Description

This is an exceedingly simple idea. I recommend a short of United Microelectronics (UMC).

UMC is the 2nd largest Taiwanese fab, playing (distant) second fiddle to TSMC. Over the years it's been somewhat mismanaged and fell further and further behind TSMC. A few years ago, it announced a strategic shift -- it will no longer invest in leading edge nodes. Instead, it will basically camp on legacy nodes and focus on niche technologies. Its forward node is 28nm which is obviously quite a ways behind. 

The hope was to turn the company into Vanguard, which is focused on REALLY lagging nodes and 8in wafers but yet dominate its particular niche and generate very strong margins. 

There are some significant differences though between UMC and Vanguard. 

1. 28nm is a crowded node. It's really hard to differentiate.

2. I expect SMIC (Chinese fab) to add significant capacity and throw off the supply 

3. Because of the recent shortage, other players are adding capacity as well

The strategy hasn't worked that well for UMC until really 2020. Margins were pretty low although FCF went up a ton as capex dropped precipitously. The firm paid out a sizable dividend as a result.

But last year (continuing onto this year), the stock price went from roughly $3 to $12, before dropping slightly to $11.

A lot of you probably know about the massive capacity shortages at fabs, particularly for legacy nodes. UMC benefitted massively and aggressively raised wafer pricing. EBIT margins went from MSD to an expected 20%+ in 2021. Investors expect capacity shortages to persist for at least 12 months and maybe longer, which will allow margins to remain high. Hence, they are willing to pay a decent multiple for "peakish" earnings.

This doesn't pass the sniff test for me. We know from comments from a variety of semi companies that a lot of capacity is being put into legacy nodes. Ask any of the semicap guys and they will tell you trailing node equipment is HOT. Even UMC itself is investing.

Will legacy node demand remain tight for a very long time thereby allowing price increases vs the usual price deflation in the long term future? I personally doubt it.

Furthermore, there is a big threat on the horizon from SMIC. The Chinese fabs do not need to make money from capacity additions. This is a national security issue for China. Obviously, SMIC wants to advance -- into 14nm, 10nm and so on. But it's adding capacity at legacy nodes as well, which are much less subject to import bans. My guess is that legacy node economics will likely deteriorate in the future from more competition and excess capacity. Either case, UMC is hard pressed to respond as it's stuck in these legacy modes as opposed to trying to advance.

Overall, investors think "this time is different" as everyone in semis is complaining about tight capacity. This level of pricing power is unprecedented. They see the backlogs building and think it will continue forever.

The short case is that if you wait for another 12 months, the picture will look very different. Normalized earnings power is not that different from what it was 18 months ago and may indeed be worse as a lot of capacity got added (and will be added) at trailing nodes, perhaps causing excess capacity for a very long time. You also have "quasi irrational" competitors in the wings in the form of govt supported Chinese fabs.

Catalyst

ending of shortages