One of the unsung heroes of the semiconductor industry is Taiwan Semiconductor (TSM 6.22%). While just about every chip house had its own in-house manufacturing in the early days of the semiconductor industry, the economics eventually weren't sustainable. Taiwan Semiconductor, followed by a number of other players, pioneered the "fabless" model in which the design houses were separate from the actual manufacturing.
This allowed smaller chip players without the revenue bases to sustain their own chip production factories to still design and sell competitive semiconductor products. Taiwan Semiconductor is the unequivocal leading foundry, but as it enters its next phase, several questions need to be answered.
Threats beginning to mount
At the 28-nanometer technology generation, TSMC has a near monopoly on leading-edge capacity. This was largely due to the other foundry players making a key strategic error in their move to manufacturing processes that used a high-K metal gate as the gate dielectric. There were two competing approaches to implement a high-K metal gate strategy -- gate-first and gate-last. The former promised a density improvement, but gate-last was easier to build and offered performance advantages.
The IBM Fab Club -- Samsung, Global Foundries, and IBM -- went with a gate-first approach. TSMC went with a gate-last approach, which Intel did at its 45-nanometer process. TSMC was able to be first out of the gate with its high-K, metal gate 28-nanometer process. As a result, it gobbled up the majority of the leading-edge foundry business at that process technology. Eventually, Samsung and Global Foundries figured their issues out, but the lead that TSMC commanded was significant.
It seems that TSMC will have the playing field largely to itself at the 20-nanometer node. But then, it will face competition at the FinFET nodes. If TSMC doesn't end up having a lead at the FinFET node over the foundries, then this poses a significant risk to the robust revenue outperformance that it has seen over the last few years, as market share comes under pressure.
Additionally, with TSMC spending heavily on semiconductor capacity, if it is wrong about estimating the demand for chips built at its factories, then the company will start getting hit with non-cash utilization charges -- a hit to gross margin.
History tends to favor TSMC, but the competition is unprecedented
In the past, TSMC had robust competition from other semiconductor foundries. But TSMC ended up commanding the lion's share of the profitability in this space -- the vast majority of the stand-alone foundries struggle mightily to grow their top and bottom lines.
That being said, TSMC is now facing a new wave of competition from both Samsung and Global Foundries. The former is a juggernaut with highly lucrative mobile and DRAM businesses to help it fund future growth. The latter is a fast-growing name with nearly infinite resources, thanks to backing from the Mubadala Development Company, backed, of course, by the Saudi Arabian government. The competition will be incredibly fierce.
With that said, TSMC has fought many difficult battles before and has emerged victorious. As long as the company is able to offer competitive prices and is able to maintain a technological edge over Samsung and Global Foundries, it should have the vast majority of the mobile apps processor orders. In particular, it is imperative that TSMC lure away Apple, which is a major buyer of leading-edge as well as trailing-edge wafers. This not only rips away a large chunk of Samsung's foundry orders, but it also is a nice revenue stream for TSMC.
Samsung, in particular, is in an interesting position as an IDM that competes with its own customers. For instance, Samsung is working feverishly on its own cellular baseband and apps processors in a bid to displace Qualcomm's (QCOM +0.16%) parts from many of its phones. The twist here is that Qualcomm is a foundry customer of Samsung's. Why should Qualcomm wish to enable its potentially largest long-term threat if it can help it?
Global Foundries is another pure-play foundry, but its execution track record is less than stellar. As long as TSMC can stay a step or two ahead on technology and yields, then there's no reason why TSMC would lose material business to this particular player. Although, it's unwise to underestimate just how much Mubadala is willing to back its currently deep-in-the-red investment.
Foolish bottom line
With TSMC's business on the cusp of getting much more competitive, and with new management in place, it will be very interesting to see how the company transitions into the next phase of its corporate life. History tends to favor TSMC, but it's hard to forget how bad things got when recently retired CEO Morris Chang retired the first time.
