Summary

• TSMC remains a top AI investment due to its dominant market position, strategic alliances, and cutting-edge technologies, despite geopolitical risks.

• In the long term, I expect to see TSMC's key margins stay where they are now (on a seasonality-adjusted basis) and probably even go up slightly.

• OpenAI's partnership with TSMC for in-house AI chip fabrication is a strong bullish sign, reinforcing my conviction.

• Despite potential geopolitical risks, I maintain a "Buy" rating on TSMC, expecting a 22%+ total return this year.

• I am Oakoff Investments, a quantitative research analyst. I run the investing group Beyond the Wall Investing where I specialize in uncovering mispriced stocks based on proprietary Wall Street information.

JHVEPhoto

My Thesis Update

I initiated my coverage of Taiwan Semiconductor Manufacturing Company Limited (NYSE:TSM) in late November when the stock dipped slightly below the $200/share mark. At the time, I called TSMC "one of the best AI plays," meaning that despite clear and well-known geopolitical risks around the firm, its "dominant market position, strategic alliances with Apple and Nvidia, and cutting-edge technologies" were making its current valuation too cheap to ignore at the time. Fortunately for my short-lived coverage history, TSM stock recovered from its little dip and outperformed the broader market quite meaningfully:

Seeking Alpha, Oakoff's coverage of TSM

Today, I still think that TSMC is too cheap to ignore: beating a critical player in the global tech supply chain, TSMC will likely have a strong pricing power amid the heated war between different LLM providers, GPU makers, and other AI-related niche players. I'm not afraid of their reduced revenue guidance because a one-off event drove the updated estimate. Also, the recently announced partnership with Open-AI only strengthens my conviction. Let me elaborate on all that.

Latest TSM's Results And Events

In mid-January, TSMC reported for its Q4 FY2024, closing its fiscal year in a strong way, delivering a sequential revenue increase of ~14.3% in NT dollars (~$26.37 billion, which is a 33.10% YoY growth), primarily driven by "strong demand for its cutting-edge 3-nanometer and 5-nanometer tech", according to the press release. That was more than enough to beat the consensus, and the same is true regarding their EPS figure (although the positive surprise magnitude was slightly lower compared to what we saw in the previous few quarters).

While the ramp-up of the N3 technology did exert some dilutive pressure on margins during the quarter, the stronger-than-expected demand factor made it possible to offset the negatives, so the gross margin increased to 59% (vs. 57.8% in Q3 FY2024 and 53% in Q4 FY2023). The existing operating leverage led to further efficiencies down the income statement, i.e., TSMC's EBIT margin went up by 150 and 740 bps QoQ and YoY, respectively, which is insanely good. As a result, the ROE increased to 36.2% from 28.1% - that's in a matter of 1 year:

TSMC's IR materials

What I really liked in Q4 numbers was the fact that the contribution of advanced technologies to TSMC's consolidated revenue structure continued to grow. I mean, we see that about 74% of wafer revenue in Q4 came from advanced technologies (7-nanometer and below), which is up from 58% for FY2023. In particular, N3 contributed ~26% to wafer revenue during Q4, while N5 and 7-nanometer technologies accounted for 34% and 14%, respectively.

TSMC's IR materials

That means that the operating margin increase we saw for the quarter is likely to stay with us longer than if it was just driven by cost cuts or one-off events. The demand for TSMC's advanced tech pushes its highest-margin segments to the upside, allowing them to use their pricing power at full scale as there are basically no similar-moat competitors in the market right now (at least I can't see any).

Of course, we saw a negative thing a couple of weeks after the Q4 numbers came out: TSMC's management warned that they're likely to report revenues at the lower end of their guidance because of the January earthquake in Taiwan.

TSM said that it expects revenue for Q1 2025 to be closer to the lower end of the guidance range of $25 billion and $25.8 billion, due to the earthquake that occurred in January.

The chipmaker estimates losses from the earthquake to be about NT$5.3 billion (~$161 million), after insurance claims.

Source: Seeking Alpha News, notes added

So yes, the seasonally weak Q1 will likely be quite weak, but I have to note here that the main driver for that weakness is actually a one-off event. On the larger picture, we're unlikely to see any business deterioration going forward as the management maintained their guidance for solid margins during the reporting period:

The company maintains a Q1 gross profit margin between 57% and 59% and an operating profit margin between 46.5% and 48.5%. TSMC said that it is working to recover the lost production, and there is no change to its full-year outlook.

What I think should be at the core of our focus is the firm's partnerships as the AI race is getting more competitive.

As Seeking Alpha reported recently, OpenAI is reportedly finalizing the design of its in-house AI chip and plans to partner with TSMC for fabrication. As Reuters reported, OpenAI tries to reduce its reliance on existing chip suppliers like Nvidia (NVDA) and "optimize chip designs specifically for AI inference workloads." In other words, they're likely looking toward creating a vertically integrated business structure, which is a nice way to potentially invest the money from the upcoming new round of funding. But as OpenAI spends, TSMC earns, and TSMC appears to be the most likely partner for OpenAI if they are indeed pursuing in-house chip manufacturing. So, on a net-net basis, this news seems to be quite negative for Nvidia and benefits TSMC.

In the long term, I expect to see TSMC's key margins stay where they are now (on a seasonality-adjusted basis) and probably even go up slightly as the firm's advanced tech occupies a higher revenue share over time. This way, the current projections regarding the firm's sales and EPS are likely too pessimistic, even if they imply nice 3-year CAGRs of 12.26% and 8.93%, respectively. In particular, I'd wait for higher estimates regarding TSMC's bottom line as the current consensus seems to miss the bigger picture of today's operating leverage sustainability, i.e., it factors in a way higher revenue growth rates rather than earnings per share growth rates.

TSMC Stock Is Too Cheap To Ignore

While I'm thinking of higher margins for TSMC in the long run, the management gave some food for thought for the medium term. While looking at FY2025, the management anticipates a gradual reduction in the dilution impact from the N3 ramp-up, TSMC's overseas expansion should keep exerting some pressure on margins "due to higher costs associated with a smaller scale, supply chain, and ecosystem maturi,", based on the earnings call transcript. So TSMC's executives project that U.S. fab margins "will be lower than overall margins by 2-3% over the next 5 years."

Anyway, the net 2-3% impact doesn't look too severe, and I still think that TSMC's valuation multiples should adjust higher, driven by the longer-term marginality improvement:

YCharts, Oakoff's notes

Today's forwarding P/E ratio of 23.07x is a 22% cut compared to the TTM figure, while the TTM figure itself isn't that high compared to the IT sector's 25.36x. I should remind you here that TSMC's valuation should be theoretically inflated by the fact that the firm is "one of a kind" because of its wide moat.

I believe TSMC's fair P/E multiple for FY2025 lies in the range of 26-30x. Considering the firm's past EPS annual surprises, I'm taking an average surprise of ~0.5% and applying it to the FY2025 EPS of $9.02. Multiplying everything on each other, I get a price target of $253.8, which is ~22% above the last closing stock price. Add to this TSMC's dividend yield of ~1.5% for FY2025, and you'll get a nice GARP stock right away.

Risks To My Thesis

I may be underestimating the geopolitical impact on TSMC in the long term. First, we know that the battle between the US and China is getting new shapes after the AI boom, and now TSMC has to make complex maneuvers to "sell shovels" and comply with the US at the same time, which is adding a discount to its risk profile. After Taiwan Semiconductor spokesman said recently that the company "complies with all applicable laws and regulations and is fully committed to complying with the new export control rules announced," we saw the stock's price negative reaction as the market participants likely started to price in lower revenue in the future. So, I think these kinds of statements and reactions will influence the stock price for the time being. Second, TSMC is located in Taiwan, as the company names itself suggest, and as China becomes more powerful and comes out of its local crisis, I think China will likely attempt to annex Taiwan at some point, which will help the main risk over TSMC materialize.

Xi has several key objectives as president, including completion of the PLA modernization by 2027, the annexation of Taiwan, progress of the Belt and Road Initiative, and most importantly, the National Rejuvenation by 2049 (the cen- tenary of the PRC).

Source: U.S. Department of Defence

I also may be wrong in the way I value TSMC stock today. I mean, what if the TSM stock trades at 25x earnings next year, and its EPS will be 5% lower than currently expected? The resulting price target will drop to $214/share, just 3% above today's stock price.

Your Takeaway

Despite the risk factors that I've described above, I still think that TSMC is one of the greatest AI plays that few people are talking about. Although the management warned about the negatives coming from the one-off event that should result in lower-than-anticipated revenue figures for Q1, I believe the recently announced plans of OpenAI to use the firm's tech for its future in-house developed chips are a very bullish sign that gives some kind of bias confirmation to my conviction. I maintain my "Buy" rating on TSMC and wait for 22%+ total return this year.

Good luck with your investments!

Editor's Note: This article discusses one or more securities that do not trade on a major U.S. exchange. Please be aware of the risks associated with these stocks.