AI servers drive TVS and ESD demand for Yea Shin Tech

The real story isn't a micro-cap OTC listing—it's the structural repricing of discrete protection devices in the AI server stack. Yea Shin's push to lift 6-inch wafer share from 35% to 50%+ is a bet on nonlinear per-board TVS/ESD content growth. In H100/B200-class GPU modules, every buck converter in the PDN demands a TVS clamp; board-density scaling means absolute device counts multiply faster than unit server shipments. 6-inch remains cost-competitive for mature-process discretes—smarter than chasing advanced nodes. On compliance, feeding the international AI server supply chain implies exposure to North American OEMs. The 6-inch node sits below most export-control thresholds, functioning as an inadvertent shield. But it also caps the ceiling: as architectures shift toward chiplets and co-packaged optics, protection requirements migrate toward higher integration. The pure 6-inch window may close within 18–24 months. Competitively, Littelfuse owns automotive TVS, but the data-center segment is fragmented across Vishay, ST, and NXP. Yea Shin enters as a second or third source—price elasticity wins initial volume, but design-in lock-in is the real moat. Once tier-1 OEMs freeze BOMs for next-gen platforms, late entrants get locked out. 12–24 month call: AI server capex cycle remains intact; volume and ASP tailwinds for discrete protection devices persist into mid-2026. The risk is a 6-inch capacity glut triggering a price war—a replay of the 2019 LED-driver IC overcapacity episode.