On a day he described as dominated by macro noise, stock-picker Serenity used the backdrop to restate—not soften—his conviction in the two themes at the center of his book: AI photonics and memory.

The demand signal. Serenity anchored his case to a set of capex forecasts he reads as a telegraph for where AI spending lands. He flagged NVIDIA's (NVDA) projection of roughly $1.3 trillion in hyperscaler capital spending—about $100 billion above one bank estimate—alongside reported multi-hundred-billion-dollar spending plans tied to Anthropic and OpenAI, and even Pentagon interest in funding the buildout. NVIDIA is the AI chip leader Serenity uses as his reference point: rather than own it, he front-runs the upstream suppliers its spending makes scarce. His conclusion: the capex flows hardest into photonics and memory, citing a research estimate that DRAM and NAND alone could absorb 68% of hyperscaler capex in 2027.

Memory. That framing feeds directly into his core longs. Micron (MU) and Sandisk (SNDK) are his primary memory bets on a structural "supercycle"—the thesis that AI demand for DRAM, HBM (high-bandwidth memory stacked beside GPUs) and NAND flash outstrips supply, letting producers hike prices at fat margins. He argued that with leading makers steering capacity toward high-end DDR5 and HBM, legacy-memory suppliers face little new supply—an environment he sees as durable regardless of short-term market fear.

Photonics. On the optical side, Serenity pointed to rising optical content per GPU (tied to NVIDIA's Rubin roadmap and Meta's (META) and Alphabet's (GOOGL) custom TPU chips) and a sharply revised 2027 optical-module forecast. He named the laser and transceiver suppliers he expects to benefit from a light-source shortage: Lumentum (LITE), his large-cap benchmark; Coherent (COHR), the vertically integrated incumbent he holds as a lower-beta core; Applied Optoelectronics (AAOI), his Made-in-America transceiver bet; and Sivers (SIVE), the tiny Swedish maker of InP (indium phosphide) continuous-wave lasers that is his single highest-conviction pick as the chokepoint light source for co-packaged optics.

The most concrete new datapoint came from crowd-sourced channel checks Serenity has been running. He reported that lasers across the 70–200mW power range are seeing rising average selling prices, and that Sivers itself has raised prices—a shift he argues supports gross-margin expansion for laser suppliers in coming quarters.

Around the edges. Serenity noted a report that trader Leopold's "Situational Awareness" activity included options in AMD, Bloom Energy (BE), CoreWeave (CRWV) and others; he attributed the options tilt to elevated implied volatility rather than reading it as a signal. AMD is a name he treats as a demand driver for his upstream picks rather than a holding, while CoreWeave remains his preferred short among AI clouds on debt concerns and Bloom Energy sits only in his broader power-buildout basket.

Lighter moments rounded out the day: Serenity said his "bottleneck thesis" had apparently surfaced in a quant fund's research pipeline, complained about paid engagement networks crowding out authentic voices on the platform, and—half-jokingly—asked followers for the next obscure instrument after discovering a short-dated tanker-freight futures ETF (BWET).

*This is derived commentary, not investment advice.*