Samsung and TSMC Slicing the Apple Pie in 2015

[Daniel Nenni]

This is from UBS analysts Eric Chen, Samson Hung and Sunny Lin:

We believe TSMC’s share price in 2015 will be supported by:

1. increasing Apple business
2. the expectation of a larger cash dividend
3. rising book value valuation
4. improving profitability

First, our revenue contribution estimate from Apple is 9.4% for 2015, up from 5.6% in 2014E, even though we assume TSMC’s share at the Apple A9 and A9X application processor (AP) is 50%, down from 100% at the Apple A8 AP foundry business. Second, our cash dividend forecast for TSMC is NT$4/share. We expect TSMC to double its FCF by 2015 due to its well-managed capex and high earnings growth. Third, we estimate its book value per share will grow 20% YoY. Fourth, we forecast its EPS to grow 17% YoY in 2015. Conclusion is TSMC can raise by 30%.

Sounds reasonable to me. If however the Apple split is A9 for Samsung and A9x for TSMC as I have said it will not be a 50/50 split. More like 70/30?

 

[user_2013101]

I may or may not be correct, but my guess is that TSMC will produce at least 50% of A9s, more likely 70%. And, can’t rule out 100% of A9s to be produced at TSMC.

I had written quite a bit on this subject. Readers can refer to the following posts. (And other posts in the thread)

https://www.semiwiki.com/forum/f302...lts-dicscussion-5334-post18060.html#post18060

https://www.semiwiki.com/forum/f302...lts-dicscussion-5334-post18075.html#post18075

https://www.semiwiki.com/forum/f302...lts-dicscussion-5334-post18112.html#post18112

https://www.semiwiki.com/forum/f302...lts-dicscussion-5334-post18130.html#post18130

 

[Daniel Nenni]

I may or may not be correct, but my guess is that TSMC will produce at least 50% of A9s, more likely 70%. And, can’t rule out 100% of A9s to be produced at TSMC.

Where do you think the loss of FinFET market share in 2015 that Morris Mentioned will come from if not Apple? I do not see anyone else shipping significant FinFET volumes in 2015. Are you saying Morris is mistaken?

 

[astilo]

I think they assumed 50%, just based on an iphone and ipad split. Of course, volumes are very different, so a 75%/25% is definitely more reasonable to me.
All in all, no big surprise.
Let´s face one simple reality: TSMC is not yet ready to support volume production for the new iphone. End of the story.

 

[Daniel Nenni]

I think they assumed 50%, just based on an iphone and ipad split. Of course, volumes are very different, so a 75%/25% is definitely more reasonable to me. All in all, no big surprise. Let´s face one simple reality: TSMC is not yet ready to support volume production for the new iphone. End of the story.

I agree. My opinion is that neither Samsung nor TSMC can satisfy the FinFET demand for both iPhone and iPad launch in 2015 even if yield was fully ramped, which it is not. It is also my opinion that Apple will not split manufacturing of a single SoC design. They will split manufacturing of the A9 and A9x to Samsung and TSMC respectively. I'm certainly rooting for TSMC but this is how I see it playing out based on my conversations amongst the fabless semiconductor ecosystem over the past few months. Remember, TSMC still has a whole bunch of 20nm demand to fill with Apple and the rest of the SoC/GPU/CPU companies so they cannot just switch 20nm capacity to 16nm.

 

[user_2013101]

Where do you think the loss of FinFET market share in 2015 that Morris Mentioned will come from if not Apple? I do not see anyone else shipping significant FinFET volumes in 2015. Are you saying Morris is mistaken?

Dan,

It’s not uncommon to lowball the expectations, in order to deliver positive surprises. No problem for TSMC to announce later that the company beats the estimates or ahead of schedules. But, a big problem if falling behind.

It’s a bit shaky to base the entire thesis on a few sentences at a conference call. For one thing: the TSMC capex guidance does not seem to support the assumption of building only 30% or less of A9s. I wrote a post on this: at least $8B will be spent in 16nm buildup this year, about the same amount to ramp 20nm last year.

https://www.semiwiki.com/forum/f302...lts-dicscussion-5334-post17964.html#post17964

We don’t have insider info, nor a crystal ball. We are just guessing here, right?

P.S.
TSMC continues to buy equipments at brisk pace. See the following search results:

https://www.google.com/search?q=TSM...g_1en,cdr:1,cd_min:12/1/2014,cd_max:1/22/2015

 

[astilo]

Dear user, considering you need at least one quarter to produce a 16nm wafer from start to end, and that you also need to build some inventory to guarantee a timely on the shelves delivery, enough to satisfy the iphone lovers, the capacity must be already pretty much all installed. Your capex forecast numbers may at best fit for the 2016 production. If you are instead thinking about the possibility that the 20nm capacity could be fully converted to 16nm, just look again at the above Daniel´s comment. This is not going to happen.

 

[Arthur Hanson]

Has anyone considered the Microsoft Hololens, which some say will be in developers hands in three to six months. This would suit both TSM and Microsoft and could be the reason for the smile on Morris Chang's face. The timeline would be about right.

 

[user_2013101]

Dear user, considering you need at least one quarter to produce a 16nm wafer from start to end, and that you also need to build some inventory to guarantee a timely on the shelves delivery, enough to satisfy the iphone lovers, the capacity must be already pretty much all installed. Your capex forecast numbers may at best fit for the 2016 production. If you are instead thinking about the possibility that the 20nm capacity could be fully converted to 16nm, just look again at the above Daniel´s comment. This is not going to happen.

[h=4] TSMC 16nm status [/h]

In September last year, TSMC delivered 16nm chips to Huawei-HiSilicon.

TSMC Delivers First Fully Functional 16FinFET Networking Processor

Huawei’s P8 flagship smartphone, powered by 16nm Kirin 930 SoC, probably will start shipping by the end of May.

Jan 23, 2015
MWC 2015: Huawei Event Set to take place on March 1 in Barcelona

It’s not unreasonable to assume that 16nm production has been ongoing for some time, except in smaller quantities.

With 90% equipments shared between 20nm and 16nm and the experience from 20nm, the 16nm ramp should be smooth and fast.

Phrased approach and high initial yield of 16FF+
https://www.semiwiki.com/forum/f293/samsung-strikes-chip-deal-apple-4864-2.html#post16998
https://www.semiwiki.com/forum/f293/samsung-strikes-chip-deal-apple-4864-3.html#post17016

The June or July timeframe, as indicated by TSMC, I assume, is when the volume will reach 50K+ wafers per month.

If so, I don’t see 16nm is late for A9s, probably faster ramp than Samsung and GF.

[h=4] TSM at new all time high’s[/h]

Despite all the negative sentiments and neutral or sell ratings, TSM stock price made a new all time high on January 22. The next day, another new high. TSM’s recent correction low was on Jan 14, the day before the investor conference. It’s up 16.5% in 6 trading days.

TSM Prices

TSM certainly may correct again at some point, especially if the broader market suffers a general downturn. However, my guesses on the 16nm and A9 production do not look outrageously unreasonable. The market seems to agree with me.

Someone must be happy with owning 10K shares of TSM:

https://www.semiwiki.com/forum/f302...lts-dicscussion-5334-post17928.html#post17928

[h=4] Comic Quotes [/h]

Wife: Are you coming to bed or not?

Husband: I can’t. This is important!

Wife: What?

Husband: Someone is WRONG on the Internet!

Have a nice weekend!