The Case Against ARM

[wallisweaver]

1.) No presence in servers.
2.) No presence in PCs.
3.) Tablet sales declining.
4.) Sale of high-end smartphones declining.
5.) New fabrication nodes cost too much for too little in improvement in performance/power efficiency.
6.) ASPs and margins declining.
7.) Broken foundry model doesn't fund R&D or CapEx.
8.) Absence of high end capacity.
9.) Apple is the one big performer and it seems that they have now run out of gas.
10.) IoT not growing fast enough to sustain rate of growth as mobile market matures.
11.) ARM stock price has been stuck under $55 since 2013 in spite of Wall Street doing all they can to boost the price above $55 one more time so they can get their money back.
12.) Qualcomm struggling with overheating/performance issues.
13.) Samsung's quarterly profit plunged 39 percent in the last earnings report.
14.) TSMC struggling with the 16nm node.
15). The foundries are one Intel-Apple deal from going off the road and into the boonies

 

[benb]

The case for ARM (??):
1. Low cost model suits the industry direction
2. An ecosystem of reusable generic IP suits the industry direction
3. They provide a counter-weight against titans
4. They somehow simultaneously counter-balance and are promoted by titans
5. The whole industry is shot, the negatives for ARM...are less negative

 

[Daniel Nenni]

1.) No presence in servers.

This is going to change. The Intel monopoly will be challenged and Intel will either lose market share or margin or both. The technology market is against monopolies.

2.) No presence in PCs.

Very true but tablets will get bigger and MACs will be ARM based in the near future.

3.) Tablet sales declining.

This goes against Intel as well.

4.) Sale of high-end smartphones declining.

Not according to IDC:

https://www.semiwiki.com/forum/f302/apples-china-pursuit-6022.html

5.) New fabrication nodes cost too much for too little in improvement in performance/power efficiency.

That is not the case. This is another monopoly issue. Now that TSMC has competition from Samsung costs and prices will go down. TSMC 16FFC is a low cost version of 16FF+. Less mask sets etc...

6.) ASPs and margins declining.

True for the chip companies but ARM still gets their royalties and fees.

7.) Broken foundry model doesn't fund R&D or CapEx.

This is just stupid. The combined R&D and CAPEX numbers of the fabless semiconductor ecosystem is staggering. Apple can write some big fat checks, Samsung too, and other systems companies. Intel is the one that should be worried about R&D and CAPEX.

8.) Absence of high end capacity.

Not true. Samsung and TSMC lead 300mm wafer capacity.

9.) Apple is the one big performer and it seems that they have now run out of gas.

Not true. Go ahead and short APPL and see what happens. iPod - iPhone - iTV - iWatch - iHome - iCar........

10.) IoT not growing fast enough to sustain rate of growth as mobile market matures.

Not true. IoT and wearables is moving faster than expected anyone really expected. The iWatch for example. Another boost for high end phones.

11.) ARM stock price has been stuck under $55 since 2013 in spite of Wall Street doing all they can to boost the price above $55 one more time so they can get their money back.

I can't comment on stocks. Anybody else?

12.) Qualcomm struggling with overheating/performance issues.

Qualcom switched from proprietary 32 bit SoCs to ARM core based 64 bit ones as a stop gap until their proprietary 64 bit SoCs were ready. This is a non issue. Apple caught them by surprise with 64-bit and QCOM had to play catchup.

13.) Samsung's quarterly profit plunged 39 percent in the last earnings report.

Not relevant.

14.) TSMC struggling with the 16nm node.

Not true. TSMC 20nm and 16nm will both break ramping records.

15). The foundries are one Intel-Apple deal from going off the road and into the boonies

Apple will NEVER fab at Intel. They went from Samsung at 28nm to TSMC at 20nm and Samsung/TSMC at 14/16nm and will go back to TSMC at 10nm. Intel is no where in sight.

 

[hist78]

1.) 2.) No presence in PCs./QUOTE]

The overall PC market doesn't grow too much and even was declining sometimes in recent years. ARM's presence in PC market is not critical.

 

[JohanDijkhuis]

ARM is more than high end tablets / smartphones. It can be used virtually everywhere where some compute power is need for very little power. In 40 nm CMOS a processor is smaller than an inductor, process spread in deep submicron CMOS is large, so almost all mixed signal chips use calibration. Assume that all mixed signal chips in 90 nm or below have a processor (in my experience mostly ARM, e.g. cortex M0), those that don't have will be compensated by those having more than one. This might not be a high profile business, but it is high volume.

 

[m210658]

Never say never

1.) No presence in servers.


15). The foundries are one Intel-Apple deal from going off the road and into the boonies

Apple will NEVER fab at Intel. They went from Samsung at 28nm to TSMC at 20nm and Samsung/TSMC at 14/16nm and will go back to TSMC at 10nm. Intel is no where in sight.

You may be right for now, but ...

 

[Brianhayes]

The case for ARM is that it has increased its market share from 17% in 2007 to 37% in 2014. The case against is that this rate of growth will at some point taper off. When? Who knows but not this year IMO..oil

Since I invested in ARM four years ago I have read countless "Intel will smash ARM" blogs and, to be frank, this one is shallower than most. Intel may well fab Apple's A11, A12 or 13 or whatever but ARM will still get its royalty that is unless Intel supplies Apple with x86 chips at a small fraction of the cost that they now charge the PC and server markets. For example Krsanich boasted in November 2014 that Intel were producing 400 million units and the depreciation charged in their accounts for 2014 was $8.5billion giving a cost of capex over 2x TSMC's ASP in the smartphone market. I don't think Intel will reduce its prices to make that kind of loss.

 

[ippisl]

>> 1.) No presence in servers.

Paypal recently started to use ARM servers. Says it's a big saving on TCO. There's a whole ecosystem starting to form. Also china is developing a very strong dislike to intel with government "encouraging" companies to use use other architectures, for example chinese chips built with "open power". So using open design could become a global trend, and it's good for ARM, and very bad for intel.

>> 2.) No presence in PCs.

Chrome-os based devices sales are growing, some are ARM. Google is serious about that.Soon they'll run android apps , and you could buy cheap compute stick that runs it(with ARM). I'm pretty sure this stick will be much better than intel's recent crummy "compute stick".

>>7.) Broken foundry model doesn't fund R&D or CapEx.

Not broken, and it'll probably see huge growth(extremely fun application that requires high compute) when virtual reality becomes popular because it depends mostly on good gpu's and nvidia/amd, leaders in this area, are based on the foundry model.

 

[benb]

One thing that impressed me recently with ARM: Although the A57 is a dog, they quickly fixed the problem with A72. Other companies such as Qualcomm, Intel, AMD and to some degree Nvidia have not shown this nimbleness.

 

[lefty]

One thing that impressed me recently with ARM: Although the A57 is a dog, they quickly fixed the problem with A72. Other companies such as Qualcomm, Intel, AMD and to some degree Nvidia have not shown this nimbleness.

Yeah, the A72 seemed to fix all the problems with the A57. I wonder if AMD's upcoming K12, or Qualcomms next ARM core will become redundant because of the A72.