Why Semiconductors Rip Highest and Crash Deepest

TL;DR Semiconductors carry as much beta as alpha — when the market drops, they drop more. I see SMH's 38.2% retracement as a reasonable first pullback target, but I refuse to let recency bias convince me they "always go up."

On Friday the SMH semiconductor ETF fell nearly 9% from its top in a single session. Almost a vertical drop. Watching it, I was reminded of what this asset class actually is.

The core issue is high beta

When semis go up, they're phenomenal. But that added alpha — those extra gains — comes bolted to higher beta versus the market. When the market drops, semis drop more. Historically and usually, that's exactly how they behave. They're also deeply cyclical, capable of crashes as epic as their rallies.

So I think SMH has room to pull back. The 38.2% retracement looks like a reasonable starting point for that pullback. Will I buy it there? Only if the price action lines up with my macro scorecard.

I need the technical level and the fundamentals to marry up

A good price alone isn't enough. The technical setup has to align with the fundamentals — the macro.

Right now, growth is coming in better than expected. PMIs for both services and manufacturing beat and sit well into the 50s, and consumer confidence — while not great in absolute terms — surprised to the upside versus expectations. The problem is inflation. CPI and the producer price index are rising faster than expected, and bonds are selling off as yields climb. Friday's jobs report was solid, and the market freaked out over what that means for rates.

I actually read good jobs as a positive. I'd rather have a strong economy with a little inflation than a weak economy with low inflation. But if inflation gets out of hand, it eats into consumer strength and corporate earnings, so I stay cautious on the current readings. My composite score sits at +4 — neutral, but teetering just below the bullish threshold of 5.

The real AI winners may be small businesses

There's nothing wrong with AI itself. I use it every single day. It's genuinely revolutionizing my own small business.

In fact, when we look back on all of this, I think one of the biggest winners will turn out to have been small businesses all along. While the mega-caps build the AI infrastructure, operators like me — and countless other founders — are reaping enormous benefits from the ability to code and lean on AI agents. It's easier than ever for an individual to build a business. It levels the playing field.

The trap of recency bias

And yet AI can absolutely get overhyped — just like railroads, electricity, and the internet before it.

Knowing your history is how you avoid repeating it. After the 2001 crash, semiconductors took roughly 6,178 days — about 17 years — just to get back to break-even, and that doesn't even account for inflation. I bring this up because new technologies breed the belief that something "always goes up." Only a couple of months ago, people were saying gold and silver never go down.

Recency bias is the default setting of human psychology. When semis rise every single day, sitting out feels foolish. So I don't declare the bull market over — but if it is over, I want to get out of the way. If I do buy, it won't be with heavy leverage and a wide stop. It'll be a small, calculated position betting on the bounce and continuation of what I believe is a genuine technological revolution.