INTC - Intel

[Palantir]

Is it feasible for Intel to sell ARM chips? Given their fabrication edge, they could totally adopt this standard and hedge their bets and possibly win within the ARM space while pursuing x86 chips in mobile as well...

[FrankArabia]

Yes they can....

 

the hardest thing about breaking into the ARM market is getting the manufacturers/app writers to adopt a new coding system....that is a significant barrier to entry.....but I suppose that is why INTC is working with Android to circumvent some of this.....

[ItsAValueTrap]

They are currently selling ARM-based products.

 

Infineon wireless uses an ARM core in its wireless designs.  Intel bought Infineon's wireless division and uses the technology in the Medfield chip for smartphones.  Intel is one of ARM's bigger customers.

 

Intel used to have a division which designed chips based on the ARM architecture (Xscale), but sold off that division to Marvell.

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

 

Nvidia's CEO has remarked that Intel should build chips for other semiconductor design companies:

http://www.forbes.com/sites/briancaulfield/2012/03/22/nvidia-intel-should-let-us-build-chips-in-its-factories/

 

My take on that here:

http://glennchan.wordpress.com/2012/10/26/intels-moat/

 

[FrankArabia]

good stuff glenn

[Ross812]

Intel can make ARM chips, but it looks like they are reluctant to do so thus far because this would pressure their profit margins. Intel may be interested in taking on the orders for a huge customer like Apple, but their manufacturing efficiency lies in making several great one-size-fits-all chips. I don’t see Intel contracting out for smaller jobs like TSM or Global. It looks like Intel is betting on demand for their new line of Haswel mobile chips and redesigned atom chips in 2013. Contracting out older foundries to make ARM based chips is always on the table if necessary. Intel is spending ~10B on two new production facilities in Arizona and Washington. This is a bet on their advanced proprietary chips because they don’t produce their flagship chips outside of the US and Europe. I suspect they put the plans in motion in 2010 when they were projecting years of growth ahead in the traditional PC business. Hopefully the mobile demand is there to fill these new foundries to capacity. Time will tell…

[FrankArabia]

Ross, what is your rationale behind Power Corp?

[Ross812]

I bought Power Corp in late 2008 during the financial crisis. My rational was using POW.TO to tag along with Paul Desmarais. So you could say that it was a jockey pick. I liked that Power Corp went into the financial collapse well capitalized and I expected them to begin acquiring troubled banks. Desmarais was not as aggressive as I thought he would be so my reason for buying didn’t play out as I had envisioned, but off a cost basis of ~13 I’m making a ~9% dividend. I’ve been selling it off slowly if I need money to buy something else. I recently sold a little bit on last Friday and added to my Altius position and bought Fairfax for the first time. Some of the discussion on the board lately about a housing collapse coming in Canada and the fact that I think BAC and BK are much better investments are helping me sell it. POW.TO has gone from a ~10% position to ~2% across my accounts in the last couple years.

[ItsAValueTrap]

I don’t see Intel contracting out for smaller jobs like TSM or Global.

Intel is currently making chips for two FPGA startups.

 

One of their FPGA customers has a FPGA product with an atom core on it.  I am guessing that this is a similar idea to FPGAs with ARM cores on them.  Xilinx and Altera dominate the FPGA market... they do not manufacture their chips with Intel.  Intel has tried to design FPGAs in the past.

 

Some other information here:

http://www.programmableplanet.com/author.asp?section_id=2042&doc_id=247093

 

----

In general, I think that the trend is this.

 

Many different semiconductor companies are increasingly banding together to form foundry alliances and partnerships.  This is mainly because fabs benefit from economies of scale and pooling their manufacturing together increases those economies of scale.  One model is to build fabs and to sell that capacity.  Another model is to enter into joint-venture partnerships.  Formerly everything was vertically integrated.

 

The argument against these alliances is that the partners may have conflicting goals.  (I don't know too much about this area.)

 

My guess is that the trend towards fragmentation will continue.  Most (non-analog) semiconductor design companies are getting rid of their manufacturing divisions, e.g. AMD spinning off Global Foundries.  Eventually this could turn into an inflection point where a contract fab has revenues similar to Intel's... taking away Intel's manufacturing advantage and opening the door for AMD to compete with Intel effectively.  I see this as Intel's biggest threat.  Intel dethroned IBM as the world's leading semiconductor manufacturer, and somebody else (probably TSMC) may do it to Intel.

 

Right now Intel is way ahead of TSMC.

[Guest rimm_never_sleeps]

what matters is exactly what they are manufacturing. you can be the world's largest chip manufacturer, but if you are primarily building chips for products that are less and less in demand, your scale doesn't help. it actually hurts.

[Ross812]

Thanks value trap. I just read your write up on Intel on your blog. I thought you had a lot of good information on there. Do you see think Intel is refusing to contract manufacture ARM SoCs because they don’t want to help perpetuate ARM chips? It seems like Intel is betting their future growth on being relevant in Windows 8 and Android mobile platforms. It seems like the best fit for Intel manufacturing any ARM chips would be Apple. Intel would not be cutting into their target market for x86 Android and W8 mobile, and would avoid helping Nvidia and Qualcomm perpetuate ARM Android and W8 RT. It would benefit Apple too because it would eliminate at least a portion of their supply problems. This of course is all speculation and flies completely contrary to all the articles calling for Apple to leave Intel… 

[mikazo]

Does anyone have any thoughts on who might make a good CEO? I've looked into Intel briefly but need to do some more digging before buying. Looks like it's getting cheaper by the day.

[FrankArabia]

if you're INTC, why would you want to be a pure foundry? Simply being a manufacturing for chips is not a great business as you're not in the drivers seat nor are you able to dictate any terms. TSMC has been doing well because the demand is there but once innovation turns, they will be a leg behind as I suspect that will be the case soon.

 

Building chips is almost a pure commodity business. The only way around it is to innovate chips by making them faster and better all in the meanwhile anticipating new trends and staying ahead of them. This requires you to be part of the whole process.

 

When something new arrives, ARM will have the same problems INTC is having right now, with the added difficulty of having to coordinate the new capex with its foundry partners.

 

 

[ItsAValueTrap]

Do you see think Intel is refusing to contract manufacture ARM SoCs because they don’t want to help perpetuate ARM chips?

Reasons you could give would be:

1- They want to help the smartphone SoC design side of their business.

2- It would be really nice for Intel if x86 were a viable option for smartphones.  Then it could simply re-use its existing low-power x86 designs.  This is better than making a custom ARM design (which is what Apple did) as well as making low-power x86 designs.

3- The design side of things can have very nice margins.

 

Reasons against it would be:

Intel could eventually get killed by a contract fab company if that company reaches the critical mass needed to compete with Intel.

 

if you're INTC, why would you want to be a pure foundry? Simply being a manufacturing for chips is not a great business as you're not in the drivers seat nor are you able to dictate any terms.

That is Intel's strength in my opinion.  They design CPUs as well as AMD (sometimes AMD had the faster CPU) but Intel does not enjoy a special advantage there.  They beat AMD because of their scale in manufacturing.

 

On the other hand, semiconductor manufacturing has been like airlines if you were in memory.  Intel was originally a memory company and exited the memory business.  Semiconductor manufacturing is a nasty business that sometimes has brutal price competition.

I don't think that Intel/AMD are going to get into a price war since AMD doesn't have the cash to do it right now.  So unless Intel gets super paranoid about its moat and market share, I might expect Intel to "raise" its prices so that it and AMD start making more money.  An argument against doing so is to make sure that TSMC and the other contract fabs don't get too big.

 

Building chips is almost a pure commodity business.

Memory chips are more or less a commodity business.  Some companies make faster memory (on the retail level you can buy premium RAM from companies like OCZ) but faster memory with lower latencies has almost zero effect on overall system performance.

 

CPUs are a little different.  Intel right now makes the fastest CPUs and nobody can beat them.  Having cutting-edge fabs is a huge advantage in making the fastest CPUs.  You can use a larger die area to make the CPU faster but cost goes up (especially since yield problems go up when area goes up).

[FrankArabia]

thanks for the insight. I'm new to tech. I like the numbers and the industry economics I see.

[Palantir]

Intel's battle with ARM about Fabs

http://www.siliconvalleywatcher.com/mt/archives/2012/09/intels_battle_w.php

 

 

This is a really good article.

[ItsAValueTrap]

Intel's battle with ARM about Fabs

http://www.siliconvalleywatcher.com/mt/archives/2012/09/intels_battle_w.php

 

This is a really good article.

 

I'm not sure how true that is.  Intel is using its older fabs to make smartphone SoCs.  Medfield is produced on a 32nm process, not Intel's latest 22nm finFET process.  The Medfield chip will only migrate down to 22nm FinFET at the last quarter of 2013 according to leaked roadmap plans.  By Q1 2014, Intel may be producing chips on a 14nm process.  (Delays may push everything a little later than that.)  But clearly the idea is that Intel will use its older fabs for smartphone SoCs.

 

Volume-wise, smartphone SoCs is not that big of a market.  Currently the market is a little over 7B in revenue.  The reason why it isn't a big market is because these are low-cost parts compared to desktop and server CPUs.

[Guest valueInv]

Intel's battle with ARM about Fabs

http://www.siliconvalleywatcher.com/mt/archives/2012/09/intels_battle_w.php

 

 

This is a really good article.

 

Great article!! Articulates the hidden risks in Intel. They could go south really quickly.

[mikazo]

Interesting article on the possible future of semiconductor manufacturing:

 

"Research discovery could revolutionise semiconductor manufacture"

http://www.lunduniversity.lu.se/o.o.i.s?id=24890&news_item=5962

[boilermaker75]

Interesting article on the possible future of semiconductor manufacturing:

 

"Research discovery could revolutionise semiconductor manufacture"

http://www.lunduniversity.lu.se/o.o.i.s?id=24890&news_item=5962

 

Silicon is pretty cheap. I am guessing the cost of a 12 inch silicon wafer is no more than $100-200, which is insignificant in the scheme of semiconductor fabrication. I played this university game of doing semiconductor research for a long time, now I just teach and do admin tasks. The main product is usually the students you graduate, not the technology developed. There may be practical uses of what they are developing, but probably not anything they have yet envisioned, especially if this is revolutionary.

 

GaAs was often referred to as "the semiconductor of the future." Then people started adding "and always will be."

 

It is near impossible to compete with standard silicon processing, which is always improving because of the billions of dollars being poured into further research and development.

[txlaw]

I'm sure you guys saw the news on INTC's bond issuance.  Pretty unbelievable rates.

 

http://www.sec.gov/Archives/edgar/data/50863/000104746912011005/a2212074z424b2.htm

 

Only 4.25% for the 30 year bonds.

[mcliu]

Any idea why Intel equity is so cheap?

 

Anyone have an updated thesis or perspective?

[StubbleJumper]

I'm sure you guys saw the news on INTC's bond issuance.  Pretty unbelievable rates.

 

http://www.sec.gov/Archives/edgar/data/50863/000104746912011005/a2212074z424b2.htm

 

Only 4.25% for the 30 year bonds.

 

 

Am I alone in thinking that people gotta be bonkers to buy a 30-year security in a technology firm and only get a prospective 4.25% return?  You're going to lose maybe 2-3% to inflation (if you are lucky), perhaps 1.5% to Uncle Sam for taxes on the coupons, and then you face long-term credit risk from a company that may not be a long term winner in the tech-race?  So, like at best you'll get your principal back with basically no return after taxes and inflation, and at worst a permanent loss of capital?

 

Am I too much of a chicken-ass pessimist?

 

 

SJ

[mcliu]

I'm sure you guys saw the news on INTC's bond issuance.  Pretty unbelievable rates.

 

http://www.sec.gov/Archives/edgar/data/50863/000104746912011005/a2212074z424b2.htm

 

Only 4.25% for the 30 year bonds.

 

 

Am I alone in thinking that people gotta be bonkers to buy a 30-year security in a technology firm and only get a prospective 4.25% return?  You're going to lose maybe 2-3% to inflation (if you are lucky), perhaps 1.5% to Uncle Sam for taxes on the coupons, and then you face long-term credit risk from a company that may not be a long term winner in the tech-race?  So, like at best you'll get your principal back with basically no return after taxes and inflation, and at worst a permanent loss of capital?

 

Am I too much of a chicken-ass pessimist?

 

 

SJ

 

You can probably say that about any bond issuance in general since they're all priced off of Treasuries.

[StubbleJumper]

I'm sure you guys saw the news on INTC's bond issuance.  Pretty unbelievable rates.

 

http://www.sec.gov/Archives/edgar/data/50863/000104746912011005/a2212074z424b2.htm

 

Only 4.25% for the 30 year bonds.

 

 

Am I alone in thinking that people gotta be bonkers to buy a 30-year security in a technology firm and only get a prospective 4.25% return?  You're going to lose maybe 2-3% to inflation (if you are lucky), perhaps 1.5% to Uncle Sam for taxes on the coupons, and then you face long-term credit risk from a company that may not be a long term winner in the tech-race?  So, like at best you'll get your principal back with basically no return after taxes and inflation, and at worst a permanent loss of capital?

 

Am I too much of a chicken-ass pessimist?

 

 

SJ

 

You can probably say that about any bond issuance in general since they're all priced off of Treasuries.

 

 

Yep, but I'm more comfortable that KO, MCD, BAC, WFC, CAT, MMM, etc will not experience financial distress at some point in the next 30 years resulting in a permanent loss of capital.  I have a great deal more difficulty with the assumption that over the next three decades nobody is going to usurp INTC's market leadership.  It may very well work out that INTC will continue to be dominant for 30 more years, but I'd want to be adequately compensated for making that particular bet!

 

Cheers,

 

SJ

[ItsAValueTrap]

Am I alone in thinking that people gotta be bonkers to buy a 30-year security in a technology firm and only get a prospective 4.25% return?

IBM has 100-year bonds.  (!!!)