ESMT sees higher 4Q26 ASP as DDR2, DDR3 supply tightens

The DDR2/DDR3 supply squeeze is not a cyclical blip—it is the structural vacuum left after Samsung, SK Hynix, and Micron systematically exited legacy nodes to chase HBM and DDR5. ESMT's aggressive capacity lock-in at foundries and OSATs exposes a deeper shift: as the Big Three redirect wafer starts toward cutting-edge products, industrial, automotive, and embedded IoT applications are forced onto a shrinking supplier base, and pricing power is irreversibly migrating to the supply side. On the technology stack, DDR2/DDR3 manufacturing relies on 0.11–0.18 μm mature nodes. ESMT's capacity reservations directly compete for the same mature-node pool at UMC, Powerchip, and other Taiwan, China foundries, inflating manufacturing costs across the entire legacy ecosystem. At the OSAT level, ASE and Amkor are seeing rising utilization on legacy lines but show zero appetite for capex expansion—a rigid bottleneck. Competitively, Nanya Technology is the most direct rival yet faces identical capacity constraints. The more critical risk: if 2027 pricing stabilizes as ESMT projects, the Big Three could re-enter under a 'niche market' banner, leveraging scale cost advantages to reclaim the segment. ESMT's window is likely 18–24 months. Long-tail verdict: DDR3 will enter quasi-scarcity pricing while DDR2 accelerates toward obsolescence. ESMT's ASP uplift is a passive consequence of supply contraction, not demand pull—its gross margin ceiling is capped by industrial customers' cost tolerance, not market elasticity. This is not a scalable growth track.