THE PUPPET MASTER

(TSMC), (MSFT), (NBIS), (GOOGL), (SSNLF), (INTC)

There’s a phone call that happens every month between Cupertino, Seattle, and a modest office building in Hsinchu, Taiwan that determines more about the future of artificial intelligence than all the flashy product launches and CEO tweets combined. 

Most investors have never heard of this conversation, but the numbers flowing through it make OpenAI’s latest funding round look like pocket change.

What they’re discussing on those calls is Taiwan Semiconductor Manufacturing Company (TSMC), and what’s happening right now reminds me of the early days of the internet boom, except this time the infrastructure play is crystal clear. 

Although virtually unknown, this company is responsible for manufacturing the actual surfboards everyone else is standing on. 

Every single advanced AI chip that powers ChatGPT, Gemini, or whatever shiny new model launches next week gets manufactured in TSMC’s fabs. Not some of them. All of them.

Let me give more context about the current setup.

When Microsoft (MSFT) announces a $17 billion dollar deal with Nebius (NBIS) for AI infrastructure, or when Google (GOOGL) quietly acquires 50 hectares in the Netherlands for data centers, they’re not just buying land and concrete. 

They’re creating guaranteed demand for the most advanced semiconductors on the planet, and there’s exactly one company capable of manufacturing them at scale. The economics are breathtaking when you think about it. 

Every data center expansion announcement from the tech giants translates directly into chip orders that won’t be filled by anyone except TSMC.

Let’s look at the numbers. 

TSMC’s August sales jumped 34% year-over-year, which sounds impressive until you realize they’re already generating over $100 billion in annual revenue. Companies this size aren’t supposed to grow at startup rates, yet here we are. 

Their EBITDA exploded from $20 billion to $73 billion in less than five years while simultaneously increasing research and development spending. 

On top of these, the semiconductor foundry market itself is expanding at nearly 8% annually, but TSMC isn’t just growing with the market. They’re actually stealing shares from competitors. 

Their foundry market share jumped from 31% to 38% in 2024 alone. Samsung (SSNLF) and Intel (INTC) keep promising they’ll catch up on advanced node manufacturing, but the reality is that TSMC’s lead in three and five nanometer processes has actually widened. 

I’ve spoken with engineers who work on these production lines, and the technical barriers to entry aren’t just high, they’re getting higher every quarter.

Analysts are projecting earnings per share of $11.32 for next year and $13.46 for 2027, which would put the stock at less than 20x forward earnings. 

That’s where things get compelling for patient investors. 

The current price at around $260 reflects a lot of optimism, but it doesn’t fully capture what happens when AI spending shifts from experimental to essential infrastructure. 

We’re still in the early innings of data center buildouts, and the competition between AI platforms is intensifying rather than consolidating.

The management team raised their full-year guidance during the second quarter earnings call, which caught most analysts off guard. They’re typically conservative with projections, so when they signal confidence in sustaining 20% revenue growth through 2029, it’s worth paying attention. 

Most assume growth will decelerate significantly by then, but that also assumes AI demand plateaus. From what I’m seeing in enterprise adoption patterns, we’re nowhere near peak AI infrastructure spending.

The technical setup presents both opportunity and risk right now. 

TSMC’s RSI is approaching 70, which traditionally signals overbought conditions, and the stock has run up substantially since I last checked. 

Short-term momentum traders might take profits here, especially with broader market volatility creating uncertainty around high-beta technology positions. That’s actually creating the setup I prefer for long-term accumulation.

The geopolitical elephant in the room deserves mention, though it’s not the risk most people think it is. Yes, TSMC’s manufacturing is concentrated in Taiwan, and yes, that creates theoretical supply chain vulnerabilities. 

But here’s what the headlines miss – TSMC is building advanced fabs in Arizona and expanding operations globally precisely because they understand the strategic importance of geographic diversification. 

The U.S. government isn’t just allowing this expansion, they’re subsidizing it through the CHIPS Act because they recognize the national security implications of semiconductor independence.

For those positioned in AI plays, TSMC offers something unique: exposure to the entire ecosystem rather than betting on individual platform winners. 

Whether OpenAI maintains its lead or Google’s Gemini takes market share doesn’t matter to TSMC’s revenue stream. They manufacture the chips regardless of who writes the software. That’s the kind of positioning that creates sustainable competitive advantages.

The current valuation multiple of approximately 25x earnings isn’t cheap, but it’s reasonable given the growth trajectory and market position. 

More importantly, any significant pullback toward $200 would represent an attractive entry point for patient capital. The long-term thesis remains intact even if short-term momentum stalls.

Given the technical overbought condition and the solid fundamentals, this looks like a buy-the-dip situation rather than a chase-the-momentum play. 

Wait for weakness, accumulate on any meaningful pullback, and let the AI infrastructure buildout do the heavy lifting.