Apple would be crazy to put all eggs in one nest
here's some comment from ASML about TSMC expansion plans (from seeking alfa UBS presentation):
note TSMC 16nm FinFet = 20nm design rules - once again die shrink somewhat muted.
Due to the long lead times ASML is probably the best early indicator - if Apple decided to go with TSMC than they must have make the decision a long time ago ...who knows ...perhaps today they would decide differently
- the transcript is only 4 pages - so check it out
So what I believe whether it is a 20 nanometer plainer or the 20 nanometer or 16 nanometer what TSMC is calling. But I think as far as I know, it’s actually 20 nanometer design rule with the finFET.
From a litho perspective, there is no real difference between the tools that required for the plainer or the finFET process. So the combined 20 nanometer and 16 nanometer again whatever it’s going to be 20 nanometer or 16 nanometer, now will drive the 1.7 time that the spend over the 28 nanometer node
So yeah, we don’t know I have an indication today on what the ramp of the 20 nanometer process will look like currently. But they’ll have to install a critical amount of capital. So actually there are some internal estimates on our part again not necessarily linked to what our customers have said, but most just looking at typical ramps of new nodes, they might suggest that there could be a fab work or about 40,000, 45,000 wafer starts to put in, which in the mid 2013 and mid 2014, and 45,000 wafer start fab at 20 nanometer is about €1 billion in little investment. Again versus the 600 or so for the 28-nanometer investment, it’s quite substantial.
"Apple would be crazy to put all eggs in one nest"
[Let's see what happens with the iWatch. It should be a good litmus test of which way Apple intends to move. If they really are working with Intel on it, it would indicate strongly that there is more to come...]
Sentiment: Strong Buy
Years ago, Steve Jobs made it clear that he did not want Intel as a supplier. He indicated that they were too slow and "stodgy", I believe was how he described it. He is gone and perhaps things have changed. If Intel's new CEO comes from outside and has a "mobile" background, that will signal that the company is going to change course. If, on the other hand, they go with an insider like Stacy Smith, that will signal something entirely different to the market. I have owned the stock for about 8 years and we need some mgmt shakeups to push the share price northward
Problem is Apple doesn't have a choice. If they want to improve their phones performance, they have to go to the 20nm process node, and there is no guarantee TSMC or anyone else can do it. So they are stuck with 28nm which means NO IMPROVEMENT in their next phone.
Intel on the other hand with their 22nm FinFet's can provide huge performance and battery life gains, over their current ARM chips.
Its a BUSINESS decision, not that Intel is stodgy or not.
If Intel's new CEO comes from outside and has a "mobile" background,
what do you mean by mobile background - you mean low power?
Apple needs to look at many aspects when selecting a manufacturing partner -
and it's probably a trade off / compromise.
"He indicated that they were too slow and "stodgy"
Intel selected 3 small top notch fabless companies and demonstrated that they can efficiently work with the design teams of the 3 companies and can bring (just announced by Achronix) a leapfrogging chip into production. I think these 3 companies were a learning vehicle for Intel because they need to work with and listen to the outsider design team.
I think that is a very important skill Intel might have lacked in the past but this has been fixed by working with the outsiders.
Many might disagree with me but P.O. laid the foundation (granted somewhat late) for mobile.
As Bohr said Intel neglected SoC but they are making great effort to catch up.
Though SoC and CPU are two different beasts Intel can leverage the process technology and learning from CPU processing and apply it to SoC manufacturing which is a huge advantage compared to ARM foundries.
A fab can not change from CPU to SoC production on the fly - very sophisticated S/W is used to run a fab and it takes some time to make adjustments.
Though SoC and CPU use the same tool set the process flow differs between CPU and SoC.
From: Intel Tips 22nm SoC Recipes, 14nm Process
google it - it gives comparision between SoC and CPU
Bohr also discussed Intel’s SoC process recipes for the 22nm node. In previous years, Intel developed a standard “one-size-fits-all” CPU process technology for a particular node.
Starting at the 32nm node, Intel developed a CPU (P1268) and SOC (P1269) process. The CPU and SoC processes have identical feature sets, but the SoC version incorporates a set of recipes specifically for device designers.
For example, Intel will provide a standard CPU process at 22nm, internally called P1270. Based on a tri-gate transistor structure, the CPU process incorporates high-speed logic circuits and interconnects.
The SoC version, dubbed P1271, makes use of a low leakage technology, dense interconnects and passives. It will also include 1.2V low-power and 1.8V thick-gate options.
Intel will provide some four SoC recipe options for designers: high-performance, standard performance, low power and ultra low power, according to Bohr.
The technology is aimed for a range of applications, including the booming mobile space. Intel is seeking to propel its x86-based processors in the mobile space-and displace the ARM camp in the process.
There is no chance of apple using atom in phones. Look at the performance difference between apples own swift core to atom, then look at the die size difference. Unlike android iOS is a native environment, meaning you have an large ISA hurdle to jump over.
Intel's 22nm redesigned Atom will leapfrog anything from ARM. As for porting IOS apps, It just requires a recompile and test (3 months at most). Apple maintains a tight control of Apps, so its possible to port all the apps to x86 in 3-6 months. Apple did it with Mac OS, they'll do it with IOS.
The benefits of going Intel far outweigh the benefits of staying with the DEAD END Arm.
I did a search on Samsung/Glofo FinFet SEM - I did not find anything - just schematics how it supposed to look like (I searched because some claim the shape of the fin would be different compared to Intel - no real picture!!! Not one - just schematics ... perhaps semiconductor guy can come with some SEMs/cross sections).
Think about the box/cube - memory cube = big #$%$ to reduce overall power - not only the SoC
I am curious - what criteria do you think Apple is applying when selecting a manufacturing partner?
Do you think they could actually consider future 450mm wafer capabilities down the road? -
AMD u is correct - it's a very lengthy process
That would require AAPL to have a smart business sense. They don't have any sense of urgency as they have 139 billion in reserve. Now if Timmy & co. were to make a play for INTC, that would put everyone else in a world of hurt. Little chance, unless Einhorn puts the screws to him and AAPL decides to spend the cash rather than give it back to the stockholders.
It takes a couple of years to transfer from one chip to another. Apple has NO CHOICE but to go with Intel, if they don't want to be a mediocre player like the rest of the ARM phones. No other fab has the capacity and the technology that Intel has.