The Memory Supercycle Just Hit Its First Real Test

Micron capped a quarter that people will study for years, and the market rewarded it by knocking the stock down 8%. If you only read the price action you would think something broke. Nothing did. This is the first real shakeout of a cycle that is still very much intact, and how you read days like today decides whether you panic or accumulate.

The lawsuit is a signal, not a threat

This week Micron, Samsung and SK Hynix were hit with an antitrust class action alleging they held back DRAM supply to inflate prices during the AI boom. Forget the legal merits for a second. The useful signal is what the suit implies. Memory pricing has run so hard that plaintiffs now assume it has to be collusion rather than a genuine shortage. When your pricing power starts drawing lawsuits, your near term demand is not the problem. That is a top line worry disguised as a headline risk.

Why this cycle looks structural, not cyclical

I have traded enough memory cycles to be naturally skeptical, because memory has always been the most brutal boom and bust corner of tech. What keeps me long this time is HBM. High bandwidth memory is genuinely hard to build, yields are punishing, and adding capacity takes years, not quarters. Micron has its allocation locked under binding contracts, and the read across the group agrees. Sandisk is up around 700% this year on the same NAND tightness and Bernstein just slapped a 3,000 dollar target on it. Roundhill even launched a memory basket ETF so retail can own the whole shelf at once. Three names and a fund all shouting the same thing about supply is not a coincidence.

The SK Hynix listing is the capstone

On July 10 SK Hynix plans to list ADRs on the Nasdaq targeting around 29 billion dollars, with close to 60% of the global HBM market behind it. Think about what that raise is for. It exists to fund even more HBM capacity, and you only race to fund capacity when your customers cannot get enough. That is the loudest confirmation yet that this is a multi year buildout rather than a single hot quarter.

How I am actually positioned

I am staying long Micron and Sandisk as my core memory exposure, and I use the Roundhill basket for the piece I do not want to actively manage. I trimmed a little into the run because taking profit is never wrong, and today I added a small amount back under 1120. The real risk here is not the lawsuit, it is hyperscaler capex. If even one large buyer signals a pause, these names are priced for perfection and will correct fast. Until the demand data actually turns, I let the winners run with a hand near the exit.

Not financial advice, just how I read the setup.