In February 2023, TSMC observers were surprised by the news that Berkshire Hathaway had sold its nearly US$4 billion stake in TSMC. Why would Warren Buffet sell the shares of the leading semiconductor company of the world? After all, this is a company with fat 55% profit margins and whose revenues and profits have more than doubled over the last five years. It is a company that essentially has a monopoly on the most advanced semiconductors without which the world would grind to a halt. 

There is no question about TSMC’s dominance of a critical technology. However, what Warren Buffet may have realized, upon closer examination of TSMC’s finances, is that its Achilles heel may be what it has chosen to do with its enormous profits since 2020. And what’s even more important is what that tells us about what TSMC’s management is likely to do in the future with the shareholder’s money.

Prior to 2018, TSMC spent about 30% of its revenues on capital investment. This changed with the global boom in semiconductors that started in 2020.  From 2018 to 2022, TSMC’s revenues and profits more than doubled.

But TSMC’s capital spending increased even more. By 2022, its capital expenditures reached 48% of revenues. As a result, none of the windfall profits TSMC earned after 2020 was returned to the company’s shareholders. All of it was plowed back into the company in the form of spending on new facilities, including TSMC’s new facility in Phoenix, Arizona. 

Capital expenditures of this magnitude are rare in mature companies such as TSMC, and are typically only seen in startups. But there are also good reasons for us to be skeptical that the enormous investments will eventually pay off for TSMC. 

First, there is clear evidence that improvements in chip technology have become enormously expensive. Moore’s law may not have come to an end so far, but it has only been delayed because of the increasing amount of money leading chipmakers such as TSMC and Samsung have been willing to spend. 

But at what point is it too costly to keep Moore’s law alive? Is it in the interest of TSMC’s shareholders to spend even more money for marginal improvements on semiconductor technology?  Is it possible that TSMC passed that point already, when improvements in quality simply cannot justify the enormous capital expenditures needed? 

In recent years, most of TSMC’s competitors have decided that they cannot justify the costs needed to push out the frontier of semiconductor technology. Companies such as Global Foundries and Taiwan’s UMC decided instead to focus on the so-called “mature” nodes. The disadvantage of course is that their profit margins are much smaller than that enjoyed by TSMC.  But the huge advantage of focusing on mature technologies is that most of their profits go to the shareholders because their capital expenditures are much lower than TSMC.

In the past, some of the return to TSMC’s investments have come precisely because its competitors dropped out of the market for the advanced chips. But this is not the situation TSMC faces today as the effective monopolist of high end chips. Its future profits will come only because of higher demand from better performance of its chips or more applications of its chips. There are no more competitors left to cede the market to TSMC. 

And the fact is that the shareholders of the companies that have ceded the market of high-end chips to TSMC have done better than TSMC in recent years. 

Take UMC, for example. UMC is much smaller and receives much less publicity compared to TSMC.  But its shareholders have done better than those of TSMC. Every dollar of an investor’s money in UMC in early 2018 was worth 3 dollars and 33 cents by the end of 2022. The equivalent number for TSMC’s owners over the same period is 1 dollar and 90 cents. The gap in shareholder return is even larger if one chooses the peak of TSMC’s stock price (December 2021) as the end period. Every dollar invested in TSMC in early 2018 was worth 3 dollars by December.2021. The equivalent number for UMC was 5 dollars and 55 cents. 

The difference is what UMC has chosen to do with its windfall profits compared to TSMC. UMC chose to return much of the profits to its shareholders (through higher dividend payouts) instead of spending the windfall on new facilities in Taiwan and the U.S. In contrast, TSMC’s dividend payout was unchanged despite its record breaking profits between 2020 and 2022. 

What could be going on is a phenomena one sees in many companies with windfall profits, in that managers believe that profit windfalls are not temporary but permanent. This is seen very clearly in oil companies that go through boom and bust cycles that are made even worse by their investment decisions. In periods of boom, oil companies invest as if the boom in oil prices will be permanent. Of course, they almost never are for oil companies, and their profits drop by even more than necessary when oil prices drop because of the overhang of previous investments that are unprofitable with low prices. 

To be sure, financial markets (and Warren Buffet) could be wrong.  It could be the case that the boom in semiconductors after 2020 is not temporary but is instead a harbinger of exploding demand, say from artificial intelligence or autonomous vehicles, in the future.  Perhaps TSMC is on the cusp of the next technological revolution in semiconductors, and TSMC’s customers will be willing to pay a premium for better chips.  Or perhaps TSMC’s customers in the U.S. will pay substantially more for chips made from its Arizona facility compared to identical chips made in Taiwan. 

Whether TSMC continues to flourish is critical for Taiwan and the world. But for a company to prosper, it is important not only to know when to invest, but also when not to.  

The enormous investments made by TSMC in the last three years may eventually pay off.  But it would be wise for TSMC to indicate clearly the circumstances at which they will say “no” to further spending.  

About the Author

Chang-Tai Hsieh is the Phyllis and Irwin Winkelried Professor of Economics and PCL Faculty Scholar, University of Chicago Booth School of Business

Have you read?

• Will PX-Mart and President Chain Store monopolize Taiwan’s retail? 
• Evergreen Marine profits jump 9-fold as era of mega vessels dawns 
• Fuzhou pepper buns and the failure to scale

Uploaded by Ian Huang