Taiwan Semiconductor Manufacturing (TSM 1.60%) is the world's largest and most advanced contract chipmaker, and its top clients include Apple, Advanced Micro Devices, and Qualcomm. The Taiwan-based chipmaker also remains ahead of its closest competitors -- Intel and Samsung -- in the ongoing "process race" to produce smaller, denser, and more power-efficient chips.

That technological lead arguably makes TSMC the world's most important chipmaker. That's why its revenue grew at a compound annual growth rate (CAGR) of 15% from 2002 to 2022 in USD terms, and why it wasn't surprising when Warren Buffett's Berkshire Hathaway took a $4.1 billion stake in the chipmaker last year.

TSMC's foundry in Nanjing, China.

Image source: TSMC.

But over the past few months, Berkshire Hathaway -- which usually focuses on long-term returns -- unexpectedly sold its entire stake in TSMC. During Berkshire's annual shareholder meeting in early May, Buffett cited geopolitical tensions as the main reason for that abrupt sale. Should investors follow Buffett's lead and avoid TSMC? Let's review three red flags for the Taiwanese chipmaker's future -- and if they could ruin the bullish long-term thesis.

1. The geopolitical tensions

From 2008 to 2016, Taiwan's tensions with China eased under President Ma Ying-Jeou, whose Beijing-friendly Kuomintang (KMT) party opened up more trade and travel links with mainland China. But in 2016, the KMT's loss to the independence-leading Democratic Progressive Party (DPP) caused China to cut off most of those direct ties.

Under President Tsai Ing-wen, those tensions escalated as Taiwan sought to strengthen its alliances with the U.S. and Japan. Meanwhile, the U.S. blacklisted a long list of Chinese tech companies and chipmakers as the trade war intensified.

TSMC is caught in the middle. In 2020, the U.S. forced it to stop producing chips for Huawei and other Chinese chipmakers. The Trump administration also persuaded TSMC to build a new plant in Arizona, which will start producing 3 nanometer chips in 2024. That marks the first top-tier foundry TSMC has built outside of Taiwan -- but the KMT has blasted the move as an attempt by the DPP to "gift" TSMC to the U.S. in exchange for diplomatic support. 

To make matters worse, U.S. government officials -- including former national security advisor Robert O'Brien -- continue to openly discuss the idea of blowing up TSMC's Taiwanese plants if China attacks Taiwan. Those comments seem speculative and inflammatory, but they highlight the strategic importance of TSMC to both the U.S. and China -- and how it could quickly become a major flashpoint in a military conflict between the two superpowers.

2. Intel and Samsung aren't giving up

TSMC has stayed ahead of Intel and Samsung over the past few years, thanks to its early adoption of ASML's high-end extreme ultraviolet (EUV) lithography systems, which are used to produce the world's smallest and densest chips. But Intel and Samsung have also been installing more EUV systems to catch up to TSMC, and that competitive gap could narrow considerably over the next two years. TSMC plans to start mass producing its 2 nm chips in 2025, but Intel plans to reach its 2 nm node by 2024 while Samsung intends to achieve that milestone by 2025.

It's unclear if Intel and Samsung can catch up to TSMC, but they'll both have lots of government support. The U.S. government will subsidize some of Intel's efforts through the CHIPS and Science Act, which grants subsidies and tax breaks to U.S. chipmakers that manufacture their own chips. South Korea's government also recently approved a similar bill -- the K-Chips Act -- to subsidize the growth of Samsung, SK Hynix, and the country's other chipmakers.

TSMC could lose a lot of orders to Intel and Samsung if they achieve their 2024 and 2025 goals -- but those risks don't seem to be baked into analysts' expectations for TSMC's revenue to rise at a CAGR of 11% from 2022 through 2025.

3. Its growth is slowing and its costs are rising

TSMC's revenue declined 16% quarter over quarter in USD terms in the first quarter of 2023, and it's bracing for another 4% to 9% sequential drop in the second quarter. Those declines were mainly caused by the slowdown of the personal computer and smartphone markets, as well as sluggish enterprise spending on new data center and enterprise chips.

TSMC expects the semiconductor market to bottom out in the second half of 2023, but its raw material costs are still being inflated by the war in Ukraine. It plans to absorb most of those higher costs instead of passing them onto customers -- but it still plans to spend about $32 billion to $36 billion on capital expenditures this year to maintain its process lead. In other words, TSMC could struggle with slower growth and higher costs for at least the next few quarters. 

Is TSMC still worth buying?

These red flags all challenge the bull case for TSMC, but I believe it still has plenty of things going for it. Its stock is cheap at 16 times forward earnings, and I'd still invest in this industry bellwether unless a war actually breaks out over Taiwan (in which case many other stocks -- including Apple -- would also crash) or Intel and Samsung actually achieve their ambitious goals.