SVR4, BSD, Linux, everything). So that's what Intel did, and won this game.

I am sad at not seeing AMD chips in MS Surface Pros and other toys, but that's not something that Intel illegally contrived at AMD's expense.

Re:Goodbye AMD

[serviscope_minor]

I am sad at not seeing AMD chips in MS Surface Pros and other toys, but that's not something that Intel illegally contrived at AMD's expense.

When the A64 came out it stomped all over the P4. Despite Intel's fab advantage, te A64 was faster, cooler and cheaper than the P4.

At that point, Intel outright bribed Dell and other large vendors to NOT use AMD chips. That is flat out illegal and denied AMD a huge amount of revenue when they most needed it to capitalise on their advantage with intel.

So despite having a big advantage in superior tech, Intel used illegal tactics to starve AMD of cash for as long as it took for intel to catch up.

Re:Goodbye AMD

[unixisc]

The only company that had a policy of not buying AMD was Dell - everyone else at the time - Compaq, HP, even Gateway was okay w/ buying AMD. And at the time, AMD was catching up, while Intel was still obsessed about the Itanium and their desire to kill off the x86 architecture at 32-bit. It wasn't until a bit later that they too embraced the AMD path, when Microsoft made it clear to them that they wouldn't be supporting a separate 64-bit architecture different from what they had already done for AMD. The reason that AMD went on to fall flat on its face was that it never was good at maximizing yields even from their own fabs

Re:Goodbye AMD

[serviscope_minor]

The only company that had a policy of not buying AMD was Dell - everyone else at the time - Compaq, HP, even Gateway was okay w/ buying AMD.

No: Intel also bribed HP to cap AMD sales at 5% of the total. They bribed AMD as well. Intels bribery severely hapmered AMDs sales. Intel has had to pay out billions for this wrongdoing. However they may as well treat that as a successful investment since it completely stalled AMD when they were beating Intel.

So, money well spent, if you have no ethics.

no way???

[sribe]

The only other manufacturer with 14nm capacity is Intel and there's no way Intel will sell them some capacity.

Riiiight. Because Intel never ever sells fab capacity.

Oh wait, they started doing that in 2010. Oops.

Re:no way???

[drinkypoo]

Riiiight. Because Intel never ever sells fab capacity.

AMD going to Intel for fab capacity is like the Palestinians going to the Israelis for rocket technology.

Re:no way???

[Anonymous Coward]

>Palestinians going to the Israelis for rocket technology.
they were selling weapons to Syrian rebels up until recently.

Re:no way???

[Anonymous Coward]

Your analogy is largely invalid. AMD notably receives a lot of benefits from Intel, because of antitrust lawsuits (in part of course because Intel abused moral and the law), and has been for quite some time. The problem is precisely that no one really does anything about Isreal governement decisions, and it abuses this situation all the way, without significant consequences...

Re:no way???

[Lunix Nutcase]

Riiiight. Because Intel never ever sells fab capacity.

The sentence doesn't say that. It says it wouldn't sell AMD or Nvidia, the companies referred to by the "them" in that sentence, some of their capacity.

Re:no way???

[Anonymous Coward]

Every respective arms dealer supplies both sides. Winners are just as bad for business as losers.

Re:no way???

[Anonymous Coward]

Entertaining Story.

Intel has major ties to Israel; they have fabs there as well as Major Engineering Departments and management.

AMD's major stakeholders include Abu Dhabi, of the United Arab Emirates.

That statement holds a lot more truth than you might think.

Re:no way???

[Anonymous Coward]

AMD going to Intel for fab capacity is like the Palestinians going to the Israelis for rocket technology.

Israel sends them rocket technology daily. WHOOSH!

Re:no way???

[Anonymous Coward]

Riiiight. Because Intel never ever sells fab capacity.

AMD going to Intel for fab capacity is like the Palestinians going to the Israelis for rocket technology.

I frequently do business with my compeditors. A dollar is a dollar. We don't care whose dollar it was, if we perform the contractual obligations, it becomes our dollar. I am always happy to take their money and we treat them fairly just like any other customer- so they come back. And they frequently do. We have decades-long productive relationships with most of our competition. If our deals weren't beneficial for both companies we wouldn't continue them- which leads me to conclude that a company refusing to do business with the competition is leaving money on the table.

Re:no way???

[unixisc]

The only other manufacturer with 14nm capacity is Intel and there's no way Intel will sell them some capacity.

Riiiight. Because Intel never ever sells fab capacity.

Oh wait, they started doing that in 2010. Oops.

Do they sell capacity in their most advanced process? I'd imagine that they sell capacity in their depreciated fabs w/ older processes. However, since Intel is so far ahead of others, even that is a major boost for customers (like Altera), and b'cos Intel has some of the best practices in handling fabs, they do know that they'll get quality there w/ it! But I wouldn't be surprised at Intel reserving its most advanced processes for things like Xeons or i7s

Re:no way???

[unixisc]

Precisely this!!! When I worked in the semiconductor business, we had Samsung as a customer, and their flash memory division was our competitor. When we asked about why they preferred us, reason was what any other customer might give - we had a better cost effective solution. Intel's flash too was our competition, but when Intel made their i740 graphics cards back in the 90s, our flash was on their reference design as well. There is no reason for any company to be religious about not using a competitor's part if that provides them what they want. That is somewhat different from the Israeli-Pali analogy above, where Israel giving rockets to Hamas would inevitably result in Hamas firing those from Gaza into Sderot

Re:no way???

[Anonymous Coward]

Israel? No! Israel is pretty much neutral in the Syrian Civil War, and views both sides as enemies - be it Hizbullah, Free Syrian Army or ISIS

Re:no way???

[sribe]

Do they sell capacity in their most advanced process?

They have for prior nodes, though not at first when it's brand new.

I'd imagine that they sell capacity in their depreciated fabs w/ older processes. However, since Intel is so far ahead of others, even that is a major boost for customers (like Altera), and b'cos Intel has some of the best practices in handling fabs, they do know that they'll get quality there w/ it! But I wouldn't be surprised at Intel reserving its most advanced processes for things like Xeons or i7s

I've read that their announced goal for 14nm capacity will be more than all other fab capacity worldwide combined. If true, that seems to indicate aggressive expansion into selling capacity.

Re:no way???

[bluegutang]

If Israel sold rockets to the Palestinians, the rockets would be used against Israel and Israelis would die. You don't get your citizens killed for money.

If Intel sold fab capacity to a competitor, it might hurt Intel's own CPU sales. But lost money can be made up for with other money. For the right price, why wouldn't Intel do this? Particularly as Intel has an interest in keeping companies like AMD alive to avoid antitrust action, while Intel is looking for customers for its custom foundry.

Re:no way???

[unixisc]

Abu Dhabi is a stakeholder in Global Foundries, not AMD per se

Re:no way???

[drinkypoo]

For the right price, why wouldn't Intel do this?

AMD can't afford the right price. That much of a shrink would make them directly competitive, assuming their processor can be shrunk that much. However, just doing that would teach AMD a lot about Intel's process technology...

Particularly as Intel has an interest in keeping companies like AMD alive to avoid antitrust action

Alive, sure. Competitive? Nah.

Build your own fab

[Karmashock]

Just one. Nothing out of control. Intel has something like a dozen of them. Just build one. Then you have the ability to expand your own capacity using Intel's model of "EVERYTHING EXACTLY THE SAME" where in Intel copies their fabs exactly down to the furniture.

Re:Build your own fab

[SuricouRaven]

Fab facilities are tremendously expensive. This isn't something you can throw together in your basement. A fab is going to cost at least a billion dollars - and that's not even for state-of-the-art 14nm stuff. TMSC's fab cost ten billion dollars to build. That's just the construction cost - semiconductor tech is constantly changing, so if you want to make the latest goodies like high-performance GPUs there's also the need to constantly puchase new and better equipment.

Re:Build your own fab

[confused one]

AMD had fabs. They had several. The cost of operating them was a heavy burden on their bottom line; so, they spun them off in 2009 (that would be Global Foundries) and decided to go fabless. Now, they're experiencing the consequence of being dependent on 3rd parties to do their manufacturing for them. TSMC is building more fab capacity -- they just can't get it online fast enough (it takes several years to build and qualify one of these fabs)

Re:Build your own fab

[Existential Wombat]

Just get a 3D printer and make your own chips. Jeez, cheapskates.

Re:Build your own fab

[Karmashock]

Intel maintains many of them. AMD is perfectly capable of maintaining ONE.

Re:Build your own fab

[hendrips]

As of September 2014:

-AMD's available cash: $950 million
-AMD's market capitalization: $2.6 billion
-AMD's credit rating: Absolute garbage
-Cost of a new Intel/TSMC style fab: $7 billion - $10 billion

It's a nice thought, but the reason that so many companies, including huge companies like Apple, IBM, and Qualcomm, have gone fabless is that fabs are astonishingly, mind-blowingly, expensive.

Re:Build your own fab

[Anonymous Coward]

Wouldn't they need two (at least)? I couldn't imagine they'd stop production for months to upgrade the fab for process shrinks every two years or less.

Re:Build your own fab

[brxndxn]

No, it isn't. The entire market cap of AMD is less than the cost of building a new fab. Intel royally fucked AMD in 1999-2003 and AMD has been in a dismal financial position ever since. Intel does not play fair - and until Intel is forced to play fair, the best you can expect from AMD is for them to stay alive. Intel was forced to pay a fine that was about 5% of where AMD should have been financially.

Re:Build your own fab

[Lunix Nutcase]

With what money do you expect them to build a fab?

Re:Build your own fab

[SuricouRaven]

The cost of upgrading a fab for process shrink is almost as much as building a new fab. They often don't get upgraded - they get downmarketed. A new fab is built for the process shrink, and the old fab is reduced to making lower-spec tech. All those little microcontrollers, glue chips, controllers and interfaces found in just about everything electronic may well come from the same fab that once churned out the state-of-the-art Pentium 3.

Re:Build your own fab

[Anonymous Coward]

[T]he reason that so many companies, including huge companies like Apple, IBM, and Qualcomm, have gone fabless is that fabs are astonishingly, mind-blowingly, expensive.

You're right, but even that really understates the difficulty. There simply aren't many engineers in the world who know the dark arts of how to make your mind-blowingly expensive fab actually get economically usable yields. Intel already has an established culture of engineers who know how to get it done, how to troubleshoot, and how to estimate which design risks are smart ones to take. A chip fab is not like some $10B printer that you just plug in and start churning out processors. If you only get 99.993% of everything right, your yield numbers are gonna be somewhere near zero. Even Intel, the masters, make new fabs as milimeter-exact replicas of one another, because there are so many thousands of variables to be controlled for.

Re:Build your own fab

[Karmashock]

That's fine. Have ONE fab. Don't sell literally every single one of them.

Re:Build your own fab

[Karmashock]

They're a multi billion dollar corporation that used to have their own fabs... not a collection of bag ladies.

Give me a motherfucking break.

Re:Build your own fab

[Karmashock]

I'm not suggesting that they rely on a their own fab but rather keep it operational so they have some logistical flexibility.

If global foundries are not giving them enough time in their fabs they can instead invest more in their own.

Re:Build your own fab

[Karmashock]

Good game for amd then. Guess they should just stick a shot gun in their mouths and pull the trigger with a toe.

Or man up and find a solution to the problem.

What you are saying is that given the current status quo, AMD cannot climb out of this box. This I agree to and understand.

However, the whole history of technological development as well as inter industrial gamesmanship is rife with examples of various companies leapfrogging the status quo by doing things differently.

AMD already tried something by going with this fabless model. The problem with outsourcing however as we all know is that you lose control. Control is probably the most underrated feature period in the 21st century. Very few people seem to grasp how critical control is to the stability and sustainability of nearly any enterprise.

You note that a big fab of the kind Intel builds costs too much for AMD to afford. Granted. Do something else. Possibly invest in a new fab with some other investors that likewise want more fab time. Or possibly consider alternative manufacturing models or systems.

The alternative appears to be biting down on the barrel of the shotgun, closing their eyes, and spraying raspberry jam all over the walls.

I find that option to be distasteful. So any other option should be considered.

Re:Build your own fab

[Ash-Fox]

I'm not suggesting that they rely on a their own fab but rather keep it operational so they have some logistical flexibility.

AMD didn't even make enough money to keep one of their fabs, never mind keep it up to date...

Re:Build your own fab

[unixisc]

Fab facilities are tremendously expensive. This isn't something you can throw together in your basement. A fab is going to cost at least a billion dollars - and that's not even for state-of-the-art 14nm stuff. TMSC's fab cost ten billion dollars to build. That's just the construction cost - semiconductor tech is constantly changing, so if you want to make the latest goodies like high-performance GPUs there's also the need to constantly puchase new and better equipment.

Not just that, TSMC needed financing from the Taiwanese government to make those fabs. Intel is fortunate that they can afford it on their own

Re:Build your own fab

[unixisc]

Uh, when AMD came out w/ AMD-64 extensions to the x86 instruction set, they had an advantage over Intel, and had a cross licensing deal. For a while, they even came out w/ better CPUs, when Intel was struggling w/ Pentium 4, Netburst and Itanium I. Reason AMD fell behind since is that Intel does a lot better job in building and maintaining their fabs, and funnels a lot of their margins into these new fabs. AMD never really figured out how to successfully run fabs and keep up w/ demand, and ultimately, they had to go fabless.

Re:Build your own fab

[Karmashock]

Exactly how is Intel not playing fair?

Re:Build your own fab

[Karmashock]

Fatalism is for losers.

Re:Build your own fab

[Ash-Fox]

Fatalism is for losers.

So, you're saying it's better to be a dead company than a loser?

Re:Build your own fab

[Karmashock]

I'm saying that any company that is dependent on a single supplier, unable to innovate, and ultimately left bidding against wealthier companies for scrap time on increasingly leveraged production lines is fucked.

So I am not saying that they kill themselves. I am saying that if they don't solve this problem they're already dead and just don't know it.

And when you're in that position... possibly some risk taking is warranted.

I repeat... fatalism is for losers. If you're between a rock and a hard place, then you can either sit there like a bug and get crushed. Or you can find a way out of that situation. Telling me it is hard etc is what losers say. They get crushed. You get out of that bind or die.

Re:Build your own fab

[Lunix Nutcase]

The cost of building and equipping a fab is more than AMD's entire yearly revenue. They don't have the cash or credit to take on such a huge debt.

Re:Build your own fab

[Karmashock]

Then they can't do it the way you are thinking about it.

But if they don't do something then they are in a very bad position.

I am not saying they have to do it the way YOU imagine that anyone must do it. Consider that there might be other ways. And really, that is an issue that could take a lot of time on the part of people that know a lot more then either of us to figure out.

Regardless, the point is that if they do not obtain some leverage they're fucked.

Re:Build your own fab

[Ash-Fox]

I'm saying that any company that is dependent on a single supplier, unable to innovate, and ultimately left bidding against wealthier companies for scrap time on increasingly leveraged production lines is fucked.

You seem confused. AMD's primary income is from selling research, not producing processors.

I am saying that if they don't solve this problem they're already dead and just don't know it.

They've always had the option to leave the microprocessor market entirely.

And when you're in that position... possibly some risk taking is warranted.

Except this still isn't their primary income, it is secondary.

I repeat... fatalism is for losers.

I repeat my question, because it really sounds like you're saying this. You're saying it;'s better to be dead than a loser?

Re:Build your own fab

[Karmashock]

It doesn't matter if they have other businesses. Their role in that market requires that they have the ability to produce products. If their lack of control over their supply chain is making their future in that market questionable then they need to make a choice.

And note... not making a choice is also a choice.

Re:Build your own fab

[Ash-Fox]

It doesn't matter if they have other businesses

Sure it does, because this is how you assess risk and determine where to focus your business.

Their role in that market requires that they have the ability to produce products.

Or exploit it for other purposes, such as selling off research, as they have done numerous times.

If their lack of control over their supply chain is making their future in that market questionable then they need to make a choice.

Tech markets tend to have a lot of various issues continuously which is reflected price changes. I remember other significant problems like when Taiwan got hit by a natural disaster and many harddrive companies were unable to produce harddrives in quantities desired due to component demands. This isn't really a different situation or something new.

Re:Build your own fab

[Karmashock]

Except even in your example that situation was not accepted or shrugged off. The industry has taken steps to decentralize the production of storage media as a result.

I am not suggesting that companies shouldn't run into problems or that problems are not normal.

Rather, I am saying that when companies run into serious problems they need to be taken seriously and addressed in a manner that they no longer pose a strategic risk to their position.

AMD clearly needs more leverage in the supply chain. How they get that is their business. But meekly waiting in line is for suckers.

Re:Build your own fab

[Ash-Fox]

Except even in your example that situation was not accepted or shrugged off.

AMD didn't shrug off their problems, they changed their firm to mitigate the risk.

Rather, I am saying that when companies run into serious problems they need to be taken seriously and addressed in a manner that they no longer pose a strategic risk to their position.

AMD's position in my opinion is 2nd best against vendors when it comes to processor/chips, this hasn't compromised that. They have other revenue streams, so they can survive a hit.

AMD clearly needs more leverage in the supply chain.

Their leverage is that they can exit the market and kill industries in the process. This however is equivalent to a nuclear bomb in my mind and shouldn't be used to fight some temporary resource contention issues which is apparently affecting the wider industry too. They're probably looking at alternatives right now, but I suspect the options aren't very desirable.

Re:Build your own fab

[Karmashock]

Which is the problem with nuclear weapons. Which military would you rather have... Some French ICBMs or India's conventional military with no nuclear weapons?

I'd take India's military any day. You can actually use it. Last resorts that you don't want to use are often no better then having no resort at all. All they let you do in many cases is put a stick in the eye of the person that killed you.

Look, I get your point... I do. Get mine... they should be able to produce their product smoothly without getting bumped out of line by Apple.

Intel would *love* to sell fab capacity...

[Anonymous Coward]

...for the right price. There's no way for AMD or nvidia to agree to that price.

Re:Intel would *love* to sell fab capacity...

[Lunix Nutcase]

Intel would *love* to sell fab capacity

Intel has no need to sell their fab capacity.

Shrinking your competitor's capacity

[BenJeremy]

It's the same reason we import oil in the US - use up the capacity elsewhere and horde your own.

Re:Shrinking your competitor's capacity

[ArhcAngel]

Actually oil companies have been exporting crude products abroad (They still can't legally export unrefined crude oil) for a few years now because they can make more selling them in other countries. In fact this past June some restrictions were lifted and they started exporting unrefined ultralight oil

Why I bought INTC

[Anonymous Coward]

Glad to here you all understand this now. This is why I bought Intel stock 2 years ago. INTC the best chip fab in the world. Ultimately Apple, AMD, and nVidia will lose. Oh yeah, and I will profit! ;)

Re:Why I bought INTC

[Anonymous Coward]

Arrogant asshole. Die in a fire.

Re:Why I bought INTC

[Anonymous Coward]

Ultimately Apple, AMD, and nVidia will lose.

You hopefully realize that only one of these three is competing directly with Intel. Hopefully. But your arrogance has likely blinded you to your stupidity, and you'll inevitably be unable to accept this fact.

Apple is Intel's customer. As long as Apple remains profitable, Intel will likely profit as well.

AMD is a direct competitor, but Intel has a vested interest in keeping them around. Without them, there's not enough money in the world to keep the DoJ out of Intel's ass. AMD is on cockblocking duty so Intel can save a bit of cash by not buying so much in the lube-and-lawyers department.

nVidia is not a direct competitor in the general sense. True, they produce a product that Intel is trying to compete with, but Intel is so pathetically behind the power curve on this that they're essentially never going to catch up. They've already lost the war for mindshare. Gamers and GPGPU users buy "anything but crappy built-in Intel GPU's". At this point, Intel is just providing an option to remove a marginally more expensive nVidia part from the BOM of cheap, low-end devices. nVidia isn't crying the blues over losing their least glamorous, barely profitable customers. That's not competition.

You should understand the market before you go and sink money into something, and it sounds a whole lot like you know precisely dick about the market you've invested in. That's not to say it's a bad buy. But you obviously have no concept of why it's a good investment, which will lead you to make missteps later on. A fool and his money are soon parted. Enjoy what you can of it while it lasts.

I think it might be hard

[Sloppy]

Intel's ahead, but it kind of looks like they have a lot of difficulty too. There are so many different part #s, that it makes you wonder if they all tried to be the same thing, but different wafers passed different tests, and .. viola, diverse line of products!

Re:I think it might be hard

[Anonymous Coward]

Viola? What's an oversized violin got to do with this?

Re:I think it might be hard

[serbanp]

he's just Sloppy...

Re:I think it might be hard

[Sloppy]

Damn, this is even more embarrassing than the flaunting/flouting error of 2009.

You must understand

[Anonymous Coward]

You must understand. There is great profit to be had by buying stock in companies you have been taught to hate by the liberal establishment. Such companies tend to be undervalued. I have also profited massively buying Conoco Phillips (fracking!), Monsanto (genetically modified crops!), ADM (high fructose corn syrup!), AT&T (non-net neutral networking!). Keep protesting. I will keep profiting.

Re:You must understand

[Anonymous Coward]

Money won't help you sleep at night, but you seem to be fully aware of the truth, so I am hereby labeling you a sociopath. Probably a psychopath, but that's debatable.

Nokia the ultimate outsourcing warning

[jphamlore]

The ultimate story about the dangers of outsourcing is how Nokia destroyed its mobile phone business. Once upon a time Nokia and Texas Instruments had a very close working arrangement with TI being Nokia's fab partner. The two together had a complete phone solution. So how does Nokia treat TI in the mid 2000s: They decided to diversify their wireless chipset providers away from working with TI. Only Nokia forgot one thing: TI don't play in markets where it cannot be overall #1. TI will as fast as possible get out of business segments where it cannot lead. And so TI said to Nokia, bye by 2012. In 2009. By then Nokia had decided it wanted to get its ARM SoCs and wireless modems from the same supplier, and there was one natural candidate, especially since they were, and still are, the leaders in LTE: Qualcomm. Only there was one big problem: Nokia had been caught in a patent war with Qualcomm for years trying to put Qualcomm out of business. It was Nokia that wound up having to settle for billions of US dollars, and suddenly it was at the mercy of the company to whom it had been an existential threat. Oops.

Re:Nokia the ultimate outsourcing warning

[Anonymous Coward]

Can you please provide a link to "TI does not play in markets where it cannot be #1"? I think it a good strategy, just wanted to learn more about it.

Former Semiconductor Furnace Engineer Here

[Anonymous Coward]

If you want the biggest bang for your buck, a single 12" Vertical Semiconductor furnace shipped and installed - $900K - $1m USD (low end guess). I've been to Taiwan into another fab, UMC, and they have hundreds of these furnaces. But with wafers that big, you need an automated transfer system (because you can't trust people to carry $10,000 worth of substrate) Wafer boats, typically holding 13 wafers, are $2000-$3000 each. the substrate itself is incredibly expensive. And not to mention the electrical bills...running one of these furnaces requires heating a 12" Diameter tube to up to 1200C....usually 480 volts into them. With the Silicon Carbide process tubes needing to be replaced every 3-4 months, and the heating elements getting an average 1 year lifespan, that's a LOT of money.

I don't blame them for not wanting to open their own fabs!

Re:TSMC would *love* to sell fab capacity...

[Glasswire]

And push our Apple or Samsung or Qualcomm for the right price. Again, AMD and NVidia can't or won't pay what that price would be.

Intel use their old 20nm fabs for their own GPUs?

[Anonymous Coward]

This looks like a great opportunity for Intel to use their "old" 20nm fabs for their own high end (discrete) GPU's while their 14nm fabs churn out the CPUs. They'd be ahead of AMD and Nvida on process shrink even using these old fabs... Intel probably has higher yield at 20nm than TMSC has at >20nm too....

what's funny....

[Anonymous Coward]

...is that Apple is now a small fry in the cell space (and going that way in tablet).

the foundry could make more money letting AMD/NVidia use their stuff than Apple.

Re:what's funny....

[Lunix Nutcase]

the foundry could make more money letting AMD/NVidia use their stuff than Apple.

In what alternate universe? Apple had nearly $183 billion in revenue in 2014. Nvidia had $4.2 billion in 2014 and AMD had $5.3 billion in 2013. How exactly would they make more money from companies whose yearly revenues combined are less than a single quarter of Apple's revenue?

Re:what's funny....

[unixisc]

Also, most of the phones I see w/ people around me are iPhones. I rarely see a Galaxy or a Lumia. I wonder why people claim that Apple's phone demand is falling? I don't recall seeing lines for new phones from Samsung or Microsoft the way it was when the iPhone 6 was released

Why not Intel?

[darkwing_bmf]

The only other manufacturer with 14nm capacity is Intel and there's no way Intel will sell them some capacity.

Why wouldn't Intel sell some of their capacity to nVidia or AMD for their GPUs? It's not like Intel directly competes with them (yet) in the high end GPU market and high end GPUs help sell new computers with Intel CPUs. I mean if Intel needs full production to fill orders, then of course they wouldn't sell capacity. But baring that (or engineering limitations), there should be no reason for them to refuse to sell capacity.

Re:Why not Intel?

[Lunix Nutcase]

Why wouldn't Intel sell some of their capacity to nVidia or AMD for their GPUs?

Because they make more money using that capacity to make their own CPUs.

Re:Why not Intel?

[Luthair]

Given Intel offers SKUs with higher performance graphics parts (iris) they are effectively competing with the lower end of the discrete GPU market, and probably over time hope to keep raising the bar.

Re:Intel use their old 20nm fabs for their own GPU

[blackomegax]

Intel has such a shit reputation with the GPU/gamer crowd that they'd profit more by leasing those fabs to AMD/Nvidia (strictly on GPU production, i doubt they'd allow AMD to make CPU's on it), whom can actually market and sell GPU's.

Problem is they also compete against themselves

[Sycraft-fu]

Current videocards work great. You can play all kinds of games in plenty of beautiful detail. So to get someone to buy a new video card, well you have to offer them a reason. If your new card isn't enough faster, or more efficient, or enough new features, or whatever then people will say "Nah, I'm good," and stick with what they got.

So they do have a need to move forward on tech, if they want to stick around.

Not their best stuff

[Sycraft-fu]

They only have one 14nm fab right now, and it only makes Intel chips. Maybe they'd sell some 22nm, I dunno, and that might be of interest for a GPU, but they aren't selling 14nm at this time near as I know.

Re:Not their best stuff

[sribe]

They only have one 14nm fab right now, and it only makes Intel chips. Maybe they'd sell some 22nm, I dunno, and that might be of interest for a GPU, but they aren't selling 14nm at this time near as I know.

Their announced plans for 14nm fabs are vastly more than what they need to supply their own chips--in fact, more than all the world's other fab production combined. Doesn't sound to me like they're planning on hoarding it forever.

Re:Not their best stuff

[Anonymous Coward]

You are incorrect:
http://newsroom.intel.com/comm...

I do not know everything they've announced, but there is definitely more of their best stuff coming.

Also they heavily invest in R&D

[Sycraft-fu]

Intel really goes all in on R&D. It isn't just that they make a lot of money, their percentage of R&D investment is very high, and they don't cut it in down times to try and squeeze out numbers. That snowballs in the long term and is why they had a 14nm process running when others were still struggling to finalize their 20nm.

Also a lot of the architectural arguments regarding Motorola v Intel were bogus. It was programmers arguing on academic issues that maybe theoretically should matter, but didn't actually make a different performance wise in final implementation. Little endian would be a good example. Plenty of people act like that is a hugely bad choice, yet it actually matters fuck all.

Re:Also they heavily invest in R&D

[Anonymous Coward]

The problem is that you seem to answer also as a programmer rather than someone doing actual hardware work.

First of all you need to understand some basic things... it is totally different to have a process and fab that is dedicated only to your engineers and cut out for a specific purpose, to serve your own designs. So it is completely different for Intel to start manufacturing in a process than a fab that is open to anybody and will need to be able to reliably manufacture 100's of different designs from a multitude of domains and customers at high yields.

you're missing the point, GP is right

[jensend]

The article is about GPUs.

ATI was fabless, and though AMD did still own their own fabs for three years after the ATI acquisition, that was completely irrelevant since TSMC still fabbed all the chips anyways. There were rumblings about doing some chips on AMD's own fabs but it never came to pass.

So the grandparent is right. ATI / AMD Graphics and Visual Solutions and nVidia have both always been fabless.

People who think owning your own fabs is always fabulous are disconnected from the realities of the semiconductor industry. It just isn't feasible for most companies to duplicate all the huge material and R&D investments that have to be continually at full throttle just to have any chance to compete in the fab space. Gamers, who care disproportionately about retail add-in graphics cards, routinely overestimate the size of the graphics card industry; Intel's revenue is over 10x either nVidia's or AMD's. Only a fab that gets a lot more business than just nVidia's can possibly hope to compete. TSMC fits that bill.

Re:you're missing the point, GP is right

[Anonymous Coward]

lets call a fab what it really is. A glorified Factory. There is something wrong when the majority of players cant generate a new factory quickly to fill their designs.

Re:you're missing the point, GP is right

[Anonymous Coward]

These are NOT just some huge buildings in the traditional factory sense where you can just throw up a new building roll in some machines and you have a factory. It requires very careful construction, clean rooms, extremely expensive and complex pieces of equipment to manufacture the chips and billions of investment. The machines themselves take many months to build. you are talking years of time from planning to completion and that is assuming everything runs perfectly smoothly.

Re:you're missing the point, GP is right

[bloodhawk]

each new fab is using cutting edge equipment, equipment that itself is custom built and takes a long time. a fab is more the equivalent of a very large cutting edge scientific research lab rather than a factory.

Screw 'em

[Anonymous Coward]

That's what idiots get for not having their own manufacturing capacity.

Reminds me of a guy I once worked for. He got rid of all his manufacturing capacity "to save money" and oursourced all that work to a firm in China. Within 6 months he was having trouble delivering product to curtomers on time and then found his Chinese vendir selling "its own brand" of HIS product at a lower price! The guy HAD to give them the design data so they could build units for him, and the test and programming data so they could do the testing and programming ... so they had everything they needed and they were not at all concerned about US Copyright and Patent laws (Duh!)

Morons who farm-out the production of their products DESERVE to go bankrupt and get sued into the ground by their investors.

Re:Screw 'em

[Ash-Fox]

Where is the Chinese knock-off AMD and nVidia chips then that are just as good, but cheaper?

Re:Intel use their old 20nm fabs for their own GPU

[Anonymous Coward]

Shit reputation... you mean the company that sells the CPU that pretty much any gamer buys...has a shit reputation with...gamers. Riiight.

If you mean GPU, it's a non sequitur as Intel doesn't sell gaming GPUs.

No Samsung

[austinpoet]

Samsung was never mentioned in the article, only in the summary.

Samsung also offers Foundry services, used for previous Apple chips for example, so I highly doubt they use up TSMC's capacity.

Oh happy fun joy joy! Sensationalism at it's best.

[Anonymous Coward]

Of course Intel's not going to give them any of their newest 14nm cap when it's already entirely dedicated to their own products. Don't be a fucking idiot making moronically inflammatory statements like that! The author simply wants us all to forget that upgrading processing nodes is generally very expensive and doesn't happen to all fabs at once, and in Intel's case we already know exactly what's going to be using the node and have zero spare capacity. As I mentioned already they surely have some non-upgraded fabs that they probably would rent, but the author is also forgetting that jumping fabs usually isn't as simple as taking our design from fab corp A to fab corp B unchanged.

OTOH there is nothing at all stopping them from renting other process nodes from Intel. There have been many mentions over the past years that Intel had been moving to allow some 3rd party usage of their manufacturing facilities. While I haven't heard much about it recently it's probably likely because TSMC had no cap crunch.

Another odd point about this article is WTF would Samsung be using TSMC. Samsung has it's very own fabs and processes. This makes no sense at all. As a matter of fact the only thing that does make sense is that Apple is using some capacity but I'd imagine that there probably is no shortage as I'm sure that both AMD & nVidia have long term contracts with TSMC.

Also the author neglected to mention other fabs like those owned by IBM and others.

Alrighty this article is just so wrong in so many ways that it's about to make my head explode...

Re:Intel use their old 20nm fabs for their own GPU

[blackomegax]

No I mean, Intel GPU tech has a shit reputation and the brand will not hold up in the gamer market if they transition into it.

you guys are looking at this all wrong ...

[Anonymous Coward]

You support AMD to force Intel to innovate and get new tech to market otherwise they will stagnate and as consumers you will vastly overpay for subpar product over time. Yes I am greatly oversimplifying here but bear with me.

Look at IBM and the pc. How did that turn out?

You can argue about the investment streams of rev to plow into R & D to fund the next tech which then needs a new fab. You can debate if Apples current crisis is they have a marketing dept powerful enough that they have neglected the software side of their tech its scary to a end user like me not to mention they want to pay thru the nose for anything they do sell you and dont mind acting like a monopoly in the process-got to love those chips forcing you to buy pricey charger cords if you want to have a working phone.

Then again in the early days it was fear of Apple driving what innovation that came out of Microsoft but there MS got so much share they could ignore Apple on the pc front.

conclusion YOU NEVER want a sole or all but sole supplier of anything in ANY industry because you will end up with overpriced crap. Look up the history of monopolies and cartel throughout human history. Given in this posting its about a highly capital intensive research and production processes which all but prevent anyone EXCEPT mutli-billionaires or corps from being able to gather such capital -human, IP, R & D, and manu let alone a marketing organization to get the result to market means we are stuck with Intel AMD and then niche and/or specialized manu corps like cell phones. How is Nokia or Blackberry doing these days?

Re:Intel use their old 20nm fabs for their own GPU

[Agripa]

The discrete GPU market is not large enough to interest Intel and it is getting smaller as integrated GPUs gain performance.