Intel IS Building a Foundry Business

[Daniel Nenni]

According to sources all over Silicon Valley Intel is aggressively recruiting OEM's for foundry business. Intel is also actively hiring Foundry people and ASIC designers.

Here is one quote:

"Over the last few months we have been hearing comments from industry contacts that Intel is talking to OEMs about a foundry relationship. We have heard Intel is looking for ASIC designers and other support staff for this effort. More recently, we understand Intel has approached Motorola. Currently, we believe Motorola has been working with Toshiba as an ASIC/foundry vendor for cell phone components. We believe the direct-to-OEM foundry model makes sense," analyst Gus Richard with Piper Jaffray wrote in a research note, reports Tech Trader Daily.

And another:

"Likely target customers would include EMC, Cisco, Juniper, Sony, Motorola, Apple, Nokia, and other large customers of leading edge logic. Intel has clearly articulated they are interested in working with companies that want to use x86 architecture. The company is not interested in enabling its fabless competitors or ARM," said Mr. Richard.

This quote is my favorite: "I was even contacted about a foundry position at Intel." said Daniel Nenni, Industry recognized blogger.

Intel also hired away GlobalFoundries Corporate Communications VP Jon Carvill. Just another data point.

View attachment 1228

Intel will certainly have plenty of 22nm fab space!

My opinion: Intel sucks at mobile semiconductor design and has no design ecosystem to openly share. ARM owns mobile, how will Intel succeed without working with ARM?

Anybody else think Intel will succeed in the foundry business? Beyond companies they have a financial stake in?

D.A.N.

 

[achalk]

I am just curious, if all of this development can lead to split in Intel sometime in future (not necessarily immediate future)? I mean Intel splitting into a foundry company and a processor/asic/circuit design company.

 

[Daniel Nenni]

I am just curious, if all of this development can lead to split in Intel sometime in future (not necessarily immediate future)? I mean Intel splitting into a foundry company and a processor/asic/circuit design company.

Similar to what Samsung has done you mean?

 

[kingmouf]

I believe that Intel has always behaved like the 800 pound gorilla. No matter if it had good or bad products, its approach has most of the time been : "I'll outdo you" rather than "I'll outsmartyou". And the two driving forces for this strategy have always been its manufacturing superiority and its financial wealth.

That is why it has not been successful to anything outside its core business - x86 *fat* cores. Now we are reaching a point where other competitors have realized this strategy and they are working on an "outsmart Intel" strategy rather than try to outdo it as well. And it pays well.

If Intel cannot be successful on an "outsmart" strategy and its "outdo" strategy wont work anymore, then the best card it has on the table is its manufacturing diamonds. I believe that Intel eventually will have to work with other partners either as a way to promote x86 (for example, I believe that they would try to propose to Sony or Microsoft a custom x86 part for the PS4 or XBox 720 and welcome their contribution. The same could be true for Apple or other high-volume manufacturers that require high-performance parts) or properly enable its products by using third party IP that needs extensive access to manufacturing information (for example scrap the totally unimpressive Intel GPUs for another design).

In the end maybe Intel will become something like Samsung (I consider a foundry spin-off a much less likely scenario). But then it will face the same and probably even worse problems that Samsung faces. If it will manufacture ARM (or MIPS or Power) chips then it will give its competitive advantage to direct competitors. On the other hand, its most advanced processes would mostly appeal to companies like AMD, nVidia, IBM, Oracle, Fujitsu etc. But these, for one reason or another, are competitors to Intel and would not want to go to Intel for their parts. If Intel tries to sell less advanced process node manufacturing to other companies, then those companies would be much better served by competitors such as TSMC, Global Foundries etc who have a far better customer support service and more developed foundry relations.

 

[harpoon]

I think the foundry model for intel may not be disasterous if they can find a "consumer" of chips who are not their competitor.

... sony already uses ibm, and you can probably rule out amd and nvidia.

so maybe apple, motorola, nokia ? How about microsoft and google ?

The other direction they could go is to open it up to research institutions and universities (e.g. MOSIS/Europractice) and try to get some great ideas from these "smart" people ... and then BUY them ! Maybe this is their "outsmart" strategy ?

 

[aparonoff]

Intel's strategic decision

Intel is a world leader is semiconductor technology and CPU design. ARM has garnered design wins in consumer and mobile sectors, and is turning its focus to capture design wins in the server space. ARM is a real threat to Intel since now the PC market is mature and parts of the PC market is ceding to tablet PCs which ARM is the leading processor.

For Intel to consider the foundry market is a signal from Intel that it will have excess leading edge wafer fab capacity. The challenge for Intel in the foundry space is to navigate a path that will (1) enable Intel to monetize its fab and process technology beyond Intel's own requirements, and (2) to do so without adding to ARM's strengths in Intel's targeted markets. If Apple (or Nvidia, TI, Qualcomm, or other company with ARM based application processors) were to be a foundry customer of Intel, then the benefit that Intel currently enjoys in its own CPU technology due to process advantage will be diminished as those benefits would also be applicable to ARM based processors.

Any move by Intel in the foundry space will have some interesting boundary conditions based on Intel's view of its competitive position in the CPU market. These boundary conditions will impact the success that Intel will have as a foundry provider. Intel has made several attempts in its history at using its technology and IP in an ASIC model, but has given up, or in the case of Open Silicon, spun it out.

The strategic decision for Intel is not whether or not to enter the foundry business, but whether or not to spin out its semiconductor technology and fab operations.

 

[nitindeo]

How Atom and Foundry are related.

According to sources all over Silicon Valley Intel is aggressively recruiting OEM's for foundry business. Intel is also actively hiring Foundry people and ASIC designers.

Dan: In 2010 Intel did $40M of foundry business compared to $40B of total revenue. (Similar story at Samsung - $400M compared to $28B total revenue). (Also compare all of this with ARM's $663M revenue last year - just to put everything in proper perspective).

Atom and Foundry Strategy are connected in a very complicated way.

Intel has realized that for Atom to do better, their foundry strategy has to be take shape. One of the major advantages of ARM is the fact that it can be manufactured at any foundry.

At the same time, I can only imagine that Intel wafer prices are much higher than other pure-play foundries.

From Intel's perspective, there are a couple of things going on. They are not ready to give up close to $20K per wafer revenue for their processors and sell that capacity to make $5K per wafer. And the other point is that there are two major factions inside Intel - RISC vs. CISC.

Until Intel makes a decision that they are in 'processor' business - not CISC business and until they realize that they need to build the complete silicon ecosystem to do better in ARM-wrestling for mobile market, it will be status-quo.

Anyway, it will take time for that staggering $40B number to feel the pain!

But I have faith in Intel's dynamic management - as they have shown strategy change and complete business focus in the past (e.g. DRAM, peripherals, desktop to laptop change, and so on).

One thing I know for sure is that consumers are benefiting from all this. Competitive prices, excellent products and great service. Now, if only someone could do something about the other giant in Washington!

 

[Quan Gan]

Welcome Intel to the foundry bussiness.

 

[admin]

How can you really enter Foundry Business if you keep an eye on customer product architecture ?
Posted by Bruno MUSSARD

 

[admin]

Agreed, please check: http://www.achronix.com/
Posted by Raul I Lopez

 

[admin]

yes , as TSMC , Global Foundries etc .. are getting huge orders and stepping up production to meet the market demand , intel also want to have the share of the billion dollar fab business
Posted by Sivakumar T S

 

[admin]

For Intel this is a paradigm shift. They started out doing DRAM remember. Then wisely decided to make computers not semiconductors. If they are serious then it signals a new era in the Semiconductor business. The consolidation will ultimately weaken nor longer strengthen the business and the technology. The fear of error will dominate the striving to inovate and do better
Posted by Gary Hillman

 

[amit n]

With Apple becoming the world's largest semi buyer (Apple now world's largest semiconductor buyer | The Digital Home - CNET News) and using ARM for their iOS products it might make sense for Intel to get into this revenue stream as a foundry. But that means Intel would manufacture ARM based products - will/can this happen? Or will Intel create a bleeding-edge foundry eco-system around Atom to try and get iOS and Android design wins?

There was buzz a while back that Apple would buy ARM but maybe it made more sense for Intel to do so (Apple Isn't Going To Buy ARM, Despite The Market Buzz (AAPL, INTC, ARMH)). Apple is just unbelievable - re-shaping the mobile, PC, semi, cloud, music, social media, payments, ... landscapes - all at once.