SAN JOSE Ė To use Andy Groveís well-worn phrase, Intel Corp. is at an inflection point. Itís time for the company to re-think what it wants to be in the next phase of its life.
One of its options is to strategically pivot. For years, Intel has led with an identity as a microprocessor designer that has great chip-making capabilities. Maybe itís time to lead with its left foot. Intel could be one heck of a foundry that happens to have its own line of very successful products.
Face it, the market has shifted from the desktops Intel dominates to tablets and smartphones where it barely participates. But world-class semiconductor manufacturing is as valuable as ever, even as Mooreís Law slows.
Here, Intel is still tops. First with high-k metal gate transistors. First with FinFETs. Tons of capacity in leading-edge fabs all over the world. Thereís little doubt it will be first to field the extreme ultraviolet lithography that is key to next-generation processes.
Iím not saying Intel should pull the plug on processors. It has a huge position with the x86 today and is doing a reasonable job playing catch up in the new game of SoCs. This is just a re-balancing. The corporate weight shifts from the front to the back foot.
Iím not alone in thinking this way. Jim Turley of Silicon Insidersees Intelís x86 on a slooooow decline along with the PC market and says itís time for a change. ďIntel needs a piece of good news, something that shows they have caught on to next-generation products, not hoping PCs will make a comeback,Ē Turley said.
The shift may already be in the works. Intel has been making chips for a handful of mainly small FPGA companies for a couple years.
ďI think they used those deals as training wheels, a trial run for a couple years before they would consider taking on some big customers,Ē Turley said,
Samsung is a TOTALLY different case. They don't compete with Apple by making ARM SoCs, they compete with Apple by selling smartphones.
The article is merely suggesting that Intel make ARM SoCs for Apple, while selling x86 SoCs to Apple's competitors. No one is suggesting Intel should start selling smartphones.
Otellini's leaving because he's a wise man. Tough economic times, even tougher semiconductor market, competition becoming thornier than ever, he knows the semiconductor industry has some major contraction ahead. He doesn't want his legacy tarnished by something he doesn't have control over.
You know, maybe he is leaving because he is really rich, has only a few years left in this world, and wants to enjoy the rest of his life in relative peace?
I've been paying attention the last few years to how old CEO's are when they retire. Early 60s is the most common answer.
Intel already have the capabilities in designing their chips for FPGA usage which diversifies their functions in a chip. I had rumours on a Cedarview Development Board with FPGA capabilities. Also, not to mention the embedded automotive market that they are so dearly focusing on, especially the IVI projects they have been doing for a few companies for the past few years. However, as for mobile phone and tablet market, they might not have a strong foothold with Qualcomm, Samsung and Nvidia app processors flooding the market.
Intel is used to making custom chips for its custom process. I think in the short term Intel can have more success in making ASICs (making chips for other people, using its design Library) rather than Foundry (let fab less design using there Design kit) .
It's a great idea but Intel is its own industry sector---they do everything in-house, from basic architectural and physical design, through their own EDA/chip design tools that are tuned to their fab process, and of course their own fabs. It will be a challenge to integrate some pieces from third-party workflow--it probably isn't as simple as reading a VHDL from ARM and synthesizing a foundry output file.
While the obervation about shrinking PC market is valid, I see two problems with the proposed shift. First, Intel has never played in a highly competitive market with razor thin margins, which is basically what the cell phone and tablet market is. Yes, they have had to contend with competition from players like AMD, but they have always released innovative products that had good margins (not to mention high volume, which was also helpful).
Second, they have consistently had about the industry's worst track record on power efficiency--it just wasn't as high a priority in the desktop segment. Combine this with the fact that they have never done well in the embedded market, either. I'm not sure battery operated phones and tablets are an area where they are going to leverage much expertise. That may be where the volumes are headed, but whether they could succeed technically or financially in that arena is an open question.
I don't see why Intel would give access to its only key differentiator in the market. Especially to apple, so apple can compete with a better phone and limit use of Intel's own atom based phones? Doesn't make sense.
Now what Intel might do is to convince apple to use an x86 CPU in its next-gen processor in exchange for its fab. Now that would be a great strategic move. This way they have their cake and can eat it too.
Long term this will be a bad strategy for Apple as they will get an inferior mobile CPU and be held hostage by the big monopoly.
Grove had vowed that Intel will never move into consumer products as lesson learned from their Microma digital watch business in 1972. Why they should run Atom based smart phone business?
To make the new era foundry business successful meeting Intel's current margin now, they need to do:
1. Better architecture to balance power and performance. Arm architecture had done a good job in power targeting mobile application and is moving aggressively toward server and other applications by beefing up the computation muscles. What is the answer from Intel? How to integrate more function to provide Apple/Cisco with better mobile CPUs than Arm core?
2. They could adding more value than just mfg. foundry with design service and provide complete set of library. They can further optimize design and technology with thorough DFM consideration which is badly needed by advanced technology nodes. These are the advantages that Intel has against mfg. foundry giants. The design service could also add more value to customers and deserve higher margin.
3. Better customer oriented service: learning from foundry king: TSMC. They need to be humbled and attentive to customers needs not just "to teach customers how to use their technology".
4. Stringent cost reduction mind set in mfg. line. Copy exactly might need to be revamped like copy smartly with some cost consideration. Dumb fabs operation should be injected with more engineering talent to fine tune equipment and process for lower cost operation.
If Intel were to makes chips for Apple it will be entering a sub 35% Gross Margin Business. In late 90s Intel exited the ASIC business for the same reasons, namely poor GM.
if you take a look at what Apple pays Samsung to manufacture its A5[A6 etc] processors, they are getting a Superb Deal;
Would I expect Apple to pay more ? Maybe 1-3% more but not 25-30% more ?
On the other hand does it makes sense to lockout Asian foundries from the Sub 28 nm Market ? Assumption being the Ciscos, Qualcomms and Junipers/Broadcom would crank out new ASICs in an Onshore Intel Fab ; this would require a Major Protectionist Policy change which will affect the Semi Equipment Mfrs;
Will it happen ? Unlikely unless it is championed by the New Horsemen [Jacobs, new Intel CEO, Ellison of the Semi space;
Intel, what is your timeline...???
Foundry biz a logical way to go short term, yes.
Platform biz under Otellini's watch...failed.
Why not do a real "leap" and use your market brand and tech strengths to sell your own mobile products...??? Compete with your customers, yes so what. Will your home grown products resonate with end customers...heck yea.
If you do...I will invest...
On QUANTA Intel is sitting on lots of undeclared supply going know where, declared and written down, would be the end of Intel. But since it was never declared in the 1st place . . .
Currently, 100 runs Pentium to Ivy Bridge, 29.59% of production is priced below cost. That loss is higher on Celeron deleted from assessment on no monopoly overcharge values. Where Intelís industrial financial theft from AMD, Cyrix, National, Nexgen, IBM, IDT, other processor, specialty memory design fabricators, would grow in excess of $62,318,101,609 now calculated.
20.59% of Intel production priced below cost opens the foundry opportunity for Intel to displace that loss with effective fabrication that is cost price efficient. Intel accounting group would need to conduct a real audit of production output. Or the production group would need to replace the entire Intel accounting group. Regardless there will be no foundry agreement where licensing companies cannot count on the counting house for their revenues.
Itís fairly obvious that ARM will attempt to negotiate a foundry license with Intel that will certainly keep a lot of lawyers busy on all the reconciliation between AMD, Intel, and foundries established on organic best practices. Iím not opposed to Intel foundry except that the monopolist Intel should be given no advantage over any foundry who established their position on organic best practices. Which means lots of cross licensee provisionís as Intel has nothing else to trade with foundries established on organic best practice.
Otellini left because he no longer has credibility with any other nation state including Intel. Atom is his industrial success where Intel surely needed to reclaim ground in that category. Otelliniís involvement in the sales racket was his end. He has no credibility with any legitimate governance institution and rebuilding of his reputation will be required in any occupation other than retirement
On Apple and CISCO; Apple certainly seems to know how to negotiate for Intel procurement, despite the fact that Intel owes Apple system end buyers a considerable sum in monopoly overcharge values for top of line Mac Pro.
On CISCO like other communications switch gear, Intel product that was good enough for the application has always been available through secondary sources, at such incredibly low prices in relation to Intel cost, who could go wrong? The issue is Intel must now compete for that direct sale on cost price and Intel cost is going up.
Perhaps 32 nanometers will be good enough for communications switch and route?
15 years ago I worked for a company playing in the consumer market. No one would invest as it was such a dog of a space. Everyone was in IP and telecom. That space is now over subscribed. Intel have the crown jewels and need to remain strong and not move away from this. They have enpough cash and clout to survive. Apple do not invesst in core technology so the Apple lovers need them to buy it to stay ahead. Soon Apple will be a fad again with a few million lovers prepared to pay because they believe they are getting something better. Every consumer electronic company tha did this looses out eventually to a faster moving rival. Apple have already done this once, look to Sony etc for the currents. In Summary, Intel keep it in house
PC for comsumer Market is dying faster than anaything, even MS has moved to make its own tablet and is providing software for smart phones as well.
So Intel has to look past PC, but I don't think it is a good idea for Intel to go into making its own systems, Intel is no IBM,
It should however, work with Big US fab less companies to supply a superior IP & chips.
It needs to build a new eco system.
Even EETimes has this rumour:
Do INTEL really want to make the Ax chips that would replace Intel own x86 chips?
Intel also wants to enter mobile markets, then why it should give "advantage" of its foundry leader to its main rival in the mobile markets? Especially when your clients is a high profile ARM licensor?
Cisco is different market than what Intel is after. That make more sense. But Qualcomm certainly NOT!
Whoever done a global research in PC market decline? Nobody!
Everyone is picking-up story of less PC sold (based on Q-report of the PC companies), tabulated by analyst companies, and they make conclusion "the PC is DEAD!". Sigh!
And i still make report, generate customers support system on a PC. When it dies, i will buy another PC. Not mobile phone. Not tablets.
Unless the computer looks like the one in Tom Cruise "Majority Report", i still need my PC.
The number of chip manufacturers keeps dropping as do the number of competing designs. I wish Intel would find a way to effectively compete, from its own strengths, with the mobile chip manufacturers. Charlie Babcock, a senior writer at InformationWeek
The track record for Intel on power-efficiency has always been not good but their processing power is superb. However, this doesn't suit the mobile phone and tablet market. ARM on the other hand plays a different ball game, it is a better choice in battery-powered market as it started the design with an objective of low-power and by slowly adding instruction sets and processing power, it will soon be a major player in the server and pc industry because they are not over-designing and making their chips more expensive and power-hungry. Intel has a tendency of over-designing and adding component costs to achieve the processing power.
kenny.g -- you never held a 200LX? In 1991 it runs more software than smartphone do today and did it for 3 weeks on two AA batteries. The market can belong to Intel if they make the right decisions on mobile x86.
And Intel does not need to buy companies... only to provide them with foundry service... i.e. Acronix, Telit -- just hope they start taking advantage of that synergy on x86. But if they just focus on foundry they giving up the margin on end products.
In a conversation at IEDM yesterday, Intel's Mark Bohr suggests the company does have some larger volume foundry customers online or coming.
But he is also clear he expects this will never be a huge volume business for Intel or a general-purpose offering.
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.