Doosan expands CCL production in China and South Korea to meet AI demand

Doosan's $716M CCL expansion is not a capacity play—it is a strategic lock on the AI hardware stack's most overlooked bottleneck. As 224G SerDes and co-packaged optics push dielectric loss toward material physics limits, copper clad laminate is transitioning from commodity substrate to strategic input. The dual China-Korea footprint is a geopolitical hedge disguised as supply chain optimization: Korean lines serve global AI accelerator customers while Chinese lines anchor the domestic substitution pipeline for Huawei Ascend and Cambricon. This is supply-chain sovereignty embedded in capex, not mere diversification. Competitively, Panasonic's Megtron and Rogers' high-frequency laminates face margin compression as Doosan undercuts on localized delivery. Expect a 12-to-24-month repricing cycle where CCL shifts from spot-market pricing to long-term contracted volumes with prepayment terms—mirroring how HBM became a bargaining chip in memory. The real signal: AI infrastructure's bottleneck is migrating from silicon to substrates. Whoever controls low-loss resin and ultra-thin copper foil capacity controls the BOM line item no AI server can skip.