You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!

TSMC reports In Line Revenues & Penny EPS beat Slow Inventory recovery

Robert Maire

Moderator
Last week TSMC (TSM) reported revenues in line at $6.14B which in line with expectations and down about 8% Y/Y. EPS was $0.38 a penny above the street but also down Y/Y about 18%. The slow recovery in chip inventories is continuing.

View attachment 17068

Not "all shook up"...

It seems the February earthquake had little negative impact on TSMC and most of the recovery appears to be driven on the lower end of the smartphone market. 28nm remains a strong point. Revenue was more or less flat from Q4 2015.

Mix of product across technology nodes appears to be solid with continued progress at the leading edge. Guidance is for Q2 to be up nicely as the industry continues to recover and inventories continue to refill. Revenues for Q2 are expected to be up about 6-7% Q/Q, with gross margin to be between 49-51%.

The company said they expect 1% growth in semis with 5% growth in foundry and 5-10% growth for TSMC.

CAPEX off to a weak start.....
Capital spending in Q1 was relatively weak at $1.15B versus Q4 2015 at $2.6B. This is probably a bit less than expected and will likely weigh negatively on the equipment stocks. We had predicted in our report several weeks ago that Q1 results would not likely support the run up in stock prices that we have seen. TSMC's spending in Q1 appears to be light and will likely show up in the results of a weaker Q1 for equipment companies. We also suggested, several weeks ago that we see capital spending being pushed further into the end of the year ( Q4) rather than just the second half (H2) of 2016. The weak start seems to support this view.

Capex likely to be at low end of $9B- $10B...
With just $1.15B spending in Q1, that would imply a strong ramp of $1B, $2B, $3B & $4B if we took a simple ramp. We don't see TSM taking in $4B worth of tools in a quarter so we think its more likely for a slow Q2 with an increase in Q3 but a strong Q4. Given that demand remains soft we would expect that TSMC will not build out fabs "to the max" which would further support the view of spending at the low end or roughly $9B, essentially in line with 2015.

95% tool reuse at 7nm over 10nm--A coming "air pocket" for spending???
One very interesting comment that we heard on the conference call was that the company expects a 95% tool reuse rate between 10nm and 7nm which is similar to what we saw between 20nm and 16nm. 7nm is a "reworked " 10nm process flow much as 16nm was a "reworked" 20nm process flow.

We would be very careful as this suggests that equipment companies could be facing an extended weak period after the expected strong spending associated with 10nm as 7nm won't require a lot of new tools. This would imply that we likely will have a year of good spending starting in earnest in Q4 of this year slowing into the second half of 2017 then substantially weakening as TSMC focuses on 7nm.

EUV still needs a lot of work...7nm train already left the station without EUV...
While saying that EUV has made progress it appears that TSMC is still not going to use EUV at 7nm as its currently not ready for prime time and 7nm process is already being finalized. They did mention that progress was being made on source power but that reliability remains an issue. TSMC also said that "ecosystem" issues have also had progress but that pellicles sill need to be developed. TSMC said they had reasonable yields by trying one layer of EUV as a test on 7nm. This continues to support the view expressed by TSMC that they expect to use EUV at 5nm - "depending upon progress".

This is about as firm a commitment as Intel has made by saying that EUV will be used when its ready and its not ready yet.

7nm is a "fast follower" but perhaps not as full a "shrink"...
TSMC said on the call that 7nm will have a 30-40% power reduction that while good implies more of a tweak of 10nm rather than a full blown, new process. TSMC is clearly focused on shrinking while less focused on re-design at every new node. This certainly streamlines and speeds progress and gets most but not all of the benefits of a shrink.

So far, this game of "pedal to the metal" appears to be working for TSMC as they seem to have the momentum on their side in node shrinks.

Summary...
We continue to view TSMC as a great company and as done a great job of managing its financial model and throttling its spending while at the same time hurtling ahead in technology. Obviously the market hasn't been great, nor do we expect a rapid recovery but as the recovery comes along, TSMC will get a bigger and better share at better profitability.

Equipment companies obviously will not be happy about both the light capex spend in Q1 and what will likely be a flat 2016 over 2015 in capex. The 95% reuse for 7nm could lead to another year or two "in the desert" between "major" technology nodes. As we have seen in the past, not all technology nodes are created equally nor have the same level of capex. As we predicted several weeks ago Q1 may see a pull back of semi equipment stocks to get expectations back in line with reality and stocks will likely dip as that reality kicks in.

 
Top