Serenity, the small-cap stock-picker whose reconstructed portfolio this site tracks, spent Tuesday focused on two things: a blowout quarter from one of his higher-conviction holdings, and fresh data points feeding his long-running memory-shortage thesis.

## AEHR's qualification-to-volume moment

The day's headline event was Aehr Test Systems (AEHR), which jumped more than 36% following its earnings report. Aehr makes burn-in and stress-test equipment — machines that run chips hard under heat and voltage to weed out early failures — for AI processors, silicon carbide, and silicon photonics. That places it squarely on one of Serenity's favorite kinds of bets: a small "chokepoint" supplier that the larger AI names quietly depend on.

His whole case for the stock has rested on a transition from years of low-volume qualification work into high-volume manufacturing (HVM) as multiple customers finish testing its systems. Tuesday's numbers, as he read them, validated that arc. Management guided fiscal 2027 revenue to roughly $130–$150 million — implying growth on the order of 160–200% — while flagging room to go higher, and reported quarterly bookings of about $60.7 million against an effective backlog north of $100 million. Serenity highlighted a widening customer list: a lead AI-processor wafer-level burn-in customer said to be ramping hard, additional AI-processor firms evaluating the technology, a completed benchmark with a global NAND leader now weighing a development deal for HBM/NAND, and a silicon-photonics customer already in production.

His takeaway was characteristic: with qualification-to-HVM equipment names, the market tends to price the ramp that is coming rather than the trailing quarter, so he suspects the official guide understates 2027 if these customers convert. He did note the obvious risk — the stock has swung violently in a single month — a reminder to know what you own.

Notably, he speculated that the unnamed NAND partner could be Sandisk (SNDK), one of his core memory longs, given prior discussion of high-bandwidth flash. Sandisk is a NAND flash maker he treats as a pricing-power bottleneck and a template for his "supercycle" idea — the thesis that AI-driven demand is outstripping memory supply, forcing repeated price hikes with no relief expected until 2028.

## A fresh NAND price signal

Later, Serenity flagged a Trendforce forecast that SLC NAND — a simpler, more durable flash type — could rise 120–170% in the second half of 2026. That squares neatly with his structural memory view. He noted that Micron (MU), his roughly 10%-weighted centerpiece long, holds about a fifth of that market, but argued the sharpest leverage sits with smaller players whose size better fits the niche's total market. Micron, a US-based DRAM/HBM/NAND maker he calls a "Made in America" moat against Korean rivals, is simply too large to move much on SLC alone.

Instead he pointed to Taiwan-listed specialists Winbond (2344), at roughly 15% share, and Macronix (2337), near 11%, plus SkyHigh Memory via Puya, as cleaner beneficiaries of the SLC spike. He did not disclose positions in the smaller names — the mention read as watchlist mapping rather than a buy call, consistent with his habit of hunting the most concentrated exposure to a moving price.

*This is derived commentary, not investment advice.*