Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/tsmc-has-in-line-quarter-as-expected-outlook-conservative.6679/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021770
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

TSMC has In line quarter-as expected- outlook conservative

Robert Maire

Moderator
Revenues of NT$212.5B EPS of NT$75.3B- Guides NT$201-204B- Inventories up
Given the current inventory that exists in the market coupled with demand patterns and previous guidance by TSMC, the company reported more or less in line with expectation.The environment suggests we will not see a sharp rise any time soon in revenues as inventories persist and demand is lackluster. We would imaging that inventories will be coming down as they are worked off through the end of the year.

3Q15_E_5066_CpDj.jpg


Slashed Capex should be no surprise
TSMC cut its CAPEX plans for 2015 from previous $10.5-$11B to $8B. Of this amount the company said it spent $5.5B YTD which would technically leave $2.5B to spend but we would assume that won't happen and the final tally for 2015 will probably be a lot closer to $7B.

What could CAPEX be for 2016???
In our view its likely that 2016 will be flattish in terms of capex versus 2015. The issue is that its likely too early in 2016 to see significant spend on 10nm but maybe some early spend later in the year. Given current industry conditions and momentum the first half of the year will likely see relatively low capex in 2016 and be more back end loaded as we see the early phases of 10nm.

Likely rough sledding for equipment companies for next 3 quarters...
As we have joked many times in our "cockroach" theory of bad quarters, heres never just one in the semiconductor equipment industry. Its going to be harder for investors to look beyond the downturn if its at least 2-3 quarters or longer, which we still don't know.

16nm capacity is just fine, thank you...
Given that Apple has split the A9 business between Samsung and TSMC for now, we would imagine there is no real need on TSMC's part to expand its 16nm as it will likely have more than enough capacity for other customers as the initial surge for Apple wears off.

Though there could be some opportunity for TSMC given its apparent power advantage over the Samsung A9 part we would expect Samsung to further refine its process to narrow any difference that may exist. Either way TSMC likely has enough 16nm capacity in place such that we won't see significant additional spend on it.

In the lull between 16nm and 10nm...
We are kind of in that place in the cycle where we have built out 16nm/14nm but are not ramping up 10nm quite yet. Given the clear delay on 10nm by Intel and similar issues we would expect others to have we would warn that this Lull period could be longer than previous lulls in technology nodes leading to a longer downturn between spending cycles.

Where does this leave GloFo???
Given that it would appear that TSMC and Samsung have captured all of Apples A9 business and there is likely enough capacity out there at 16nm/14nm our guess is that isn't much in the way of table scraps for GloFo to go after. It could be that given the rush to get 16nm/14nm ready for Apple and GloFo's built in delay waiting for the process to be handed down from Samsung that they were just too late to the party.
In addition given that both TSMC and Samsung likely now have some extra 16nm/14nm following the initial Iphone ramp, it may be even harder for GloFo to break in to this market as they are competing against Samsung and TSMC with available wafer starts.

The bigger question in our mind is where does GloFo go from here? It sounds like Samsung will not license its 10nm process flow to GloFo which means they would have to strike out on their own with process developed in house and from the IBM acquisition. We would question wether this is indeed ready for prime time and if customers & EDA shops can get the arms around it in time to be a viable alternative at 10nm.
If GloFo can't break into the leading edge then it will be relegated to being a less profitable commodity supplier of semiconductor capacity.
Given current conditions and cost cutting, we also don't expect a lot of capex out of GloFo in 2016.

Samsung already announced a 20% capex cut...
As they had previously announced in their capex cut, we think they too have enough capacity in the near term. We don't think they need a huge amount of additional capacity for internal consumption as their own products haven't been flying ff the shelves and macro demand remains soft.

The bottom line is the only place where we expect Samsung to invest is 3D NAND and thats not a surprise nor is it upside to existing expectations. We would be careful to monitor NAND pricing and inventory as we could see a similar situation as we saw in DRAM if demand doesn't suck up the new capacity generated by 3D.

Capitulation Continues- Bolting Barn Barricade Behind Bolted Bovines
We are at that familiar point in the cycle where the negative signs have been on the wall for all to see for a long time, the stocks have dropped, and now when its too late to prevent losses, analysts are trying to cut theirs by finally capitulating and getting negative. This also usually means we are at or close to a bottom as thats when most analysts give in.

Memory pricing has been dropping through the year, foundry weakness is so old news we heard about it a year ago. The macro economy continues to be weak. Intel has been cutting back spending while cutting back their schedule to a two year cadence from three. Samsung cut their capex a month ago and now TSMC has slashed their capex......there is not much more negative news that we can think of.....maybe 3D NAND slowing

Over the last few weeks, in front of earnings, analysts have finally started to capitulate and get more negative, they are all finally cutting their capex forecasts and earnings estimates. Numbers are probably coming down, but likely not enough.

Fundamentally the two key questions are 1) how deep? and 2) how long?

The depth of the downturn seems a bit fuzzy but it seems from what we see and think that analysts may be forced to make a second round of cuts as the news flow may get worse before getting better. We would have thought it would be prudent to slash numbers more aggressively rather than coming back around to have to cut them again.There does not seem to be a large upside surprises and so far the news has been underwhelming

In terms of the length of the downturn, and according to our cockroach theory, it initially feels like a 3-4 quarter downturn which is a bit short on a historical basis but more in line with the less cyclical pattern that the industry has evolved into over the longer term.In that time period we would count the 3rd and 4th quarter of 2015 and likely the first and second quarter of 2016 (at a minimum)

Our checks within the industry indicate that most are already bracing for weakness in the first half and we will likely need at least some help from the macro economy to see a second half recovery (maybe in line with elections...)

The stocks...
Given what we just said , the stocks are likely near some sort of near term bottom where we could bounce along at these new valuation levels for a while. We may see some further resets of valuations as companies report numbers that are less than thrilling and shave a few dollars off their stock prices.

In all this there appears to be some resilience as the stocks seem to be more willing to absorb the bad news which is again another sign of a bottoming. ASML reported less than stellar results and a weak guide and after an initial down turn saw the stock regain. Maybe investors don't have a lot of alternatives or maybe new money finds the lower valuations more attractive..

For patient investors, bottom fishing may not be a bad thing. Given the that downward beta seems to be lessening those investors who can wait around for a while till we get a sense of a recovery could do well. We would also expect some short covering in here as investors were smartly short the stocks can take their money off the table and not try to get too greedy or foolish.

Robert Maire
Semiconductor Advisors LLC
 
Here are the sound bites from SA:

  • TSMC (TSM -1.7%) is off moderately after slightly beat Q3 EPS estimates. Revenue of NT$212.5B (+2% Y/Y) was in-line with prior guidance of NT$211B-NT$213B, above original guidance of NT$207B-NT$210B (a strong dollar helped), and already telegraphed by the company's monthly sales reports.
  • Q4 revenue guidance is at NT$201B-NT$203B ($6.23B-$6.29B), down from NT$222.5B ($6.9B) a year ago. A 2-analyst Thomson First Call consensus is at $6.29B. Apple's decision to partly rely on Samsung for A9 CPU production (unlike with the A8) is weighing on near-terrm sales.
  • The 2015 capex budget has been cut for the second time this year, to $8B from a prior $10.5B-$11B. A cut was generally expected, given the forecasts provided by various chip equipment makers.
  • 16nm/20nm chips accounted for 21% of wafer revenue, and 28nm chips 27%. That compares with 20% for 20nm and 27% for 28nm in Q2. 16nm (still in ramping) is expected to contribute "more significantly" to Q4 sales.
  • Gross margin was 48.2%, down 30 bps Q/Q and 230 bps Y/Y, but in-line with guidance of 47%-49%. Q4 GM guidance is at 47.5%-49.5%.

Taiwan Semiconductor Manufacturing Company's (TSM) CEO Mark Liu on Q3 2015 Results - Earnings Call Transcript | Seeking Alpha
 
Last edited:
As always the most interesting bits were in the Q&A, much of which was taken up with barely disguised questions, "What's Apple doing?".The point that struck me was CC Wei's comment the 10nm would be a short node, retracted subsequently and substituted with, "We don't know", it being fairly obvious that the Apple-TSM plan is to go straight through 10nm down to 7nm as fast as they jointly can.

I would strongly recommend anyone interested to listen to the full video. It is as much what they don't say that is informative as what they do.
 
As always the most interesting bits were in the Q&A, much of which was taken up with barely disguised questions, "What's Apple doing?".The point that struck me was CC Wei's comment the 10nm would be a short node, retracted subsequently and substituted with, "We don't know", it being fairly obvious that the Apple-TSM plan is to go straight through 10nm down to 7nm as fast as they jointly can.

I would strongly recommend anyone interested to listen to the full video. It is as much what they don't say that is informative as what they do.

I agree, it is a good read. There are some real surprises in there for us insiders. I will comment more later. Happy Friday!
 
As always the most interesting bits were in the Q&A, much of which was taken up with barely disguised questions, "What's Apple doing?".The point that struck me was CC Wei's comment the 10nm would be a short node, retracted subsequently and substituted with, "We don't know", it being fairly obvious that the Apple-TSM plan is to go straight through 10nm down to 7nm as fast as they jointly can.

I would strongly recommend anyone interested to listen to the full video. It is as much what they don't say that is informative as what they do.

Did you catch this one Brian:

C.C. Wei - President and Co-CEOTo repeat your question, you said that it will be in second half of next year and/or 2017. Our initial schedule actually said in 2017, but we might pull in because right now we have two versions of 16 FFC, one is [swing] and the other one is a [non-swing].

Why were there no follow-up questions on this?
 
FC=cost reduced 16nm, correct?
Let's say TSMC's strategy is to stay at 16nm and work on reducing the cost until they can move their massive 28nm business to 16nm. I like this strategy.
Intel's strategy is to delay and wait for EUV to mature, then bring up 10nm in a big way, leveraging EUV to reduce the cost. That seems risky.
Samsung's strategy appears to be to bring 10nm and 7nm as quickly as possible, using the "litho lite" strategy they used at 14nm. I like this strategy too.
 
FC=cost reduced 16nm, correct?
Let's say TSMC's strategy is to stay at 16nm and work on reducing the cost until they can move their massive 28nm business to 16nm. I like this strategy.
Intel's strategy is to delay and wait for EUV to mature, then bring up 10nm in a big way, leveraging EUV to reduce the cost. That seems risky.
Samsung's strategy appears to be to bring 10nm and 7nm as quickly as possible, using the "litho lite" strategy they used at 14nm. I like this strategy too.

Yes, 16FFC requires less masks and uses 7 track libraries versus 9 track, which is good for low power designs that do not require high performance (IoT). 16FFC is definitely a 28nm replacement and it is going to be a LONG time before the second source fabs can replicate this process (UMC, SMIC, etc...).

The other thing that was mentioned is 10nm and 7nm fabs are nearly identical, like 20nm and 16nm. So the time gap between 10nm and 7nm will be shorter than I previously thought.

My guess is that quite a few companies will skip 10nm and jump right to 7nm. Xilinx has already said this, others are sure to follow.
 
7nm is the promised land. The land of milk and honey and...a lower cost/transistor than 28nm.
 
Slashed Capex should be no surprise
TSMC cut its CAPEX plans for 2015 from previous $10.5-$11B to $8B. Of this amount the company said it spent $5.5B YTD which would technically leave $2.5B to spend but we would assume that won't happen and the final tally for 2015 will probably be a lot closer to $7B.

At such late moment, only two and half months to the end of year, I believe TSMC $8 billion CAPEX forecast they mentioned in their October 15th conference call will probably be very close to their final 2015 real CAPEX figure. I guess.
 
Last edited:
From the earning conference transcript:

Given the fourth-quarter guidance, TSMC should be able to deliver a double-digit growth, at about 10% to 11% year on year in 2015, thanks partiallyto the stronger US dollar.

And

Our 2015 forecast of the growth of smartphone unit shipment is 10%, PC for the industry is minus 6%, tablet minus 14%, digital consumer electronicsminus 6%. The semiconductor industry growth is about zero percent. And our fabless industry growth is minus 5%.


So I'd like to ask if TSMC can grow 10% ~ 11% while "The semiconductor industry growth is about zero percent. And our fabless industry growth is minus 5%" then who is (are) losing shares?
 
At such late moment, only two and half months to the end of year, I believe TSMC $8 billion CAPEX forecast they mentioned in their October 15th conference call will probably be very close to their final 2015 real CAPEX figure. I guess.
Lora Ho implied in the July transcript that they were looking at the capex budget and had decided to leave it alone for the time being. She said they had cut the budget this transcript for four very specific reasons. I pay far more attention to what she says than any analyst; she very obviously knows what she is doing. With two months to go it seems 10:1 on that the expenditure has already been approved.
 
The CapEx cut is only a small part due economic downturn:
about 33% or one-third of the reduction of CapEx is due to operating efficiency gains that led us to spend less money but still has the same output. Another one about 30% of the reduction is due to changes in investment projects including the conversion between 20 nanometer and 16 nanometer. Another one, about 20% of the reduction is due to changes in capacity schedules. Lastly, the remainder about 17% of the total reduction is due to the strengthened of U.S. dollars against euro, Japanese yen and NT dollars. We expect our 2016 capital budget will be higher than this year.
 
Back
Top