TSMC exclusively doing Apple A10

[jpiniero]

According to this referenced article:
http://www.chinatimes.com/newspapers/20150914000039-260202

Unlike the A9 (which was shared with Samsung but NOT GloFo since GloFo sucks) the A10 will be produced exclusively at TSMC. Starts in March 2016 and still on TSMC's "16" nm process.

Psst, nVidia: You might want to get something in at TSMC before they start on the A10.

 

[Trumpstyle]

This is old news. Semiwiki already said tsmc will do a10. A9 will be samsung 14 nm and a9x will be tsmc 16 ff+.

 

[Arachnotronic]

This is old news. Semiwiki already said tsmc will do a10. A9 will be samsung 14 nm and a9x will be tsmc 16 ff+.

Are you so sure that A9 isn't also TSMC?

 

[ShintaiDK]

TSMC got 30% of A9 after GloFo flopped. And 100% of A9X.

While TSMC gets all the A10. Samsung may get the A10X.

At the same timeframe Intel will get 30% of the modem allocation with the 7360 with Qualcomm for the rest 70% with their 9645.

 

[Trumpstyle]

This is from the latest tsmc quarterly report about 16 finfent for the third quarter:

Lora Ho - SVP, CFO Third quarter will be very small single digit revenue.



Morris Chang - Chairman
16 nanometer, it's very little.





Pretty obvious they lost the iphone

 

[Guest1]

16nm two years in a row? Wow things are slowing down in the fabless world. Looks like they are going to be here for a few years while they recoup their costs. Sucks to be Samsung losing all that Apple volume in exchange for Qualcomm. TSM must be thanking their lucky stars to get Apple volume now they can recoup their costs. I wonder if they have any capacity left to bring on additional customers. Maybe a Mediatek with the Helio X30.

 

[freeskier93]

This is old news. Semiwiki already said tsmc will do a10. A9 will be samsung 14 nm and a9x will be tsmc 16 ff+.

What? No. A9 is 16 nm, it would make no sense for A9 to be 14 nm only to go back to 16 nm with the A10.

 

[III-V]

I'm sure TSMC will be fine, with or without Apple going forward. They have a lot of legacy node revenue, their 28 nm was great, their 20 nm started out extremely strong and remains strong still. You don't have to spend much time in bed with Apple at all to see your profits go through the roof.

They could skip 16/14 nm altogether, and everything would still be just peachy. They had 11B cash on hand in 2014 (not sure if that was the beginning or end of 2014 -- doesn't matter). Of course, it's a slippery slope from being on top of the world to being fed to the animals, but TSMC hasn't gotten to the place they're at now by being incompetent.

I'm not well-read on Samsung's performance in the fab industry, but it certainly doesn't matter for them either. When you're so far ahead of everyone else in the food chain, mistakes are affordable.

The only real scare for either of those companies is if Intel manages to gain traction in fabbing for others. If Intel had Apple, it'd be game over -- discussion of Samsung vs. TSMC is rather silly in that light. Intel may have had a rough start, with 14 nm not going quite as smooth as planned, but again, they're still very new to the game and still very dangerous.

 

[Guest1]

I'm sure TSMC will be fine, with or without Apple going forward. They have a lot of legacy node revenue, their 28 nm was great, their 20 nm started out extremely strong and remains strong still. You don't have to spend much time in bed with Apple at all to see your profits go through the roof.

They could skip 16/14 nm altogether, and everything would still be just peachy. They had 11B cash on hand in 2014 (not sure if that was the beginning or end of 2014 -- doesn't matter). Of course, it's a slippery slope from being on top of the world to being fed to the animals, but TSMC hasn't gotten to the place they're at now by being incompetent.

I'm not well-read on Samsung's performance in the fab industry, but it certainly doesn't matter for them either. When you're so far ahead of everyone else in the food chain, mistakes are affordable.

The only real scare for either of those companies is if Intel manages to gain traction in fabbing for others. If Intel had Apple, it'd be game over -- discussion of Samsung vs. TSMC is rather silly in that light. Intel may have had a rough start, with 14 nm not going quite as smooth as planned, but again, they're still very new to the game and still very dangerous.

Depends on your definition of fine. If they are no longer going to be bleeding edge yes I agree. If you are saying they will be competing with Samsung and Intel no they are not. They will soon be N-1. I believe one of the reasons Apple constantly switches between the two is to keep them on life support for leverage against each other. Sooner or later they will run out of stamina to compete with Intel at the bleeding edge. I predict TSM will bow out before Samsung. Here's a very good financial reason why via seeking alpha.

"The foundry industry is spending about $25 billion per year to supply Apple and QCOM with about a total of $12 billion per year in product ($7 billion to Apple and $5 billion to QCOM) That can't work. The author should write about what a bankrupt and hopeless strategy that is. Then to make matters worse, Apple bounces their business, (or plans to) from TSMC to Samsung and back again. How can that work?"
-Russ Fischer

The fabless model is dying a slow death at least with regards to bleeding edge business. 14/16nm may be the last node Samsung/TSM launches on time.

 

[Arachnotronic]

Depends on your definition of fine. If they are no longer going to be bleeding edge yes I agree. If you are saying they will be competing with Samsung and Intel no they are not. They will soon be N-1. I believe one of the reasons Apple constantly switches between the two is to keep them on life support for leverage against each other. Sooner or later they will run out of stamina to compete with Intel at the bleeding edge. I predict TSM will bow out before Samsung. Here's a very good financial reason why via seeking alpha.

"The foundry industry is spending about $25 billion per year to supply Apple and QCOM with about a total of $12 billion per year in product ($7 billion to Apple and $5 billion to QCOM) That can't work. The author should write about what a bankrupt and hopeless strategy that is. Then to make matters worse, Apple bounces their business, (or plans to) from TSMC to Samsung and back again. How can that work?"
-Russ Fischer

The fabless model is dying a slow death at least with regards to bleeding edge business. 14/16nm may be the last node Samsung/TSM launches on time.

I think you dramatically underestimate TSMC and overestimate Samsung. Also, the problem with Russ' analysis is the assumption that TSMC + Samsung are spending $25B in capex solely for the Apple/Qualcomm business.

The fabless semiconductor industry is far larger than "just" Qualcomm/Apple.

 

[Guest1]

I think you dramatically underestimate TSMC and overestimate Samsung. Also, the problem with Russ' analysis is the assumption that TSMC + Samsung are spending $25B in capex solely for the Apple/Qualcomm business.

The fabless semiconductor industry is far larger than "just" Qualcomm/Apple.

Ok show me who else is launching 14/16? AMD is not launching until next year and with sales of a whopping $1 Billion that's not going to move the needle not to mention they are going with GloFo. Nvidia? Mediatek Helio is launching on 28nm. 808 is 20nm. The other great expanse of business you are referring to are from the budget vendors who want cheap but functional that sounds more like 28 than 14/16. If you are TSM are you going to dump money made from legacy nodes into bleeding edge R&D when it is going to take years to get that money back if ever and on top of that you have to fund development for the next node? The fabless business is far larger than Qualcomm/Apple but those two are the only ones willing to pay for bleeding edge nodes. You think a rock chip or Mediatek is going to do it? So all that bleeding edge will only be for two maybe three clients if you include Samsung selling to itself. Not to mention the premium SoC market is not growing nearly as fast as entry and midrange. Bleeding edge for the fabless crowd is not sustainable.

 

[ShintaiDK]

There really isnt much business outside Intel/Samsung/Apple/Qualcomm on 14/16nm. The cost barrier for designs is too high. Sure some other companies will eventually join the club. But some of those may simply lose money on it. While others can do it because they can sell the chips at up to 25000$ a pcs.

Lets see on semiconductor sales:

Accounting for current deals:

The problem here is that most of these products cant pay for it at all. It basically needs to be MPU revenue. And thats where the ball drops for most companies. This is also why Intel wins hands down in the economic aspect. With over 60% of all world MPU revenue. While the second place gets around 10%.

 

[dark zero]

That would be the last good chip (A9X) from Apple for a while.... FinFet at 16 nm will be a really BAD move.... and knowing how TSMC are "good" in their deliveries.... expecting TONS of delays.

 

[Idontcare]

This is from the latest tsmc quarterly report about 16 finfent for the third quarter:

Lora Ho - SVP, CFO Third quarter will be very small single digit revenue.



Morris Chang - Chairman
16 nanometer, it's very little.




Pretty obvious they lost the iphone

How many 16nm wafers do you think they will have shipped in Q3 even if they have the entire A9/A9X contract considering that the A9/A9X will only have been shipping in market for 5 days in Q3? (Apple release is Sept 25, Q3 ends Sept 30)

If people really wanted to know who is producing the A9, it is very easy to find out. Just come to Asia and visit the fabs, you'd be amazed how much you can learn just by sitting in restaurants at lunch time and listening to conversations between fab engineers.

My curiosity is why does anyone in these forums care whether the A9 is being produced at TSMC or Samsung?

If you are an employee of TSMC or Samsung then sure, I would be concerned from a job security standpoint. Or if you are a shareholder of either company then it becomes a profit motivation.

But outside the money angles, do people chatting on a forum really care whether or not the A9 SOC is being predominately (or solely) produced at TSMC or Samsung?

Is it a fanboy/psychological thing? Do we really have Samsung and TSMC fanboys nowadays?

Honest question, because I don't get it. I do, however, happen to live in Asia and work in the industry, and it is quite clear who has which contract and why. But who cares, and why?

 

[scannall]

How many 16nm wafers do you think they will have shipped in Q3 even if they have the entire A9/A9X contract considering that the A9/A9X will only have been shipping in market for 5 days in Q3? (Apple release is Sept 25, Q3 ends Sept 30)

If people really wanted to know who is producing the A9, it is very easy to find out. Just come to Asia and visit the fabs, you'd be amazed how much you can learn just by sitting in restaurants at lunch time and listening to conversations between fab engineers.

My curiosity is why does anyone in these forums care whether the A9 is being produced at TSMC or Samsung?

If you are an employee of TSMC or Samsung then sure, I would be concerned from a job security standpoint. Or if you are a shareholder of either company then it becomes a profit motivation.

But outside the money angles, do people chatting on a forum really care whether or not the A9 SOC is being predominately (or solely) produced at TSMC or Samsung?

Is it a fanboy/psychological thing? Do we really have Samsung and TSMC fanboys nowadays?

Honest question, because I don't get it. I do, however, happen to live in Asia and work in the industry, and it is quite clear who has which contract and why. But who cares, and why?

Best post I've seen in a while. What should matter is how the device performs, not which foundry it came out of.

 

[ehume]

. . . and it is quite clear who has which contract and why. . .

OK, who has what contract, and why did each win the contracts they did? Technology? Price? Is the 16/14nm actually at 20nm, and comparable to Intel's 22nm?

We care because we buy the devices these things go in.

 

[Thanatosis]

How many 16nm wafers do you think they will have shipped in Q3 even if they have the entire A9/A9X contract considering that the A9/A9X will only have been shipping in market for 5 days in Q3? (Apple release is Sept 25, Q3 ends Sept 30)

If people really wanted to know who is producing the A9, it is very easy to find out. Just come to Asia and visit the fabs, you'd be amazed how much you can learn just by sitting in restaurants at lunch time and listening to conversations between fab engineers.

My curiosity is why does anyone in these forums care whether the A9 is being produced at TSMC or Samsung?

If you are an employee of TSMC or Samsung then sure, I would be concerned from a job security standpoint. Or if you are a shareholder of either company then it becomes a profit motivation.

But outside the money angles, do people chatting on a forum really care whether or not the A9 SOC is being predominately (or solely) produced at TSMC or Samsung?

Is it a fanboy/psychological thing? Do we really have Samsung and TSMC fanboys nowadays?

Honest question, because I don't get it. I do, however, happen to live in Asia and work in the industry, and it is quite clear who has which contract and why. But who cares, and why?

Oh look, a page long post by idontcare implying he knows the answers to our questions but offering no information and then posing the question of why, here on one of the oldest tech discussion forums on the internet, do we "care" who manufactures what SoC or which process it may be on. Why indeed?

Well, for starters, you pretty much answered your own question in the first few lines:

If you are an employee of TSMC or Samsung then sure, I would be concerned from a job security standpoint. Or if you are a shareholder of either company then it becomes a profit motivation.

Many people on these forums would fit one of those two qualifications, so there you go. But after answering your own question you then go on to call everyone in this thread "fanboys" for giving a crap. I can see how you became an "Elite" member, you're clearly way too smart and too important to even bother to answer us "fanboys". Why did we even ask?

Why did we make this thread? Why is there even an overclocking and CPU forum when we can all rely on the benificent and all knowing genius that is Idontcare? What fools we were.


I have no idea why anybody listens to you, you clearly are out of the industry. Anybody who can still call intel 14nm a decent product or even on par with samsung 14nm in performance can't have worked that recently.

 

[zebrax2]

I can see some merit on discussing on who produces what as it could indicate the state of the node.

 

[Arachnotronic]

I have no idea why anybody listens to you, you clearly are out of the industry. Anybody who can still call intel 14nm a decent product or even on par with samsung 14nm in performance can't have worked that recently.

Idontcare's posts are always interesting, insightful, and add significant value. That's why people listen to him.

BTW, kind of strange that somebody who just joined these forums in August and has a mere 13 posts under their belt speaks as though they've been here a while. What alias did you post under previously?

 

[Borealis7]

Oh look, a page long post by idontcare

good, i like those

as for why I care? i care because more capacity goes to SoC means less capacity goes to better CPUs and GPUs, which I spend much more money on than SoCs in their phone-packaging. less capacity means products come later, and probably cost a bit more.

Many people on these forums would fit one of those two qualifications

most outlandish claim of the year. i would bet less than 1% of the CPU forum are intel/amd/samsung/whatever employees.

 

[NTMBK]

Oh look, a page long post by idontcare implying he knows the answers to our questions but offering no information and then posing the question of why, here on one of the oldest tech discussion forums on the internet, do we "care" who manufactures what SoC or which process it may be on. Why indeed?

Well, for starters, you pretty much answered your own question in the first few lines:


Many people on these forums would fit one of those two qualifications, so there you go. But after answering your own question you then go on to call everyone in this thread "fanboys" for giving a crap. I can see how you became an "Elite" member, you're clearly way too smart and too important to even bother to answer us "fanboys". Why did we even ask?

Why did we make this thread? Why is there even an overclocking and CPU forum when we can all rely on the benificent and all knowing genius that is Idontcare? What fools we were.


I have no idea why anybody listens to you, you clearly are out of the industry. Anybody who can still call intel 14nm a decent product or even on par with samsung 14nm in performance can't have worked that recently.

WWYBYWB?

 

[arandomguy]

My curiosity is why does anyone in these forums care whether the A9 is being produced at TSMC or Samsung?

If you are an employee of TSMC or Samsung then sure, I would be concerned from a job security standpoint. Or if you are a shareholder of either company then it becomes a profit motivation.

But outside the money angles, do people chatting on a forum really care whether or not the A9 SOC is being predominately (or solely) produced at TSMC or Samsung?

Is it a fanboy/psychological thing? Do we really have Samsung and TSMC fanboys nowadays?

Samsung has had fanboys for quite some time since they became a rather consumer facing company with their entry into smartphones. This is something the tech crowd seems to not like to refer to or ignores while being derisive against those in the non tech crowd that favor Apple.

Due to the association now of Samsung/GF and correspondingly with AMD there is also some carryover from the pro AMD or anti Intel/Nvidia groups. Similarly the Samsung/ARM association means that Samsung process technology has some relevance in the whole ARM group/Intel confrontation.

The perception is that these design wins can be used as indicators. Perhaps Apple using Samsung would mean their process is somehow better or further ahead, this could be interpreted by some as boding well for AMD's prospects with Zen. Apple using TSMC would be spun as capacity that won't be available to Nvidia or the process is ready, as always interpretation tends to fit ones preconceptions.

 

[NTMBK]

I wonder how much corporate politics plays into decisions like this? If TSMC's performance/cost/whatever was 5% behind Samsung's, would Apple still go with TSMC to deny the revenue to their competitor?

(Note I am an outsider with no insights into how Samsung and TSMC's processes compare. Just a curious observer )

 

[Phynaz]

Oh look, a page long post by idontcare implying he knows the answers to our questions but offering no information and then posing the question of why, here on one of the oldest tech discussion forums on the internet, do we "care" who manufactures what SoC or which process it may be on. Why indeed?

Well, for starters, you pretty much answered your own question in the first few lines:


Many people on these forums would fit one of those two qualifications, so there you go. But after answering your own question you then go on to call everyone in this thread "fanboys" for giving a crap. I can see how you became an "Elite" member, you're clearly way too smart and too important to even bother to answer us "fanboys". Why did we even ask?

Why did we make this thread? Why is there even an overclocking and CPU forum when we can all rely on the benificent and all knowing genius that is Idontcare? What fools we were.


I have no idea why anybody listens to you, you clearly are out of the industry. Anybody who can still call intel 14nm a decent product or even on par with samsung 14nm in performance can't have worked that recently.

Oh look, a page long post by a newb.

IDC is one of the pinacle posters on this forum. He has earned everyone's respect. You would do well to show him the respect he has earned on this forum, if you want anyone to respect your ideas.

 

[Arachnotronic]

IDC is one of the pinacle posters on this forum. He has earned everyone's respect. You would do well to show him the respect he has earned on this forum, if you want anyone to respect your ideas.

Extremely well put.