backtop


Print 16 comment(s) - last by Hector2.. on Jan 21 at 10:49 AM

Company leaders expect strong results for 2014, despite a tablet/smartphone slowdown in Q1

As slowdown struck the high-end side of the smartphone and tablet market in Q3 and Q4, many investors feared we would be in for a rocky ride in 2014.  While those concerns still remain to an extent, many industry leaders have now suggested that the slowdown may be more of a short-lived blip than a long-term development.  Many companies believe that by Q2 smartphone sales will be picking up, as flashy entries like the Samsung Electronics Comp., Ltd.'s (KRX:005935) (KRX:005930) Galaxy S5 (expected in Q2) and Apple, Inc.'s (AAPL) iPhone 6 (expected in Q3) make their market debut.

I. Third Party Fab Market is Expected to Revive Soon

Reinvigorated sales would would be good news for investors in companies like Samsung or Taiwan Semiconductor Manufacturing Comp., Ltd. (TPE:2330) (TSMC). 

Samsung derives a major part of its revenue from third party chipmaking, despite the bulk of its profits coming from its smartphone and tablet sales.  Hence its investors are concerned about a double hit from a slowdown.  TSMC is solely devoted to third-party fabrication, so in a way it is even more sensitive to general industry-wide trends.

The latest development comes courtesy of the latter fab firm, who announced its Q4 2013 earnings early Today at the close of the Taiwanese market.

TSMC
[Image Source: Cult of Mac]

In Q4 TSMC saw a net profit of NT$44.8B ($1.49B USD) (NT$1.73 per share), better than the figures of NT$42B ($1.40B USD) and NT$41.8B ($1.39B USD) that analysts surveyed by Thomson Reuters I/B/E/S and Bloomberg, respectively, were expecting.  That's still down 13.7 percent from TSMC's strong Q3 2013 performance, but it wasn't as bad as was expected.  On a year to year basis TSMC still mustered a modest 7.7 percent profit growth.

Revenue stacked in at NT$145.8B ($4.85B USD), a hair higher than the on average NT$145.4B ($4.84B USD) that the 23 analysts surveyed by Bloomberg expected.  Again, this was down 10.3 percent on a quarter-to-quarter (QoQ) basis, but up 10.9 percent on a year to year (YoY) basis.

II. Strong 2014 Predicted for TSMC

TSMC's forecast is where things get interesting.  In Q1 the story is mixed.  Analysts were expecting NT$141B ($4.69B USD), but TSMC forecast revenue of NT$136-138B ($4.52-4.59B USD).  At worst, that would represent a 6.7 percent drop from Q4's revenue.  And yet there's also a positive surprise in the guidance as improvements in high end (high margin) chip sales were expected to be strong enough to drive TSMC to a NT$44.5-46.5B ($1.48-1.55B USD) profit (likely a small growth from Q4), versus the analyst expectation of NT$43.7B ($1.45B USD)

For the year TSMC's expectations were even rosier.  It say its expects double digit growth in both revenue and profit for the fiscal year 2014 (FY2014).  That indicates an expectation of a strong Q2-Q4 2014 sales run, as the profit growth expectation in Q1 is about 3.7 percent at best.

TSMC chips
TSMC is still seeing strong orders for 28 nm processors.  [Image Source: Reuters]

TSMC's biggest rival -- South Korea's Samsung is reporting its Q4 results on Jan. 24 and the results are not expected to be good, judging by the company's preliminary sales forecast, released before CES.  Where as TSMC's issues are more isolated to sales alone, Samsung is also being hit by a strong won, the currency of its home nation.  This strengthening has created difficulties for Samsung as it is making it harder to bring its money back home to spend on new tehcnology.

Samsung and TSMC both play the roles of "mobile arms dealers" of sort, or, more aptly mobile ARM dealers.  Both rely heavily on third party orders of chips using ARM Holdings Plc's (LON:ARM) ARM architectures and mobile instruction sets, technologies that dominate today's mobile processor market.

TSMC Fab3
TSMC controls 50 percent of the mobile chip market. [Image Source: TSMC]

At last estimate TSMC was the biggest processor producer in the mobile space, responsible for an estimated 50 percent of production by volume.

III. Samsung vs. Intel vs. TSMC -- Roadmaps

Intel Corp. (INTC), the auburn ruler of the personal computer processor space, stirred much fear and speculation among investors in TSMC and other fabs when it announced in Oct. 2013 that it would be manufacturing chips for Altera Corp. (ALTR).  The particularly heated point of controversy was the fact that these FPGAs included 64-bit ARM cores.

Intel currently enjoys a modest process lead over TSMC.  TSMC in 2013 was in its second year of building 28 nanometer transistors, while Intel was in its second year of building 22 nanometer transistors.  That said, Intel has saved its best chip lines for some time for traditional PC processors, building mobile chips on older facilities on larger nodes.

TSMC is starting its first 20 nm mass production this quarter, which will put it ahead of Intel -- if only briefly.  Intel may still have a slight edge in power thanks to its power-saving 3D transistor design commonly referred to as FinFETs (Fin field effect transistors).

Intel spent $11B USD last year on semiconductor research and fab development, while TSMC spent slightly less -- $9.7B USD.  That was still a record expenditure from TSMC, who expects to bump investment again this year.  Samsung spent $22B USD in 2013, making it the biggest spender, capital investment-wise.

 
TSMC roadmap
[Image Source: ExtremeTech]

IDF 2013
[Image Source: Jason Mick/DailyTech LLC]

samsung die shrinks

[Image Source: Samsung]
 
The next node for TSMC is the 16 nm node (which will finally bring in FinFET designs), expected to be taped out late this year (Q4 2014) or very early next year (Q1 2015).  The next node for Intel  -- 14 nm -- was already tape out last year.  Next year it plans to tape out 10 nm chips, in preparation for a 2016 product launch.  Samsung is also expected to tape out 20 nm this year, reach volume production next year.

So arguably TSMC is about a year behind Intel in process, at present, and Samsung is a year behind TSMC.  Globalfoundries, a fourth major player, is thought to be a little behind Samsung.

IV. No Minor Threat

Intel is a serious threat to TSMC.  The rumor that it's expected to devote its best lines to third party chip production are likely baseless, but even if it opts to primarily push its own mobile product, that's dangerous to TSMC due to its process leadership.

Samsung is dangerous for different reasons.  As the world's top smartphone maker, it can cut costs by controlling the entire supply chain, something TSMC can't do.  It also is spending much more on chipmaking research and fab development that Intels or TSMC.

Fab worker
Samsung may trail TSMC in market share and process, but it's a deep-pocketed and dangerous rival. [Image Source: IntoMobile]

Still, TSMC is confident in its future path.  Chairman Morris Chang, who recently stepped down from the CEO spot at the fab firm, expects mobile and tablet chip revenue will grow roughly 35 percent in 2014.  New co-CEO , reigning alongside CC Wei, remarked during the earnings call, "[TSMC's 28 nm is] scaling is much better than what Intel said, though still a little bit behind.  [TSMC is] far superior [than Intel or Samsung] in mobile execution."

TSMC's Mr. Liu insists that his firm is more less tied with Intel, calling Intel's claims of process supremacy "erroneous and based on outdated data".  He may technically be right, but it won't be long before Intel has product on the market that once more leapfrogs TSMC in node size.

But it should not be discounted that Intel has thus far been unable to leverage its node advantage into sales in the mobile market, and that TSMC has dominated that market despite some struggles making smaller transistor nodes.

Intel has had some recent struggles of its own, delaying the release of its first 14 nm processor Broadwell from Q3 2014 to Q4 2014.  It also announced it would hold off on 14 nm upgrades to its Arizona fab, out of concern dwindling PC sales would mean insufficient orders for 14 nm chips.  The Arizona facility was one of three facilities that had previously been announced as a source of 14 nm production.

Intel CEO
Intel CEO Brian Krzanich [SOURCE: bgr.in]

Intel CEO Brian Krzanich said last week that Intel would be "pragmatic" in leveraging its process expertise to win third party clients, but in a nod to TSMC's prowess in the budget space added, "[We are] not going to get into a price war with TSMC."

Sources: TSMC, Reuters, Bloomberg



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

AMD Fab
By pmaciel on 1/17/2014 5:36:23 AM , Rating: 2
Why could not AMD change from GloFo to TSMC to build the Kaveri APUS, taking advantage of more advanced lithography?




RE: AMD Fab
By Ranari on 1/17/2014 7:44:37 AM , Rating: 3
Globalfoundries used to be AMD's fab unit, and part of the spinoff/separation details made it so AMD is still contractually tied them to order so many units per year using their lithography, and is required to produce all of their chips with GF by a certain dates until that contract date ends.

I agree though, I think the deal hamstrings AMD.


RE: AMD Fab
By phatboye on 1/17/2014 3:29:32 PM , Rating: 4
It does, the entire purpose of being Fabless is that you should be able to shop around to which ever fab gives you the best contractual deals.


RE: AMD Fab
By blaktron on 1/17/2014 6:02:30 PM , Rating: 2
So the AMD / GF split was originally planned to avoid AMD being less than 50% US owned, which would put the x86 license at risk. SO they spun off GF, and signed deals for X number units to make the new company attractive to investors. Had nothing to do with shopping fab space, although that would be an advantage if you forget about the rest of the deal.


RE: AMD Fab
By coburn_c on 1/17/2014 8:49:50 PM , Rating: 3
quote:
So the AMD / GF split was originally planned to avoid AMD being less than 50% US owned, which would put the x86 license at risk.


I'm not sure if this is some sort of revisionist history or just misinformation, but I've never heard this before. The foundry spinoff was a desperate bid for liquid capital after the heavy leveraging to acquire ATI put AMD in serious risk of bankruptcy in the wake of the financial crisis. It was, in fact, the spinoff, if anything, that put the x86 licenses in jeopardy:

http://betanews.com/2009/03/16/intel-fires-a-warni...


RE: AMD Fab
By ilt24 on 1/18/2014 3:12:03 PM , Rating: 2
I think your both right. AMD could not longer afford it fabs but because of their cross license agreement with Intel could not outright sell the fab, so Globalfoundries was originally created with AMD having 50% ownership.

As the guys from Abu Dhabi added capital and AMD didn't, the ownership fell below 50% and Intel complained. Soon after this there was a settlement of AMD's lawsuit with Intel, which included removing the restriction of AMD having to make it's x86 wafers.


RE: AMD Fab
By coburn_c on 1/18/2014 6:05:14 PM , Rating: 2
Did you read the article? AMD never owned 50% of Global Foundries and Abu Dhabi never owned more than 8% of AMD. There was nothing right about his claim.


RE: AMD Fab
By ilt24 on 1/18/2014 11:04:13 PM , Rating: 2
From the original 2008 press release:
quote:
At the same time, the Mubadala Development Company will increase its current investment in AMD to 19.3 percent on a fully diluted basis.

quote:
Upon closing, AMD will:

Have equal voting rights with ATIC in The Foundry Company;
Own 44.4 percent of The Foundry Company on a fully converted to common basis;

It wasn't a 50% ownership, but a 50% voting right that AMD had, which they did to try and get around the 50% clause in their cross license agreement with Intel...so I still thing you where both right.

see: http://www.globalfoundries.com/newsroom/2008/20081...


RE: AMD Fab
By Ringold on 1/20/2014 4:52:47 AM , Rating: 3
From a practical perspective, I don't think any one that looks at it thinks AMD could lose the x86 license, no matter what business deals they strike. Why? Neither the US nor EU would allow for Intel be the only real competitor in the vital x86 realm. They're already touchy as it is.

If Intel didn't agree to new deal, the entire Western worlds regulators would be coming down on their heads. Intel's smarter then that.

Similar reason why Intel hasn't been more aggressive then it has, in advancing its technology or lowering prices. It'd destroy AMD (still trying to catch up to Sandy Bridge and it's own damn Thuban X6's in some scenarios), which would be a legal nightmare.


Does not make sense
By asliarun on 1/18/2014 5:01:10 PM , Rating: 2
quote:
Intel currently enjoys a modest process lead over TSMC. TSMC in 2013 was in its second year of building 28 nanometer transistors, while Intel was in its second year of building 22 nanometer transistors.


quote:
"[TSMC's 28 nm is] scaling is much better than what Intel said, though still a little bit behind. [TSMC is] far superior [than Intel or Samsung] in mobile execution."

TSMC's Mr. Liu insists that his firm is more less tied with Intel, calling Intel's claims of process supremacy "erroneous and based on outdated data". He may technically be right, but it won't be long before Intel has product on the market that once more leapfrogs TSMC in node size.


quote:
The next node for TSMC is the 16 nm node (which will finally bring in FinFET designs), expected to be taped out late this year (Q4 2014) or very early next year (Q1 2015). The next node for Intel -- 14 nm -- was already tape out last year. Next year it plans to tape out 10 nm chips, in preparation for a 2016 product launch. Samsung is also expected to tape out 20 nm this year, reach volume production next year.

So arguably TSMC is about a year behind Intel in process, at present, and Samsung is a year behind TSMC. Globalfoundries, a fourth major player, is thought to be a little behind Samsung.


Some of these statements don't make sense.

In various parts of the article, you say that Intel has a modest lead - or agree with TSMC's statement that there is virtually no lead.

Yet, according to your own article (what I have quoted in the end) Intel seems to have taped out 14nm 12-18 months before TSMC has even taped out 18nm. If you adjust for the fact that TSMC has more or less settled on half nodes to compete with Intel, Intel seems to have a good 24 month or 2 year advantage over TSMC.

Does that really seem like a "modest lead"?

If anything, Intel has actually extended its process lead over its competitors in the last 10 years or so. I remember a decade ago when Intel only enjoyed a 9-12 month lead. And a single misstep can derail such a lead.

Plus, Intel has FinFet shipping today in its 22nm node when TSMC will have FinFet only in 2016. That in itself is a good 3 year lead.

Don't get me wrong - Intel has made some massive screw ups in terms of ignoring emerging trends and neglecting power efficiency over pure performance.

However, as a pure foundry, I am not sure how much marketing hogwash someone can throw when they claim that TSMC is in the same technical level as Intel.

The only thing you can really give TSMC is that it does SOCs better than Intel. That is only because TSMC makes sure that inefficient SOC design is paid by its client and not by TSMC itself - so they don't care. Intel has less design experience in SOCs but that too is rapidly changing. And Intel cannot tolerate inefficient SOC design as much as TSMC as it is more vertically integrated so it hits Intel's margins at the end of the day.




RE: Does not make sense
By melgross on 1/18/2014 7:41:17 PM , Rating: 4
Except that Intel has had delays the past three cycles. They're about a year later than they would have originally been with 14nm, because of the generational lapses over the past few years.

Of course, others have had even more problems. I remember AMD, a couple of years before the split, claiming that their 12 month lag behind Intel would shrink to 6 months, but instead grew to 18 months. That was one of the reasons they sold out. That lag was eating into their high end sales, as every product came out late (among other things).

So now Intel is almost 6 months late with 14nm. But they claim that 10nm is on track for 2016. I really wonder. A lot of chip experts have been saying that 10nm could be the last easily attainable chip reduction. Some are saying that it could be THE last. Yet, Intel is showing a roadmap with 7nm, and even 5!

I have a feeling that 10nm will again be late. By how much is anyone's guess, but I wouldn't be surprised if it was more than 6 months, and possibly even a year.

As far as 7nm goes, well, good luck with that!


Reallllly...
By coburn_c on 1/17/2014 3:48:10 PM , Rating: 1
quote:
"[We are] not going to get into a price war with TSMC."


Sounds like grounds for a price fixing investigation, time to bleed some money EU.




RE: Reallllly...
By ritualm on 1/17/2014 5:30:57 PM , Rating: 1
I don't think Intel is in a sound enough position to engage in protracted price wars against a major low-cost foundry for the sake of new customers.

Then again, many of TSMC's claims have to be taken with a heaping spoon of salt. 16nm FinFET isn't coming online without another 6+ months of delays, why? Because this is TSMC we're talking about.


RE: Reallllly...
By melgross on 1/18/2014 7:20:52 PM , Rating: 2
Not at all. It just means that they won't try to price their chips so low as to kill their profits. There's nothing wrong with that. It's done all the time.

The law states that it's ok for companies to arrive at the same pricing, as long as they don't get together before to set those prices. So if one company has a price, and another matches it, it's perfectly legal.


Intel
By deltaend on 1/18/2014 8:15:46 AM , Rating: 2
Intel is far better off than people give it credit for. In the mid to high end pc market, Intel is the clear leader. In the datacenter, there is pretty much only Intel. For every single mobile device that you add, you create a demand for cloud services which in turn forces datacenters to add more Intel servers. For every mobile device that offloads data to a pc in order to gain disk space or graphics processing, you are likely to find an Intel chip powering it. In essence, the mobile market actually creates more demand for pc's, it's just that people are more likely to throw away their mobile device and buy a new one every other year than they are their pc.




TSMC Superior ?
By Hector2 on 1/21/2014 10:49:59 AM , Rating: 2
With their decades of experience working with customers, I think it goes without saying that TSMC would have a superior infrastructure to manage all of that. This is their strength.

However, this means little to Intel presently and gives TSMC no advantage now. This is because Intel is working with only a few external companies currently and Altera is a sophisticated & experienced business partner not needing a lot of hand-holding.

Only as they ramp up their foundry volume and start adding many more foundry customers, will it be important for Intel to have a stable & robust foundry infrastructure




"We shipped it on Saturday. Then on Sunday, we rested." -- Steve Jobs on the iPad launch














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki