To The Moon

When Stanley Druckenmiller quietly loads up on a stock, it’s usually worth a closer look. His recent interest in Taiwan Semiconductor Manufacturing Company—better known as TSMC—has sparked plenty of market chatter, and I’ll admit, it got my attention too. At a glance, TSMC is riding high on the AI wave, delivering staggering growth numbers, world-leading chip capabilities, and a forward roadmap that reads like the blueprint for a $3 trillion company. But as always, it's not just about hype—it’s about whether the fundamentals can keep pace with the valuation.

Advanced Nodes, Advanced Profits

TSMC’s Q2 earnings offered a glimpse into just how dominant the company has become. Net profit soared 60% year-on-year to NT$398 billion, with revenue touching nearly US$30 billion—a 39% increase from the same period last year. Most strikingly, 3nm and 5nm chips now generate over 60% of wafer revenue. These aren’t your average semiconductors. They’re the core logic driving $NVIDIA(NVDA)$ AI accelerators, $Apple(AAPL)$ newest mobile chips, and the brains behind an increasing number of high-performance computing systems.

Where logic meets light — the AI arms race begins

In other words, $Taiwan Semiconductor Manufacturing(TSM)$ is no longer just a manufacturer—it’s the beating heart of modern compute. As far as I’m concerned, that moat is getting deeper with every new process node it perfects. That 2nm milestone, expected to hit mass production in late 2025, could be a game-changer in performance-per-watt metrics that matter deeply for AI workloads. And with Intel still struggling to match yields and Samsung trailing in reliability, the crown appears secure for now—though perhaps less permanent than it seems.

Valuation: More Nuanced than Meets the Eye

You're right to question whether the 24.3x forward P/E is truly 'reasonable' in the grander scheme. It feels fair today—given the 30% revenue growth guidance and the AI euphoria priced into tech. But we both know semiconductors don’t grow in straight lines. TSMC’s current margins and growth trajectory are flattered by a sharp AI upcycle that could normalise faster than investors expect. Historically, when inventory gluts arrive—and they always do—they’ve been painful.

So, the real question becomes: can TSMC sustain a 20–30% annualised growth rate into the second half of the decade? I’d say probably not—unless another disruptive compute paradigm takes off. A more grounded case might assume mid-teens growth with margin normalisation, especially once the initial datacentre buildout matures. That would push the valuation closer to the edge of being fully priced, particularly if gross margin compresses back toward the 50%–52% range from its current highs.

The Competitive Landscape: A Moat with Moving Edges

TSMC’s moat is substantial, but it's not immune to erosion. Intel’s comeback ambitions aren’t laughable anymore. Under Gelsinger, they're throwing $100 billion-plus into a five-year plan to reboot their foundry business, and they’ve already won early commitments from $Qualcomm(QCOM)$, $Amazon.com(AMZN)$, and even the U.S. Department of Defense. Execution remains Intel’s Achilles heel, but capital and geopolitical backing are in their favour.

Samsung is equally hungry, with deep pockets and an aggressive roadmap to close the yield gap in GAA (gate-all-around) transistors. Their push into 2nm—targeting commercial deployment by 2025—could put real pressure on TSMC if they solve yield and defect rate issues. If either rival succeeds, pricing power could shift back toward customers faster than TSMC expects.

Over the past year, TSMC has surged while Intel has slumped—underscoring the market’s growing conviction in Taiwan’s technical lead.

In short, TSMC’s lead is real, but not locked. It’s more of a moving target than a fixed moat.

The Numbers: Strong, but Temporarily Supercharged

TSMC’s 43% profit margin and 34.5% return on equity are fantastic—but they’re not necessarily sustainable. These figures are being boosted by peak utilisation rates on bleeding-edge nodes, and a very favourable pricing environment. Historically, TSMC’s net margin has averaged closer to 35%, and ROE in the mid-20s, which is still elite—but clearly lower than today’s supercharged level.

If we see just a modest inventory correction or delay in AI capex, these metrics could deflate. Free cash flow is robust at NT$700 billion, but that’s already after significant capex—and the company is guiding for even heavier spending through 2026. If the top line stalls, the free cash flow could shrink quickly. TSMC is essentially in 'build it and they will come' mode, which works well… until it doesn’t.

YTD: TSMC flirts with upper Bollinger band after sharp recovery

Cyclical Realities: A Bull Market in Capital Spending, for Now

The broader risk here is the cyclical nature of the business. The AI boom feels secular, but we’ve heard that tune before during crypto mining, smartphone super-cycles, and 5G infrastructure buildouts. At some point, hyperscalers may hit the pause button—either due to saturation or broader macro headwinds. Already, signs are emerging of cautious spending post-2026, especially if energy costs or regulation slow AI training scale.

Combine that with the capital intensity of global expansion—15 fabs across Taiwan, the US, Germany and Japan—and you’re looking at one of the most aggressive capex cycles in corporate history. That’s bold, but also risky if the return curve flattens. It’s the classic fixed-cost dilemma: if demand surprises to the upside, TSMC wins big. If it stumbles, the downside is magnified.

How I’d Frame the Investment Now

I still like the company—but I’m a little less certain about the timing than I was a few quarters ago. The story remains compelling, the financials are pristine, and the technology lead is undeniable. But the price is catching up to perfection, and the upside looks increasingly tied to continued AI investment velocity.

I wouldn’t say I’m pulling back from the stock. But I’d want to enter on a correction—perhaps during a broader market rotation or a temporary pause in AI-related euphoria. TSMC is likely to be one of the most important companies of the next decade. That doesn’t mean every quarter will be smooth sailing.

Final Thought

If you’re buying TSMC now, you’re not just betting on AI—you’re betting on continued leadership, operational excellence, and geopolitical stability, all at once. That’s a lot of things that have to go right. For Druckenmiller, that risk-reward may still make sense. For the rest of us, a bit more patience could offer a safer entry point.

The bet: brains, borders, and belief in precision leadership

@TigerStars @Daily_Discussion @Tiger_comments @Tiger_SG @Tiger_Earnings @TigerClub @TigerWire