Chips and Shadows

Amazon. It’s a name that’s been thrown around like a marked bill in a back alley. Big, sure. Dominant? For now. But the city always changes. There are other players moving in the shadows, building empires on silicon and promises. Two, in particular, have a scent of something more. Broadcom and Taiwan Semiconductor. They’re betting big on this AI craze, and the house might just win.

The usual suspects talk about market caps, growth rates. Numbers. I deal in probabilities, and right now, the odds are shifting. Amazon’s a two-trillion-dollar giant. TSMC and Broadcom are breathing down its neck, but chasing a number isn’t the same as building something real. Still, the numbers tell a story, if you bother to listen.

The Hustle

Amazon’s currently sitting pretty at $2.29 trillion. TSMC’s at $1.76 trillion, Broadcom a little further back at $1.57 trillion. Analysts expect Amazon to creep up at a 13% clip in ’26, 12% in ’27. Predictable. The kind of growth you expect from a well-oiled machine. If they keep that up through ’28, they’ll hit $3.25 trillion. A tidy sum. But sums can be illusions.

For TSMC and Broadcom to surpass that, they need an 84% and 107% jump, respectively, over the next three years. Ambitious? Sure. Impossible? Not if this AI thing keeps humming. The question isn’t whether they can, but whether the hype will hold. The market has a habit of delivering cold showers.

Silicon and Smoke

AI. It’s the new gold rush, and these two companies are holding the shovels. TSMC, they make the chips that run everything. The logic, the memory, the whole shebang. Nearly every AI player worth a dime relies on them. As long as the money keeps flowing, they’ll keep printing chips. It’s a simple equation, really. A little too simple, perhaps.

Broadcom’s playing a different game. They’re building custom AI chips, tailored to specific needs. A cheaper alternative to Nvidia’s GPUs, they say. But custom means limited. Like a bespoke suit that only fits one occasion. If a client needs flexibility, they’ll still go with the general-purpose option. But Broadcom’s betting that hyperscalers know exactly what they want, and that’s a gamble.

Broadcom’s CEO, Hock Tan, is talking about $100 billion in AI revenue by the end of ’27. They were at $8.4 billion just a quarter ago, up 106% year over year. That’s a fast burn, if it holds. They expect $14.8 billion this quarter. If they keep that pace, they might just push past Amazon. But numbers on a balance sheet don’t account for the unexpected. A supply chain hiccup, a design flaw, a change in the wind – any of those can derail a runaway train.

Analysts are predicting over $150 billion in revenue for Broadcom by the end of ’27, about 120% higher than their current total. They need a 107% return to surpass Amazon by ’28. Based on the current trajectory, it’s plausible. But the future is a blurry photograph, and projections are just educated guesses.

TSMC isn’t growing as fast as Broadcom, but they’re no slouch either. Analysts are predicting 31% growth in ’26 and 24% in ’27. If they maintain a 20% pace in ’28, they’ll see a 94% increase in revenue. They need 84% to pass Amazon. Solid. Reliable. Like a well-maintained engine. But even the best engines can stall.

These stocks will deliver returns, no doubt. The question is, will they be enough? And at what cost? The market is a fickle mistress. Don’t mistake a temporary surge for lasting value. Load up on shares if you must, but remember: every fortune has a shadow.

2026-03-11 21:03

The Hustle

Silicon and Smoke

Read More