print-icon
print-icon
Add ZeroHedge as a preferred source on Google

Korea Crashes Again

Portfolio Armor's Photo
by Portfolio Armor
Friday, Jun 26, 2026 - 12:31

Working on microchips, Studio Ghibli-style.

Korea Crashes; AI Buildout Continues

In yesterday’s post, we argued that the AI buildout still has legs.

Micron’s earnings were the latest evidence. The company’s results showed that AI memory demand remains intense, supply remains tight, and the bottlenecks created by the buildout are still showing up in real revenue and earnings.

Then Asian tech stocks sold off overnight.

Korea was hit especially hard, with memory and AI-related names giving back part of their Micron-driven gains. That raises the obvious question: if Micron’s report strengthened the AI-infrastructure thesis, why are some of the most AI-exposed markets still trading like this?

The answer is that both things can be true.

AI is real. It’s a revolutionary technology, and we’re still early in the economic consequences. It will show up in self-driving vehicles, robotics, drug discovery, defense, industrial automation, software, logistics, and the replacement or augmentation of skilled human labor.

But the market is still trying to figure out who actually benefits.

Earlier this year, that uncertainty hit software stocks. The “SaaS apocalypse” trade was partly about whether AI would help incumbents, commoditize their products, or force customers to rethink what they were willing to pay for.

Now we’re seeing a related debate about the largest AI and hyperscaler names. The buildout is real, but the costs are real too. Investors are asking who captures the economics: the companies buying the infrastructure, the companies selling the scarce components, or the companies one layer deeper in the supply chain.

That’s where our focus is narrowing.

Bottlenecks Over Buzzwords

We don’t want generic AI exposure for its own sake.

In this market, we’d rather have exposure to real bottlenecks: memory, optical networking, timing chips, power systems, cooling, energy storage, passive components, and the other physical pieces that have to exist before the AI future can scale.

We’re also putting a little more weight on fundamentals.

That doesn’t mean we’re turning into a backward-looking value screen. Many of the most interesting bottleneck names won’t look cheap on trailing numbers. But all else equal, in a choppy tape, we’d rather lean toward companies with at least okay fundamentals than names where the entire case depends on the market ignoring current weakness.

The other adjustment is structural.

Some of this volatility may be less about fundamentals and more about market structure: crowded AI positioning, levered products, retail leverage, factor rotations, and violent swings in the most popular expressions of the memory trade. We want to get paid for volatility.

That’s the thread running through today’s trades.

The Next Bottleneck

One of today’s themes is capacitors.

Zero Hedge had a good piece Thursday (“Capacitor Frenzy”) on multilayer ceramic capacitors, or MLCCs, as a possible next AI bottleneck.

The basic idea is simple: AI servers don’t just need GPUs and memory. They need power delivered smoothly and instantly to those chips. MLCCs help regulate current, smooth spikes and drops, and filter noise so the system can run reliably.

As AI racks get more powerful, the number and quality of these components matter more.

The market already learned this lesson with memory. A boring component can become explosive when it turns into a bottleneck.

We already have an open trade in one MLCC-related name. Today, we’re adding more exposure to that theme.

We’re also adding exposure to precision timing for AI infrastructure, another essential part of high-performance computing systems. And we’re keeping one foot in biotech infrastructure, where the theme is not one binary drug trial, but the tools and containment systems needed for injectable drugs, biologics, and GLP-1-related demand.

Those are the kinds of themes we want here: not vague AI adjacency, but tangible infrastructure.

Using Structure Again

Volatility is high, spreads are wide, and several of these names have already moved. So we’re using options structures to lower our entry cost and give us asymmetric upside relative to risk.

In one of today’s MLCC-related trades, the call we want is trading around $13. Instead of simply buying that call, we’re aiming to enter a trade structured around it for about one-tenth of that cost.

That’s the approach we want in this market: find real bottlenecks, filter them for names that also pass our technical and fundamental screens, and use option structures that let volatility work in our favor.

If you'd like a heads up in real time when we place those trades later today, you can subscribe to our trading Substack/occasional email list below. 

Friday Afternoon Update 

Our trade alert went out this morning. We got some good entries during the early flush. 

 

Contributor posts published on Zero Hedge do not necessarily represent the views and opinions of Zero Hedge, and are not selected, edited or screened by Zero Hedge editors.
0
Loading...