I own shares of Intel (INTC -0.38%), but I do not own shares of Taiwan Semiconductor (TSM 2.84%). However, as an investor, I am constantly testing, retesting, and trying to break hypotheses for the good of my long-term returns. With shares of Intel up significantly from my cost basis, and with shares of semiconductor foundry giant having pulled back about 11%, I wondered whether it may make sense to swap out some Intel for Taiwan Semiconductor.
I decided against it; here's why.
Taiwan Semiconductor's business looks great ...
As the mobile computing explosion occurred, the companies that really grabbed the proverbial bull by the horns were the fabless companies -- most of which utilized Taiwan Semiconductor's factories -- as well as Samsung Electronics. Intel, the company that had traditionally owned consumer-oriented computing, was nowhere to be found.
This opened up a whole new frontier for Taiwan Semiconductor, which in 2012 found itself overwhelmed by the demand profile for its latest leading-edge chips at the 28-nanometer node. This overwhelming dominance at 28 nanometers has led Taiwan Semiconductor to post not only record sales but also extremely healthy gross margins, as the result of having essentially 100% utilization.
... but is the company at peak earnings?
Though I have very little doubt that Taiwan Semiconductor's 28-nanometer technology is mature and probably cheaper to build than that from the fairly new competitors to this space, Taiwan Semiconductor is likely to see some margin pressure at the low end -- particularly as the company's lead customer, Qualcomm (QCOM 0.73%), has inked deals with alternative players for this node going forward.
What's really concerning, though, is that Taiwan Semiconductor -- which had commanding leads at 28 and 20 nanometers -- now appears to be falling behind Samsung at the very bleeding edge. In fact, TSMC admitted that its share of the 14/16-nanometer node would be behind Samsung's during 2015 -- though it expects a rebound in 2016 -- because both Qualcomm and Apple are reportedly moving some orders over to Samsung, according to Digitimes.
In short, TSMC's position as the dominant foundry supplier at the leading edge is now under very serious threat, which may bode poorly for revenue growth and/or margins in late 2015 and beyond.
Two outcomes: highly profitable duopoly or margin disaster
There are two logical (and, unfortunately, diametrically opposed) outcomes possible here. The first is that Samsung and TSMC could both enjoy high margins in a tightly controlled pricing environment. However, it's just as likely that in a scramble to get orders from Qualcomm and Apple, the two leading-edge customers that matter, Samsung and TSMC will both get into a race to the bottom and foundry margins get crushed.
So, why do I like Intel better?
It's true that Intel has largely missed the mobile market so far. However, in sharp contrast to TSMC, which is at the top of its gross and operating margin game, Intel's net income is significantly depressed by its large investment in mobile.
In fact, as Intel bears will be all too keen to point out, Intel is annualizing a nearly $4 billion loss as it invests in the mobile market. However, despite this loss, Intel is on track to do over $10 billion in net income this year. If Intel is able to bring this mobile business to breakeven, then Intel will enjoy a fairly dramatic profit and margin uplift.
More to the point, there's still room for things to get much better since mobile is so horrific -- and that means an increased share price if Intel can deliver, which remains to be seen.
Additionally, it doesn't look as though any of the ARM Holdings (ARMH)-based server offerings will serve as credible threats to Intel's Xeon stronghold, nor does it appear that Intel will do anything but continue to gain share in the PC market as it advances both its Atom and Core architectures. Should evidence to the contrary appear, it would then be time to re-evaluate the thesis.
Foolish bottom line
At this point, while I have tremendous respect for what Taiwan Semiconductor has accomplished, it's really looking as though there's a chance that margins and profitability are close to a peak while Intel's business could see significant leverage once mobile revenues become material. As such, I prefer to remain long Intel rather than swap out to Taiwan Semiconductor at this time.
