This is the first in a series of columns on the strategic direction of several major technology companies that have faltered of late. We’ll start here with Intel, follow in a couple days with Cisco followed by Microsoft, then see where it goes from there.
At Intel’s annual shareholders’ meeting last week the company talked about moving strongly into mobile chips and selling its stillborn OnCue over-the-top video streaming service, but the most important story had to do with expanding Intel’s manufacturing capacity. This latter news is especially important because if you look at the square footage of 14 nanometer fab facilities Intel says it will be bringing online in the next two to three years it appears that the company will shortly have more production capacity than all the rest of the semiconductor industry combined.
Not just more 14 nm production capacity, we’re talking about more total production capacity than all Intel competitors together.
This is fascinating news for several reasons. First, at $5+ billion per fab you don’t add 3-4 new ones without a darned good reason for doing so. Second, it takes so long to plan and build these plants that this part of Intel’s strategic plan had to have been in the works for years before the company ever mentioned it in public. So it’s not like new Intel CEO Brian Krzanich moved into his office and said, “Let’s build some new fabs.”
That’s not to say Krzanich has been without impact on the company. He clearly (and properly) killed OnCue, which was Intel’s third try at building a media empire and wasn’t any more thoughtfully done than the first two failures. Watch for Intel to sell OnCue to Verizon for $500 million as planned, but with sweetheart financing from Intel Capital so it doesn’t cost Verizon a penny upfront.
This is not a criticism, Brian: I’d do the same thing.
Increasing the fab capacity makes sense, too, though maybe not as much sense as it made a few years ago when the idea was first presented. Then PC sales were still growing and excess production capacity promised the sort of low prices that could finally kill the hated AMD.
Only today AMD isn’t that much of a threat. There are other players like Qualcomm in the mobile space and even Apple that are far scarier than AMD.
So what’s a Krzanich to do? He’s making the best of a difficult situation.
Understand that Intel is far and away the best semiconductor manufacturer the world has ever known. Its only real competitor in manufacturing is TSMC in Taiwan and even that’s not a close race. Intel has the best technology, the best yields, and because of the way things work in the semiconductor industry, they have the lowest manufacturing cost per chip.
What they don’t have, however, is the best mix of chips to build. Desktops are in decline, the market is all GPUs and mobile with a huge flash RAM opportunity on the horizon — all areas where Intel is at a disadvantage.
So Krzanich has Intel looking into building chips for others, entering the foundry business. On the face of it this move looks really, really stupid. But the more I think about it the more sense it actually makes.
What’s stupid about competing with TSMC and others in the foundry business is the profit margins aren’t good at all. Typically a larger fabless semiconductor company will pay around $6000 for each foundry-processed wafer, yet the same wafer filled with Ivy Bridge processors can easily generate $400,000+ in sales for Intel. That’s the whole idea behind vertical integration — to take all the profit from every stage of the business.
Admittedly there’s plenty of R&D and other expenses that need to be covered by that $400,000 wafer, but the idea of making it all yourself to reap all the benefit makes sense… except if you don’t know what to make.
That’s Intel’s problem. They can build the darned things better than anyone else but they don’t necessarily have the right product mix to build for the current market. And even if they get their product mojo back tonight that won’t have much effect on their business for another two to three years.
Brian Krzanich can’t wait two to three years for new mojo. He needs mojo right now.
And that’s why Intel is suddenly interested in the so-called foundry business. They are soon to have excess 22 nm and then 14 nm production capacity and will be able to easily undercut any other manufacturer at those feature sizes, all the while offering superior performance. Traditionally this would lead to a market bloodbath with most Intel foundry competitors dying and production costs eventually going back up as companies fail and capacity is cut back. But I don’t think that is what’s happening in this case.
If Intel drove the per-wafer price up to $12,000 or even $20,000 it wouldn’t make enough difference to Intel’s bottom line to be worth the anti-trust risk. They’ll still be cheaper, just a little bit cheaper.
I believe Intel is entering the foundry business mainly as an industrial intelligence operation.
As a chip company exclusively manufacturing its own designs Intel competes with most of those fabless semiconductor companies, but as a foundry — especially the cheapest best foundry around — those same companies will open their kimonos to Intel (with strict NDAs in place first, of course).
But this time the NDA doesn’t matter in this case because Intel’s purpose isn’t to steal trade secrets — it is to find companies to buy. Once you buy the company the NDA dissolves.
Intel needs new product lines sooner than Santa Clara can design them itself. More importantly, having rightly lost some confidence in its ability to predict and lead the market, Intel needs a few astounding ideas from outside and this is by far the easiest way to find those.
Suddenly Intel manufacturing engineers will have a view of the chip market they never had before. Find the best products that are close to market, open the checkbook and buy them up. What better way to find those new products that really work than by manufacturing them in the first place?
If I am right, Intel is emulating Cisco’s 1990s strategy of buying ahead of the next technology wave, though in this case leveraging its superior fab technology to figure out that next wave.
It might even work.
I’m thinking the big fish to land for Intel is Apple. The chip inside the 5s is darn good. Apple has a real problem with Samsung.
Instead of Intel buying other companies for intellectual property, this might be a defensive move to avoid being bought by Apple.
Thanks Bob for this piece.
I was aware that Intel was going in this direction, but I didn’t quite understand why.
I still don’t understand the complete source of Intel’s cost advantage in production but that they have that and better quality/skill puts them in a position to execute this strategy.
I wonder what would happen if a company insisted on a NDA that didn’t permit Intel to act on their intellectual property in any way other than manufacture the chip? What would Intel do then? Not make the chip? Would that attract DoJ? I know, this is highly hypothetical as any chip maker might prefer to be bought out by companies with big pockets.
There is another way to look at this. Intel’s business is an evolutionary process. With each new family of chips comes new fab lines. The bulk of their profit comes from the new products. As sales decline on the old products, production drops on the old fab lines. At that point the investment in the fab lines has been paid off. Through Intel’s business evolution they end up with nice, idle, fully paid off fab lines. Any use they can find for these fab lines will be easy money. Going into the fab business could be very good for Intel, and maybe the consumer too.
.
Making products for others is a great way to get tuned into market trends.
.
Bravo for Intel.
.
It won’t be long until Intel will start making ARM chips. They could either buy a license or make them for others. With their production capabilities, that could be game changer.
Yesterday’s CPU and graphics chip line becomes tomorrow’s chipset lines. The downside to this model is mobile design is more “system on a chip” instead of northbridge/southbridge style setups.
And Intel already has a license to make Arm chips (I thought they were making them for a while now, but I guess not): https://www.extremetech.com/computing/169853-hell-freezes-over-intel-announces-plan-to-fab-arm-processors
They have done this before through downturns.
Except this time they are already over capacity. They have delayed opening their latest new fab, Fab 42 and have laid off workers at the New Mexico site.
I agree, this isn’t the first time Intel has approached the fabless semiconductor companies and offered to fab for them.
My question is, why is Intel suddenly having this crisis in their R&D? They used MDF and their tick-tock chip design to strangle AMD for a decade, constantly bringing out new chips and buying up market share.
While AMD beat them to the punch several times, Intel was always able to get a good-enough chip to market soon after because of their incredible cash flow that enabled their design engineers to catch up.
So WHY is Intel so badly lacking in direction for their chips? Is AMD now no longer spurring innovation? Does Intel really have such a dearth of creative engineering vision?
Intel isn’t lacking in R&D. The latest bay trail processor already competes well with ARM. CPU wise they beat ARM. GPU wise they are still lacking a little, but they have made huge leaps. Anandtech already called Intel a true GPU manufacturer. At some point they are going to beat ARM.
Intel to offer the cheaper version? That would be the day. What’s likely is theirs will be more expensive because they’ll think they are ahead in process node technology. Also, I wouldn’t expect Nvidia or Qualcomm to allow Intel to make their chips anytime soon, and discover their secrets.
AMD might want to do it, but in that case it’s probably Intel that will refuse to make AMD’s chips at 14nm (and later at 10nm). My point is this is definitely not going to be an “easy sell” for Intel.
What you’re saying is that Intel will be building ARM chips. This is an obvious move as the X86 chipset market is beginning to shrink. The Atom processor is just not quite cutting it. Even if it does get competitive with current ARM designs, software and operating systems would have to be rewritten.
Part of this may be a way of keeping Apple, a customer that Intel spent a long time wooing, and has been extremely profitable for Intel. Apple isn’t interested in AMD, and they constantly want Intel’s best, fastest, and most profitable chips. Apple doesn’t build $300 systems that use whatever is in the bargain bin. Apple is a different company than it was five or six years ago. It’s now iOS and the Ax chipsets. There’s always rumors that Macs will switch to the Ax chips, but it’s more likely that the Ax chipset will mature and iOS will be able to replace a 15 year old OS X. Apple will always pay a premium for its chips.
Another reason is that Intel can make more than that $6000 per foundry process wafer. Apple’s Ax chip is designed by Apple. Apple isn’t using the off the shelf Snapdragon, it designs its own chips, and Apple is big enough to do just that.
If you’re making Android phones, and you use that Snapdragon chip, you can only compete on price. Yes, you can make a “premium” Android phone, but you’re simply competing against all other Android phones with the same processor. As Samsung found out, you can’t even pretend you’re phones are better and faster because the tech press knows better.
However, what if you had a customized ARM chip? A chip that is optimized for your particular handset? Not only would you have an ARM chip that can beat the competition, but you have one which couldn’t even be compared to the competition because no one else has one quite like it? This is where the big money is. Intel will not only fabricate the chips, but can also design them. That can bring up to $100k per foundry process wafer.
HTC, Motorola, LC and HTC aren’t simply competing against Apple and Samsung. They’ll soon be competing against a wave of generic Chinese phones. These companies can also buy the same off the shelf Snapdragon processor, use the same display, and the same OS and software. How can these companies compete against clones that are 1/3 the price? If you don’t want a WinPC style spec race to the bottom, you have to make a product that can’t be duplicated. What better way than a customized chip that those clone makers cannot touch?
“Even if it does get competitive with current ARM designs, software and operating systems would have to be rewritten.”
Android already runs on X86. Windows runs on X86. And iOS compilers also output both ARM and X86 code.
[…] Link. Good luck buying Apple or Samsung. […]
I’m not so sure Intel will be doing reverse-engineering on everybody else’s designs. last time I was reading 12 trade mags a week, the way fabless worked, was GizmoCo was designing their own logic, and shipping the file to whatever fab to reduce that to wafer, produce, and either crack off the chips into die or sending the wafer out whole. besides, that’s what NDAs are for, “don’t look inside the box.” more likely it’s a combination of market research and maintaining full employment until they need to put something of their own in that plant. if market research says to buy GizmoCo because that was the cheap way to gain customers and kill a competitor, so be it.
Where does Global Foundries fit in the big picture ?
Remembering the idea in your “iOS 7’s Secret” column…
.
Apple and the industry are rushing to displace desktop and notebook CPUs with smartphone CPU/SOCs.
.
Using BlueTooth Low Energy to communicate with ‘vacuum’ displays, keyboards, and mice!
.
Apple and Google, you argued, are rushing to this future computing paradigm as fast as they can with Apple arriving first with its A7 64 bit processor.
.
With the key chip in the phone, Apple and Google, will compete for Intel fab production. I think Apple would rather run with TSMC than Intel as Intel’s CEO sits on Google’s Board.
.
I don’t see Apple being so trusting that it shares fab facilities with Google while leaving Intel the possibility of being the king maker to the highest bidder between the two.
.
In fact, Apple may buy a foundry rather than be beholding.
And Al Gore sits on both the Apple and Google boards. so what’s your point?
What if a company manufactures its chips in Intel fabs, and then when the new Intel chip comes out claims that Intel stole their tech?
What if a company that infringes on Intel patents asks to have its chips manufactured by Intel, and if Intel sues claim a breach of the NDA?
I see potential for lawyers to make lots and lots of money…
Great to see you back on form with this incisive, informative piece. Hope the scar is healing up ok!!
Hi Bob,
You wrote:
“What’s stupid about competing with TSMC and others in the foundry business is the profit
margins aren’t good at all. Typically a larger fabless semiconductor company will pay around
$6000 for each foundry-processed wafer, yet the same wafer filled with Ivy Bridge processors
can easily generate $400,000+ in sales for Intel. That’s the whole idea behind vertical
integration — to take all the profit from every stage of the business.”
But it looks to me that TSMC makes profits comparable to Intel given the revenues for
both companies (it actually looks like TSMC is more profitable per dollar of revenue):
TSMC
(Net sales/Net income)
2008: 333,158/81,822
2009: 295,742/89,293
2010: 419,538/164,320
2011: 427,081/137,103
2012: 506,249/158,655
2012: $17.5B/$5.5B
https://www.tsmc.com/download/ir/secFillings/20F-2012.pdf
INTC
(Total Revenue/Net Income)
2010: $43B/$11.5B
2011: $54B/$13B
2012: $53B/$11B
http://finance.yahoo.com/q/is?s=INTC+Income+Statement&annual
Sure, the *highest* margin parts for Intel don’t have any counterpart at TSMC, but
on average I don’t see the problem with Intel being a foundry with foundry-type
profits.
Frankly speaking, as they like to say in Japan, I never quite know what to say about these TSMC numbers. The company occupies a singular place in the economy of Taiwan and can spend any amount of money on new fabs with unlimited financing at miniscule rates. Yes, the numbers look good, but why can’t UMC do as well? They operate in different universes is why. So does Intel.
On a profit margin basis, Intel and TSMC may be comparable. However, on a return on equity basis, Intel is far superior. This is because TSMC have to invest vastly more money in order to obtain the same revenues as Intel (by selling many more wafers). ROE is key, and Intel’s will decline if they move to becoming a foundry.
It could be true that Intel will look to acquire and vertically integrate custom foundry customers, but customer #1 is Cisco, and it’s definitely not going to acquire Cisco, which even you have listed as #2 on your list of ailing behemoths. But Cisco is just a tryout for them.
Intel is doing custom foundry because of mobile. Period. The end. Handsets have no margin for a OTS Intel mobile processor when the big players are integrating whatever they want into the main chip, fabbing it to spec and saving power. Intel can’t make a chip that’s everything to everyone unless their profit margin or power consumption is sacrificed by adding everything for everyone to their standard product.
There’s history in other consumer spaces here, too, where Intel won by default the first Xbox CPU, and then lost it because it was unwilling to sell the same specs at lower cost, preferring to sell newer OTS CPUs at the same cost and faster specs and maintain their margin. Margins were more important than the Xbox business.
But when you talk about the companies driving the disruption to Intel’s business model of making high margins on the latest CPUs, and the companies they would acquire… they’re not chip makers. They are system makers who design chips to spec. Like Apple. Like Cisco. Like Sony, Nintendo, Microsoft, Brocade, Juniper, Huawei… there’s no space for Intel to buy those companies absent those companies birthing a defacto industrywide standard chip that Intel can borg onto and pop out at higher specs and higher margins than everyone else. So what’s that? Intel’s going to buy which company that has a defacto monopoly on mobile device CPUs? Maybe that’s ARM, but no effing way can they buy ARM unless every monopoly law on earth is ignored, and ARM’s not even in the business of controlling who manufactures its designs.
Which gets you back to… on mobile, Intel’s traditional business model is fcked. And maybe just being a foundry like those poor disrespected boring penny pinching Taiwanese is the best they can do. And you know what? That’s not terrible because Taiwan does it. It’s indistinguishable from magic, even if TSMC is doing it outside the headlines. Samsung too. High tech, high revenue, shitty margins. Intel, you may be in for it too.
Unless they fab for Qualcomm, who they can’t afford to buy.
Yes Intel could buy Nvidia to get decent GPU designs. The problem is if chip designs get to be a commodity, but not a standard, then buying a bunch of small niche players will not add up to a big win.
I see Krzanich building fabs because (1) that’s the side of the business he comes from and (2) in their heart Intel is a silicon company.
The hell of being the low cost manufacturer is keeping your expensive fabs fed. The delight, as Bob alludes to, is if you can have a successful in-house design so you can increase wafer value by orders of magnitude.
The way the technology industry has evolved towards mobile designs has meant there isn’t a single large chip winner. The biggest winners are Apple and Samsung, who do their own designs based on ARM technology, but even combined they are still a minority of mobile designs.
It’s a new post-PC world where there is no large overarching winner, meaning standard, and that means Intel can’t make a chip for everybody and has to sweep up the smaller players. But even that won’t save their profit margins as x86 declines. Intel will sell more chips, make less margin, but has plateaued as a technological and economic force.
“Why Intel wants to be everyone’s chip maker” is rather stating the obvious. The bigger a company gets, the more of a chore competition becomes. The answer is of course – Global Domination! Aren’t monopolies wonderful!
Also, “Desktops are in decline”. Can we stop saying this like it means something? Sure, the chimpanzee in the street doesn’t buy desktops, or even laptops, anymore because their needs are minimal and they can get by with a tablet or just a smart (sic) ‘phone. As soon as these come with an electrode in the brain for creating regular orgasms, all they will really need is a 555 timer chip and a battery… and will the highly ethical types in the commercial world shy away from selling such things. Of course they won’t.
The people who need desktops still buy desktops! Hopefully a sufficient variety (you know, that thing that is the spice of life) of computing devices will continue to be a available so that they can continue to do so. Or are you trying to talk us into a miserable future where one size must fit all no matter how poorly?
The real answer is money laundering.
Intel won’t succeed in the foundry business and they’re hitting the physics / economics wall on Moore’s law.
Here’s why I can say this:
I worked in the EDA (electronic design automation) industry for 20 years with companies like Mentor, Synopsys and Cadence. We supplied tools both for designing chips as well as for photolithography mask creation, which is used by the fabs to create the production masks. This is a non-trivial process because as the dimensions shrink, the light from one rectangle (there are other general polygon shapes, but most are rectangles) begins to optical interfere with its neighbor. I.e. the diffraction effect. This means you need to alter the shapes on the mask to account for this when you print. Kind of like distorting an old fashioned film negative to account for flaws in the enlarger lens.
I started off writing code for these tools and later went on to work with the foundries (e.g. TSMC, etc.) to implement design flows both for the fab and for the customers they were supporting. Place-and-route, physical verification, simulation, etc. etc.
Creating a true foundry is incredibly difficult as these tools are very complicated and don’t work well together. You have to create a lot of ‘glue’ code (scripts, etc.) to make everything work. TSMC had several hundred engineers doing nothing but creating & verifying tool flows. (Note the plural; different customers liked different tools just like some carpenters prefer Porter Cable saws but DeWalt drills. You can’t force them to all use one brand if they don’t want to.)
One of the guys I used to work with at Mentor here in Oregon (where Intel’s biggest fabs are and where all new fab development is done) took a job at Intel on their CAD tool flow development team. He left recently after he found out just how screwed up Intel is trying to build flows for outside chip design teams. He used to work with me supporting tool flow development at other foundries, so he knows how this should work.
The fabrication side of Intel has been able to dictate to their internal designers what they can and can’t do. What tools they can and can’t use, etc. You can’t do that if you want to work with outside customers. If they want to use Synopsys for place-and-route and not Cadence, you have to support them.
I’ve worked with Intel as a vendor and they’re probably *the* most arrogant customer I’ve worked with (followed closely by TSMC). It was their way or no way. You couldn’t even discuss that maybe a different solution to a problem might be better, they just didn’t want to hear it. That was the problem my friend was seeing and he couldn’t convince management that you’d never be able to force customers to design ‘the Intel way’.
Not a great attitude when you’re the vendor, not the customer.
This article (http://goo.gl/Yin8Rd) from the Oregonian does a good job explaining the technical challenges Intel is facing. Note that the increase in square footage is due to the problems printing at small dimensions, not because they’re adding more capacity. Much of the extra equipment is needed to overcome the physics of the problem, not to add production capacity.
Finally, this isn’t the first time Intel has gone down the foundry path. I think I still have a glossy brochure from the late 90’s when they tried to build a foundry business and that collapsed.
I wouldn’t hold my breath that they’ll get it right this time.
m
I don’t get your point about tools. If Intel becomes a fab house won’t they just accept a pg tape that will be “taped-out” by the designers/customers who will use any tool they want to create it?
> just accept a pg tape
.
Before that, fab houses have to provide a lot of data to customers, so they can feed it to their design tools. It’s called a physical design kit (PDK). The customers have in-house teams to massage data for different flows, but they need some support on the fab side.
I can see Intel doing foundry work for a few select customers, but nowhere near the variety that TSMC does. They’ll go after the bigger customers, such as Apple, if they do. Mark (above) makes some very interesting comments on the problems that Intel would face if they did general foundry work.
.
I’m sure that Apple would love to drop all their Samsung components as soon as possible. They’ve been feeding Samsung and having their hand bitten off for far too long.
Intel’s plans to produce an ARM Cortex A53/A57 ARM chip is probably a reaction to AMD’s recent announcement to do the same.
In the past Intel would have responded instead by producing an X86 chip in competition to it. They could have used their huge resources and Moore’s Law to compensate for X86’s inherent weaknesses compared to Risc designs like ARM’s. They’re still doing that with some success but they’ve gone about as far as they can go.
They’ve done almost as much as they can with the increasing number of transistors provided by Moore’s Law, like better circuitry for faster calculations, 64 bit CPU, extra cores, integrated GPU. The latest feature they’ve announced is a huge L4 cache shared between CPU and GPU, which sounds a lot like general purpose RAM, unlike the other developments which were improvements to the processors themselves.
I think this means that Intel’s move to ARM signals the beginning of the end of the x86 architecture. They’ll still push x86 CPUs as much as they can but I think it’s only a matter of time before AMD, or maybe even Intel, brings out a hybrid x86-ARM chip, followed by an irreversible move to ARM. Intel might be counting on being able to charge a premium for their chips from their more advanced fabs which are ahead of the competition in terms of resolution (currently 14nm) and capacity.
[…] I, Cringely, Why Intel wants to be everyone’s chip maker, here. Cringely pushing the analysis that Intel go to the lower margin foundry business so that they get […]
This is not some clever ruse to spy on competitors and take them over. This is an act of desperation. Intel obviously planned for a huge increase in the PC/server market, and that has turned out to be a major fail. So now they are going to be stuck with a large glut of capacity. Mistakes like this can cause a business to fail.
.
Competitors are not going to give Intel a look into their products unless the price is very cheap. Intel will be lucky to just stop the bleeding and break even on the new fabs.
.
The only way to use up that much capacity is a general purpose chip, like memory. Look for solid state drives to become very cheap in a few years, as the market is flooded. As always, over capacity is very bad for stock prices, but very good for consumers.
I’ll add one more comment. Intel needs to buy software companies, not hardware companies.
.
Their growth has always been blocked by Microsoft’s poor execution. How long did Intel have to wait for 64 bit windows… 10 years? And try as they did, they could never get Microsoft’s support for a risc replacement for x86. So they were stuck having to incrementally improve an outdated design. This allowed AMD to constantly nip at their heels and occasionally jump ahead for brief periods.
.
I say Intel’s new strategy should be to buy up cloud and virtualization tech, and create the standard for hosted services. Their goal would be to churn out massive quantities of servers built on their own chips. Google did that with android. It would allow them to innovate again and build the best chips they can, instead of being tied to a software vendor who doesn’t give a damn.
“Microsoft’s poor execution” only from Intel’s perspective. Microsoft maximized profit beautifully for themselves without a risky switch of their OS kernel.
I agree that both nature and business tend to overshoot and Intel has too much capacity. No doubt that’s why they are willing to take on other business. The cloud is in the ascendancy so it would be a growth market for Intel to enter, but there is already serious competition from AWS, Google and even Azure. It’s not a slam dunk win to build and hope business migrates their way.
Still Intel has to try something and they’re big and rich enough to well, try everything. Why not build a data center, or buy up Rackspace and Apigee? Why not gobble up startups that assist in “cloud-ifying” everything? Intel is piling on R&D expense into Android for x86, ready to fabricate ARM chips for anybody, but that is only the client side of today’s computing. Intel is nowhere in the cloud except as the dumb brains of servers, which are likely to be replaced by less power consuming ARM processors. Moving into the cloud makes sense, but I don’t see it happening. Intel has too much of a silicon and device mentality.
“How long did Intel have to wait for 64 bit windows… 10 years?” Throughout that time the main advantage of 64-bit was the ability to address more than 4 GB of RAM. Now that Microsoft, Intel, and the PC makers got together and forced 64-bit down our throats, who actually needs 4 gigs of RAM now that we are moving toward cloud processing and thin clients?
I do. I’ve got 16GB in my laptop and wish it was more so I could have a better performance balance between what I run in my VM and what I run on my laptop OS.
Are you sure RAM is the performance limiting factor? The MS experience index may provide a clue.
“And try as they did, they could never get Microsoft’s support for a risc replacement for x86.”
Um … Windows NT supported MIPS, PowerPC and Alpha for a while. At the time I thought that Intel considered these folks competition. What are you getting at here?
If you can’t beat them – Join them!
.
Having re-read some of the comments I then wondered if Intel might be better producing their own ARM designs rather than acting as the foundry for others. Intel have the best fabs with the leading technologies in silicon design. But the wrong products for today’s mobile markets.
.
Take the new technologies being rolled out for the current generation of Core processors and apply that to an ARM processor. Then, integrate various functions such as video on to the same IC to add value but importantly reduce power consumption. The end result is an off the shelf part which beats custom designs simply because the technology is more advanced.
.
If Intel goes down this route the likes of Apple would beat a path to their door!
“Increased manufacturing capacity”?
What happened?
Did the boat from Vietnam sink?
The only thing I can think of that could use up Intel’s new fab capacity is if Intel is planning on integrating ReRAM on the CPU die to replace DRAM chips.
Thanks for the ReRAM reference. Got more info here: http://en.wikipedia.org/wiki/Resistive_random-access_memory. Interesting that some think it’s based on the memristor. (MeRAM?) Didn’t the 1960s computers have non volatile RAM in the form of iron cores, as in “core memory”.
The chip manufacturing industry has out-innovated itself. It can produce more transistors per dollar per square mm at a faster pace and at faster [processing speeds than the world can produce printed characters on paper, magazines, photocopies, etc, but in 5 metal layers, in the most complex industrial factories of our time. And that was ten years ago…
Today it prints even more transistors and it will soon have more productive capacity than the world has needs for today which means our advanced technologies will be so cheap they literally disappear into our surroundings.
Local or cloud are arguments that are temporary because it won’t matter. But Intel’s shareholders won’t be pleased by its ability to add so much manufacturing capacity that it can’t use.
But the dirty little secret of the chip industry is that you can’t run your chip fab at anything below full capacity which means you are stacking up major losses if you can’t fill the fab, and you are building up massive unsold inventories (if it’s your own chips you are making) that can sink your business.
Intel has two ways to lose massive amounts of money. Intel can delay the new fabs and risk TSMC catching up or it steams ahead and floods its own markets with massive inventories. However, to say Intel is in the foundry business to spot the companies with great IP is ridiculous. Even if Intel bought some of thee fabless companies it wouldn’t provide enough volume or income to offset desktop sales.
Even if Intel owned ARM it wouldn’t solve Intel’s problem: How to make 60% margins on each finished Silicon wafer through its foundry business.
Intel simply became too good at making chips. It sells microprocessors because they are high margin but there isn’t an equivalent high volume high margin chip product to sell today.
“Intel simply became too good at making chips.” I keep stumbling over that sentence. I get the point that the high margins depend on high volume. But it can’t hurt that they are good at making chips. They just have to learn how to adapt that skill to lower volume uses or to combine chips with other components to make finished products, perhaps like Apple, or now, even Microsoft with their “devices and services” business.
IMO, the most fundamental mistake the Intel made was misunderstanding what made their core business worked so well in the past…They thought they succeeded for so long because they had the best chips and the process lead, when in actuality it was due to the Wintel ecosystem as the sole computing solution of consumers and they were the only ones with the manufacturing volume to satisfy it.
This business model broke down once the disruptive flood of ARM devices started at 2009 when users finally had a viable alternative of their computing needs. For the first time they didn’t need Windows and it’s x86 baggage, complexity and cost, or Intel and their extra performance and cost. The price of moving away from Wintel was mobility, and to most it was well worth it. Bulk of the hardware margins moved from the CPU to the end product, Apple and Google came to dominate the software part of the equation and Intel had nowhere left to go, even for the PC side it was already clear they outsold CPU performance to most people ; the main bottlenecks was UI/UX, network bandwidth and mass storage, none of which solved with a ever faster CPU.
I have to agree there. I think there desperate as one poster said earlier, this is a fundamental change of their business model and they wouldn’t have done it if they didn’t have to. That said, they do seem to have a few irons in the pot. But there sort of like dice rolls ,who knows if wearable’s is the next big thing, and even it is , the value/margins will be in the ecosystem.
PS: Intel still has an ARM license.