TSMC (NYSE:TSM) is a company I thought would be priced much higher than it is by now, and one of the companies I have not been successful in predicting so far, in terms of the stock price. The company does very well despite being a victim of a somewhat poor macroeconomic environment, but the stock price has not reflected that recently. With that in mind, I am once again going to predict the company will do well, although how quickly the market reacts to this is less certain short term.
The company has been hit by the same headwinds as other giants like Intel (NASDAQ:INTC). But while Intel revenue shrank in 2015, TSMC's revenue jumped 10.6%. Foundries in general grew 2% in 2015, so TSMC beat the industry. This is just the preamble for an even better 2016, with the company expecting 5% industry growth, and 5% to 10% growth for itself.
The main driver for this is TSMC's 16nm FinFET process, although improvements in the 28nm HPC+ process will also be of benefit. Although 16FF+ has been very successful since ramping in the second half of 2015, it will be bolstered by a very important node called 16FFC.
One of the major issues with FinFETs is the cost developing for it is much higher than with 28nm planar, as well as the cost per transistor being high. The 16FFC node has the same design rule as 16FF+, allowing companies to switch to it easily, as well as a simplified process, allowing for lower costs. Volume production begins in Q1, and this should be a very attractive and popular process, particularly considering it is known that Apple (NASDAQ:AAPL) was one of the drivers of this technology. On top of this, management expects this to have slightly higher margins for TSMC than the 16FF+ node. All totaled, TSMC believes it will have 70% of the 16/14nm foundry business in 2016, up from 50% in 2015. Considering this is a fast-growing segment, it is a significant data point. However, there can be no doubt some of this will be coming from its 20nm planar customers, as revenue for it will fall significantly in 2016.
Despite the fast and successful transition TSMC made to 16nm FinFETs, Q4 revenue was negatively impacted by customer inventory reductions and overall caution as the macroeconomic environment remained difficult. High-end phones are expected to remain soft into 2016, with lower-end products projected to have an upwards trajectory. Overall, TSMC sees weakness in Q1, but a 8% growth in smartphone shipments for all of 2016.
TSMC management also gave some insight into its technology development. 10nm is currently in intensive learning yield mode, with customer tapeouts this quarter. 7nm is on track and will offer "significant" improvements in density, power reduction and performance over 10nm. It is expected to be in mass production in H1 of 2018. 5nm is potentially the most interesting, as it will likely be using EUV lithography to simplify the fabrication process. This could potentially improve yields and simplify the process, lowering cost. It is scheduled roughly two years after 7nm, so around 2020.
TSMC projects capex for 2016 to be $9 to $10 billion, up 10% to 20% from 2015. The company is expecting a 10% increase in capacity, the vast majority going for the 10nm node. One point Morris Chang, the chairman, brought up was TSMC does not increase capacity unless they are relatively certain it will be used. So, not only is the company projecting greater revenue, and confidence in the 10nm process, it is clearly acting consistently with both.
It also is important to note that the 28nm process is doing well. The company expects revenue to be flat in 2016, with greater wafer demand tempered by a reduction in price. So, although the FinFETs are cannibalizing 20nm to some extent, 28nm clearly has a long life span ahead, particularly since TSMC continues to improve its characteristics.
However, there are two counterpoints to this. First, AMD has announced Polaris will be made on 14nm technology, and it's not clear if it will ever be moved to TSMC's 16nm process. Also, Global Foundries will be using a 22nm FD-SOI node, which should have superior characteristics to TSMC's 28nm planar processes. However, if the cost is not competitive, it will have little effect.
Another important move by the company was to start development of another fab in China. Although the production costs will be higher, building the plant there will put it in greater favor with the Chinese government and open doors that might otherwise be closed. Management believes the additional cost is more than offset by the greater market opportunities it will allow. It seems a wise move, considering the market in China and the growth in low-end phones.
I remain very bullish on TSMC. There's very little to dislike and so much to like. The company continues to do better than the industry and should continue to with a very aggressive technology road map that seems out of place in an industry that seems to be slowing down in fabrication technology advancement. The new 16FFC node can finally bring the cost of FinFETs down to a point where they are feasible in more scenarios. This and the continued success in legacy technologies like 28nm more than offset somewhat disappointing market conditions short them, while the strong technology road map indicates continued success for this well-run company.
Disclosure: I/we have no positions in any stocks mentioned, but may initiate a long position in TSMC over the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.