Serenity, the stock-picker who hunts the small upstream suppliers feeding the AI buildout, spent Saturday mostly on a single new name — a Korean optical company — while reaffirming where his real conviction lies.

Supply chains as a trade weapon

He opened by arguing that the real leverage in trade negotiations sits in frontier supply chains, not consumer-goods exports, pointing to chokepoints — critical suppliers with few substitutes — held outside the US. He name-checked ASML, the Dutch monopoly on EUV lithography machines that he treats as the benchmark for semiconductor capital-equipment scarcity; Towa, the Japanese near-monopoly in the compression-molding tools used to package high-bandwidth memory (HBM) and one of his core holdings; and LPKF, the German maker of glass-core substrate etching equipment he calls the "ASML of advanced packaging." His point: tariffing allies that hold these monopolies was self-defeating, but the relationships could instead be weaponized for better deals.

A lighter aside

He also riffed on reports of whey-protein shortages and price hikes, joking that "gym bros" had created their own bottleneck — a nod to Sandisk (SNDK), his core NAND flash holding and his go-to template for a supply-constrained pricing "supercycle."

The main event: a Korean transceiver upstart

Most of the day went to OE Solutions (Seoul-listed 138080), a small Korean optical-transceiver firm he likened to a "Korean AAOI." Applied Optoelectronics (AAOI), one of his core US longs, is a Made-in-America vertically integrated transceiver maker; Serenity frames OE as a sovereign Korean attempt at the same model, but built around scarce EML lasers (the electro-absorption modulated light sources for 800G/1.6T links) rather than the continuous-wave lasers used in co-packaged optics (CPO). He noted OE is now one of only a handful of EML players globally, alongside Coherent (COHR) and Lumentum (LITE), the two large incumbents, plus Japanese suppliers.

His verdict was measured and explicitly cautious. OE has 20-plus years of history, US and Netherlands R&D, finished CPO products (ELSFP modules with high-power CW lasers) sampling in Q3, and low capacity utilization that leaves room to grow revenue before heavy capex. But it has no confirmed customers yet for its AI growth lines, uncertain yields, and a US-institution-unfriendly small-cap Korean listing. He called it early — perhaps more interesting after Q3 sampling — and notably higher-risk than his top pick.

Sivers stays the anchor

That top pick is Sivers (SIVE), the tiny Swedish indium-phosphide CW laser maker he rates his highest-conviction position. He contrasted OE's unproven status against Sivers' design-ins across Ayar Labs, Jabil (JBL) and the foundry GlobalFoundries (GFS), plus its fabless capacity de-risking via Win Semiconductor. Jabil features as a customer validation point — it selected Sivers' lasers for its 1.6T pluggable transceivers — while GFS recurs as an ecosystem node in his CPO supply maps.

In a closing post he pushed back on a "misconception," stressing that Sivers supplies lasers across next-generation architectures — pluggables, scale-out and scale-up CPO, and NPO — not just CPO scale-up alone.

*This is derived commentary, not investment advice.*