Semiconductor Volatility Is My Friend

And It Can Be Yours Too

Every time semiconductors sell off, the same people crawl out of the woodwork to announce that “this time it’s over.” Now the script is a “10 year backlog,” “too much AI capex,” and “the bubble has finally popped.” It sounds dramatic. It also doesn’t line up with how basic economics or this industry works.

If you take their story seriously, they’re describing years of demand that exceed supply. Some say that means AI and chips are fizzling out. What it really means, however, is record profits.

Some are saying smartphone demand is lower; I even had one YouTuber try to argue that smartphones are facing a nine year backlog in the future. So the bears are all over the place and can’t even agree on what the future holds or even what is going on in the present.

Smartphones are high volume, but they’re not where the fattest margins live. The real money is in AI data centers and other premium customers that have to get silicon or nothing they’re building works. In a genuine squeeze, chipmakers serve the customers who pay the most, raise prices because demand is strong and capacity is constrained, and let low margin volume scramble for what’s left (typically older chips).

All that extra profit doesn’t vanish into thin air. It shows up as free cash flow, and free cash flow is what they use to build more capacity. Scarcity, sticky demand, rising prices, and capacity expansion are not signs of a sector that’s dying. They’re signs of one that’s doing what it’s supposed to do when the world is begging for what it sells.

Now let’s talk about the capex panic. Yes, the AI names are spending a ridiculous amount of money on infrastructure. No, that doesn’t automatically mean they’re insane. These are companies sitting on huge operating cash flow, dominant market positions, and cheap access to debt. They’re not guessing with rent money. They’re building because they think the returns are there, and right now their behavior looks a lot more like “we’re laying down a new layer of the economy” than “we’re YOLOing into a meme bubble.”

Capex only becomes a real problem when the spending outruns reality: when demand rolls over, pricing power disappears, and margins get crushed so badly that new investment never pays back. That’s not what the semiconductor and AI infrastructure complex looks like today. What we see instead is tight capacity, premium customers signing multi-year deals to lock in supply, and chipmakers getting paid well for it. You can slap the word “bubble” on that if it makes you feel smart, but the cash flows don’t care what you call them.

As far as how my semiconductor-heavy income portfolio is doing, even after the recent drop, I’m up 36.54% in total returns this year. My trailing yield is about 50%, my yield on cost is 57%, and I’m still up roughly 12% in capital appreciation. If I slip into the red, I’ll buy more. Not because I enjoy pain, but because I know the difference between a temporary sentiment tantrum and a broken thesis.

Every time the bears have yelled, “It’s over, the AI bubble has popped,” the pattern has been the same. It’s been scary headlines, volatility, and then a recovery. Volatility has been one of my best friends since 2020. I love it. If the underlying economics still make sense, volatility isn’t a warning, it’s a sale. Warren Buffett and Charlie Munger both argued that volatility isn’t risk. Sometimes it is, but in this case, it is not a risk. Rather, it’s an opportunity. Now is a great time to get into the chip sector or to increase your position.

I’ll keep collecting distributions, using the drops to my advantage, and letting math and patience do what they’ve already done for me, over and over again. Print me money.