The best entries usually feel terrible. They come on a red day, against a scary headline, when the tape is screaming that you’re wrong. Micron in late March 2026 was exactly that kind of trade.

01 / The Bad NewsGoogle drops a bombshell on memory

On March 26, 2026, Google announced TurboQuant: a new compression algorithm that makes large language models as much as 6x smaller and up to 8x faster in inference, with no loss in fidelity. Memory-chip makers, already under pressure from the broader “Saaspocalypse” software derating, completely shit the bed. Micron (MU), SanDisk (SNDK), Western Digital (WDC) and Seagate (STX) crashed instantly and kept spiraling the next day.

The market’s logic was simple and, on its face, plausible: if AI suddenly needs far less memory to do the same work, then demand for memory chips is about to fall off a cliff. Sell first, ask questions later.

02 / The ThesisJevons’ Paradox, hiding in plain sight

Jevons’ Paradox states that when a technological breakthrough massively improves the efficiency of some resource, total demand for that resource goes up, not down. The classic example is energy: when coal-burning got dramatically more efficient in the 19th century, Britain didn’t use less coal. It used vastly more, because efficiency made coal cheaper to deploy across more uses.

The same dynamic applies to AI memory. Lowering the cost per unit of compute doesn’t shrink the market; it unlocks new applications that were previously too expensive to run. Every time computing has gotten cheaper, total demand has exploded. So a 6x compression breakthrough isn’t a headwind for memory makers who innovate alongside it, it’s rocket fuel.

“The market believes efficiency in AI memory means lower demand and a death spiral for memory makers. I believe the opposite is true, and this is a textbook Jevons Paradox setup.” — Mark Jeftovic, The Sovereign Capitalist, April 2026

03 / Why MicronThe worst chart, the best business

Not all memory makers are created equal. Of the four names that got hit, Micron had arguably the ugliest chart, it had fallen below its 50-day moving average and printed a long string of red candles. That is precisely what maximum pessimism looks like, and it’s often where the fat pitch hides.

But the chart was the only ugly thing about it. Barely a week before the selloff, Micron had reported blowout earnings, and the panic was now offering that same business at a discount:

High-bandwidth memory is the component that feeds the most advanced AI accelerators on the market, and Micron’s was spoken for through the end of the year. A company with 74% gross margins and a sold-out flagship product, trading in the single-to-low-double-digit P/E range, is not a business in a death spiral. It’s a high-quality cyclical being priced as if the music just stopped, which is the textbook overreaction to a media-driven scare.

“This is a fat-pitch, picks-and-shovels, maximum-pessimism play overreacting to media hype. Love it.” — Mark Jeftovic, The Sovereign Capitalist, Trade Alert, March 27, 2026

04 / The FitA pick-and-shovel in the Post-Singularity Stack

This wasn’t a one-off punt. Micron slots cleanly into the portfolio’s evolving framework — what we call the Post-Singularity Stack, summed up as “Sound Money, Smart Machines, Scarce Resources.” The idea is to own the companies and resources that underpin whatever comes next in an AI-accelerated world, and to favor picks-and-shovels over hype.

You don’t have to guess which AI model wins to profit from the build-out. Whoever wins, they need memory, and the more efficient that memory gets, the more of it the world will use. That’s the picks-and-shovels logic: sell shovels to every miner, regardless of who strikes gold.

05 / The UpdateThe panic was the gift we thought it was

The thesis played out about as cleanly as these things do. The TurboQuant scare faded, demand for high-bandwidth memory kept climbing, and the most-hated chart in the group turned. By May 2026 the position was up roughly 50% from our entry. It has, in the portfolio’s own words, been “all upside since we entered.” We’re still holding; this remains an open position in the Post-Singularity Stack rather than a closed, booked gain. But as a real-time test of the Jevons read against the market’s “efficiency equals doom” reflex, the early scoreboard is decisive.