TSMC Is Printing Record Profits. A Naval Blockade Is the Only Thing Keeping It Cheap.

Demand for TSMC's 3-nanometer chips — the most advanced you can buy — keeps outpacing how fast TSMC can make them. When you're the only factory that can build what Apple and Nvidia need, you get to name your price. And with a line out the door, demand keeps exceeding supply.

This isn't just about revenue growing. It's about margins expanding. TSMC's costs don't spike just because AI demand doubles. The factories are already built. The engineers are already hired. Every additional dollar of revenue drops to the bottom line at a higher rate.

That margin expansion is exactly why ASML's order book next week matters so much — it'll signal how aggressively TSMC plans to maintain its pricing power.

When you can't make chips fast enough, you buy more machines. There's only one place to buy the most important one. ASML reports earnings on April 15, one day before TSMC's own earnings call.

Here's what most investors miss: ASML's order book doesn't just tell you about current demand. It tells you how fast TSMC is pulling forward capacity expansion. If ASML reports a surge in EUV orders beyond what TSMC's publicly guided Arizona timeline would require, that's a signal. It would hint that TSMC's leaders are more worried about being concentrated in one region than their polished investor slides let on.

A strong ASML quarter confirms the capacity crunch is real and ongoing. But an unexpectedly large order book could reveal something more important — that TSMC is quietly racing to diversify faster than it's telling shareholders.

TSMC has committed staggering resources to its Arizona expansion. Its outstanding guarantees to TSMC Arizona total roughly NT$350 billion. It's also lending funds to a subsidiary in Washington state. These aren't small bets — they're tens of billions of dollars flowing into American soil.

But building chips outside Taiwan is fundamentally harder. Labor costs are higher. The supply chain ecosystem that took decades to build in Hsinchu doesn't exist in Phoenix — every part of the supporting infrastructure has to be replicated from scratch.

The Arizona buildout will compress TSMC's margins in the medium term. But here's the flip side: many big Western funds hold less TSMC than normal — or skip it entirely — because their risk rules flag the geopolitical exposure. A credible U.S. Manufacturing base could unlock a new class of buyers who are currently sidelined — potentially more than offsetting the margin drag with multiple expansion. Investors need to decide which effect wins.

TSMC's market cap — the total value of a company's outstanding shares sits at roughly $1.9 trillion as of mid-April 2026. The stock has more than doubled from its October 2025 low. Yet the market still prices it at a notable discount to American tech peers growing at similar rates.

The Iran blockade this week didn't directly affect TSMC. But it rattled investors because unrelated geopolitical shocks in one theater get used as proxies to reprice tail risk — the small chance of a catastrophic loss in another. If a naval blockade can happen in the Persian Gulf, portfolio risk models quietly mark up the probability of disruption in the Taiwan Strait.

Here's where it gets specific. If you compare TSMC's forward earnings multiple to a hypothetical U.S.-domiciled company with identical growth and margins, the gap implies the market is pricing in a meaningful probability of severe disruption within the next several years. That discount is either the market being smart about a risk nobody can quantify — or it's the single largest mispricing in global equities. The April 16 earnings call won't resolve that question, but it will tell you whether the business fundamentals are strong enough to keep widening the gap between what TSMC earns and what the market pays for it.

Intel joining the project changes its credibility in a big way. This is the world's second-largest chip maker, with decades of factory expertise. But even Intel has spent hundreds of billions trying to close the gap with TSMC's leading-edge process, and it's still behind by multiple process nodes.

Building cutting-edge chips takes decades of hands-on know-how. It demands specialized equipment only ASML makes and tens of billions in capital — before you produce a single working chip.

Terafab actually strengthens TSMC's moat. It shows how badly big companies want a backup chip source — and how far off any real alternative still is, even with Intel involved. Nobody would try to build a fab from scratch unless the existing options were insufficient. That's not a threat to TSMC. It's a testament to how irreplaceable it is.