The reason why I am so optimistic about the competition between TSMC and Samsung is that neither quantum technology nor in-memory computing will dictate the outcome before 2025. Even if Samsung is able to make a breakthrough, it will not enter commercialization and make a significant impact until sometime in the 2030s.
Is there a chance that Samsung may team up with Internet giants like Google, Amazon or Facebook for the purposes of capturing cloud, artificial intelligence (AI) or autonomous driving opportunities? Not all of America's Internet giants will choose to work with Samsung. Moreover, if one sides with Samsung, the others may go to TSMC for collaboration. For example, in the case of making Apple's application processors, Samsung once gained the lead but TSMC made all-out efforts to win it back. Samsung stood a chance of defeating TSMC around 2014 but it was probably the closest Samsung would ever get to TSMC.
Samsung may have appeared to be neck and neck with TSMC in recent years. In 2019, TSMC introduced an enhanced version of its 7nm node and Samsung answered the challenge with its 6nm process. In 2020, when TSMC launched its 5nm and 6nm process technologies, Samsung claimed its 5nm node was going into commercialization. In 2021, TSMC is highlighting an enhanced 5nm node while Samsung is boasting of its 4nm process. These technological breakthroughs that Samsung has hyped up has ended up being all sizzle and no steak. The market did not favor its 6nm or 4nm process. No world-class heavyweight customers seem to have adopted Samsung's advanced nodes for volume production. TSMC, on the other hand, saw its 5nm node contribute 14% of its first-quarter 2021 revenue and 7nm node 35%. The gap between TSMC and Samsung is quite evident. We can make the assumption that TSMC has secured over 80% of the global revenue generated from 7nm and more advanced processes. Some Korean market research organizations even put it at 85%. We'll stay conservative and assume it's 80%. Based on this assumption, TSMC will enjoy a quick return on investment while Samsung's investment will go down the drain. Samsung may be able to endure this situation for two to three years but if it continues into future generations of process nodes, Samsung will have to find ways to guarantee profitability in the coming years.
Let's not forget the fact that Samsung's mobile phone business contributed 75% of the corporation's earnings in its prime despite going downhill in recent years. Although Samsung's semiconductor business generates one-third of the corporation's revenue, it contributes half of the corporation's profits; but for the most part, the contribution is made by memory chips. Samsung has to ensure that its memory chips continue to sell well and at high prices.
Losing to Taiwan in the foundry sector may not be deadly to Korea, but that may not be the case if Korea stumbles in the memory market on challenges from Micron or even some newcomers like YMTC. If this happens, Samsung will be going downhill from there. Without its edge in memory chips and mobile phones, could Samsung be shifting focus toward the foundry business?
Samsung isn't really a challenger for TSMC. No one in the foundry sector is strong enough to beat TSMC. TSMC could only be intimidated by governments. Which governments? Could it be the US government, the Chinese government or the Taiwanese government? I'll let the readers figure out the answer.
(Editor's note: This is the seventh part of a series of analysis by DIGITIMES Asia president Colley Hwang about TSMC's competitiveness and the foundry sector.)
