Among Intel's (INTC) rivals, Advanced Micro Devices (AMD) receives a flood of attention but Taiwan Semiconductor Manufacturing Company (TSM) receives very little. True, Intel and TSMC are not direct competitors - Intel's commercial foundry business is tiny, and TSMC does not design its own chips. Still, TSMC is an enabler for Intel's direct competitors since its manufacturing capabilities allow fabless semiconductor firms such as AMD and Nvidia (NVDA) to compete with Intel in its core businesses. Consequently, it is a little surprising that there is not much comparative analysis of Intel and TSMC's future prospects and their relative attractiveness for investors. This article addresses this gap in one context and asks which of the two firms is a safer pick in the long term - i.e. which carries lower long-term downside risk.
My analysis suggests that TSMC's current competitive environment and basic business model make it a safer long-term pick than Intel, although in the short term, TSMC also carries greater risk for multiple contractions. Below I support this conclusion by looking at four significant differences in the risk profiles for Intel and TSMC.
1. Resurgent vs. Struggling Competition
The first major difference between Intel and TSMC is the competitive environment in which each operates. Here, the trends for the two firms point in entirely opposite directions.
On one hand, Intel is increasingly threatened by its competitors - in particular, AMD. AMD is making very real progress in improving processor core counts and multi-thread performance with the introduction of Infinity Fabric. Moreover, AMD will be able to leverage TSMC's fabrication lead over Intel once it ships its next-generation 7nm processors built on TSMC's latest process - which, unlike Intel's comparable 10nm, is already in volume production. Armed with these improvements, AMD now threatens to capture substantial market share from both of Intel's biggest businesses: client computing with Ryzen and data center with EPYC. As such, there is a reasonable likelihood that Intel will give up significant market share in businesses that only recently were basically monopolies.
Figure 1: AMD Zen Product Roadmap. Source: AMD
Naturally, we will have to wait and see how the dynamics play out with Intel and AMD's reignited rivalry. For now, though, Intel is losing ground. Given its extensive previous delays in node transition to 10nm, there is some chance that Intel may fall further behind TSMC (and via that AMD) in terms of fabrication. There is also some chance that Intel will have trouble keeping up with AMD's progress in architecture. Both risks are compounded by uncertainty over the appointment of the next CEO and how well he or she performs. As a result, Intel's competitive situation has decidedly been worsening for the last couple of years, and this trend could very well continue.
On the other hand, TSMC has been giving its competitors a run for their money. Nowhere was this clearer than when Global Foundries recently announced its decision to indefinitely delay further investments into its 7nm process and instead focus research and development on improving its existing 12nm and 14nm processes. As a result, TSMC's position at the cutting edge of semiconductor fabrication just became even more secure. TSMC is currently the only outfit with a 7-nanometer process in volume production, and it has only two challengers remaining at this node - each almost a year behind: Intel and Samsung (OTC:SSNLF).
Nor does TSMC show any signs of resting on its laurels. Per the most recent earnings call, TSMC plans to begin volume production on 7nm+ in Q2 2019, as well as risk production for 5nm in H1 2019 and volume production in H2 2020. TSMC even stated that work on 3nm has started. Consequently, pending execution, it looks reasonably likely that TSMC will manage to maintain its lead over Intel and the rest. If so, TSMC will enjoy a near monopoly at the leading edge for at least a couple of years, and perhaps even improve its margins with the increased market power it has now acquired.
In terms of competitive environment, then, TSMC looks like the winner. It is doing far better against its competition than Intel, and so is less likely to cede market share and profits to them.
2. Potential Technological Points Of Failure
Next, let's turn to potential technological points of failure for Intel and TSMC - i.e. areas where missteps could substantially affect market share. The key difference here arises from the two firms' different business models: Intel is an integrated semiconductor manufacturer whereas TSMC is only a foundry. As such, there are two potential points of failure for Intel, but only one for TSMC. Intel's market share can be threatened by missteps in either architecture or fabrication, but TSMC's can only be threatened by missteps in fabrication.
This situation is a direct result of Intel's traditional approach to selling processors: cutting-edge architecture combined with cutting-edge fabrication. Over the years, Intel has flourished with this approach as it has trounced competitors on both fronts. Now, however, there is room for worry since Intel is falling behind on both fronts. Going forward, Intel will need to get both architecture and fabrication right in order to retain market share - i.e. Intel will lose market share if it either missteps in architecture or in fabrication. If Intel falters in architecture, then being on par in terms of fabrication will not be enough to keep competing for products with better architecture at bay; and if Intel falls behind in fabrication, then being on par in terms of architecture will not be enough to keep competing products with better fabrication at bay. Hence, there are two technological points of failure where missteps by Intel could result in a decline in market share, and Intel will need to get two things right to maintain its dominant position.
TSMC is almost as dominant at the leading 7nm and 10nm nodes, but its business model is different. Because TSMC is only a commercial foundry, it only needs to get fabrication right in order to maintain its position. Its customers can worry about architecture, and many of them - including the usual suspects AMD, Apple (AAPL), Nvidia, and Qualcomm (QCOM) - are extremely good at it. Huawei is investing into architecture at the leading node as well. Consequently, although it is perfectly possible for TSMC to lose leadership in fabrication to Intel or Samsung, there is only one technological point of failure for it: fabrication. As long as it can maintain its technological lead there, it has too many capable and diversified customers to be in danger of running out of things to fabricate. If anything, the likely scenario right now seems to be that TSMC will keep growing its top line.
Now, call me simple-minded, but it seems to me that it must be easier to do one thing right rather than to do two things right. It must be easier to get fabrication right than to get both fabrication and architecture right. If so, then it straightforwardly follows that Intel's odds of a costly misstep are higher than TSMC's - especially since TSMC has been doing very well and expanding its fabrication lead over competitors while Intel has been giving up ground in both architecture and fabrication. To me, the horse that has already been winning seems more likely to keep winning.
Thus, in terms of technological points of failure, I again conclude that TSMC is the safer pick.
3. Ability To Capitalize On Secular Growth
The semiconductor industry is gearing up for secular growth in a number of segments - data center, big data, AI, automation, and the internet of things - all of which are expected to yield growth for many years to come. Who is more likely to capitalize on this growth: Intel or TSMC?
It is my contention that TSMC is significantly more likely to capitalize on these trends. TSMC's status as "everyone's foundry" means that TSMC can get a slice of these growing pies in the semiconductor industry as long as it can provide fabrication services to whoever strikes gold. There will most likely be at least a few fabless firms that benefit from these growth trends, so TSMC will most likely capitalize on them to at least some extent. In effect, TSMC is backing multiple horses in terms of architecture and will benefit regardless of which ends up winning.
Not so for Intel. Intel's success on these growing semiconductor segments depends on whether its particular architecture and fabrication strategies for capitalizing on secular growth pan out. If Intel missteps, then it could very well get shut out of impending growth - as it was shut out of mobile processors. Intel is backing only one horse - its own - and if it loses, then Intel will struggle to capture the aforementioned growing markets.
Consider data center, for instance. This is a market segment in which Intel already has a dominant presence, but even here, Intel's position seems more precarious than TSMC's. It is entirely possible that Intel will fail to capitalize on growth here if market share losses to AMD offset growth in TAM, which may even be the most likely outcome at the moment. Intel's success depends on the success of Intel's particular approach to data center processors panning out against the competition. TSMC, by contrast, seems to be in a much more secure position against these competitive dynamics. It will almost certainly capitalize on data center growth either way. If AMD captures market share from Intel, well and good. If it does not, TSMC should still capitalize on data center growth because it will most likely continue making data center GPUs for AMD and Nvidia.
Or consider the internet of things, widely expected to grow rapidly in the coming years. Whether Intel can earn meaningful income here remains debatable, since it is not a given that its internet of things offerings will prove successful. There is a reasonable chance, but no certainties. It is still early days, and the competition is stiff. In autonomous driving, for example, Intel potentially has to contend against Nvidia, Google (GOOG) (NASDAQ:GOOGL), Tesla (TSLA), and perhaps also Apple and AMD (especially if AMD's mysterious "embedded" semi-custom win that was supposed to be ramping around now turns out to be in automotive).
The situation in autonomous driving (and potentially other aspects of internet of things) is further complicated by the software element. Nvidia is making some serious progress in software and AI, and Intel will need to either design its own software or hope for the right ecosystem to evolve. Tough to say yet if it will succeed, so there is some software-related risk here as well, in addition to the usual risks associated with maintaining leadership in architecture and fabrication. So Intel's job is not easy, although certainly, it has fair enough odds of success.
By contrast, though, TSMC's odds look much better in internet of things. It seems far more certain that TSMC will earn meaningful income in the internet of things segment because, aside from Intel and Samsung, whoever breaks into the segment will likely come to TSMC for fabrication if it can keep executing well. TSMC need not worry about architecture or software: one or more of its customers will likely figure these out eventually. Per TSMC, "By 2025 it is estimated that the IoT's installed unit base will be ten times greater than that of smartphones." Even a small proportion of this TAM going through TSMC's fabs would be enough to meaningfully contribute to its earnings and the odds that some will seem pretty high given TSMC's diversified customer base. Hence, TSMC will likely get some share of the internet of things pie one way or another, but Intel might not.
Consequently, it seems reasonable to me to conclude that the odds that TSMC will capitalize on secular growth trends in the semiconductor industry are more certain than the odds that Intel will do so. TSMC can let the chips fall where they will and serve (some proportion of) whichever fabless firms succeed. Intel cannot let the chips fall where they will, because if its own forays into these growing segments fail, then it could be largely shut out of the corresponding growth prospects.
4. Risk of Multiple Contractions
So far, we have focused on Intel and TSMC's industry fundamentals, competitive environments, and business models. On these fronts, the risk profile for TSMC looks much safer than that for Intel. However, with reference to financials and stock price, Intel looks safer. Intel trades at lower multiples than TSMC. Although the bull market has generated multiple expansions for both Intel and TSMC, the results have been more pronounced for TSMC:
Figure 2: Intel vs. TSMC - P/E Ratio, P/S Ratio, P/B Ratio (Jan. 1, 2015-Present)
Figure 3: Intel vs. TSMC - Change in P/E Ratio, P/S Ratio, P/B Ratio (Jan. 1, 2015-Present)
True, TSMC's higher multiples are well-deserved. TSMC has been rapidly growing in its top line, bottom line, and FCF:
Figure 4: Intel vs. TSMC - Revenue Growth, Net Income Growth, Free Cash Flow Growth (Jan. 1, 2015-Present)
Still, higher multiples necessarily leave room for more multiple contractions. Especially in the case of an extended stock market downturn, stocks with high multiples tend to be hit the hardest, and although both Intel and TSMC trade at reasonable multiples, Intel is priced more conservatively. Consequently, should the markets see a period of extended decline, and especially if the ongoing market turbulence proves a harbinger for a more severe downturn, it makes sense that TSMC would be susceptible to a more significant tumble than Intel.
As such, although the higher multiples for TSMC are well-justified, TSMC still has a higher risk of multiple contractions. Consequently, with reference to risk for multiple contractions, Intel looks like the safer pick.
There we have it, then. The analysis suggests that TSMC is the safer pick in terms of its competitive environment, potential points of failure, and ability to capitalize on secular growth trends, but Intel is safer with reference to the risk for multiple contractions. In the short term, this last element may well prove decisive - if the current downturn in semiconductors and tech becomes something bigger, Intel could very well prove the safer pick because its multiples have less room to contract. However, TSMC's multiples are not unreasonable given its execution, and its competitive circumstances and business model should make it the safer pick in the long-term despite the potential for a short-term multiple contractions.
What do you think? If you had to pick only one, would you go for Intel or TSMC?
Disclosure: I am/we are long AMD, INTC, NVDA, TSM.
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.