Samsung to make AMD and Nvidia HBM parts at second Texas fab, report says - The Next Web

Samsung producing HBM for AMD and Nvidia at its second Texas fab marks the execution phase of HBM supply-chain de-risking—not a mere capacity play. The technical moat in HBM resides not in the DRAM die but in TSV stacking and hybrid-bonding packaging. Landing advanced packaging lines in the US directly pressures SK Hynix's roughly 70% share of Nvidia's HBM allocation. Once customers source equivalent specs domestically, procurement logic pivots from "technical exclusivity" to a dual-weight of "geopolitical security + cost." On compliance, CHIPS subsidies reduce capex, yet export controls on hybrid-bonding equipment still constrain Texas fab ramp-up. A deeper structural risk: with production Americanized while IP remains in Korea, Samsung faces a compliance fracture if restrictions escalate. Competitively, SK Hynix will likely lock US packaging partnerships (Amkor, ASE US) within 12 months and leverage HBM4 interface-standard authority to secure Nvidia's next-gen platform. Micron will push for AMD's HBM4 allocation as a differentiation lever. 18–24-month outlook: HBM shifts from a seller's market to a regionalized buyer's market. Nvidia and AMD will simultaneously hold 2–3 domestic suppliers, compressing HBM premiums by 15–25%. Samsung's true strategic intent is not share capture but using HBM as an anchor to pull additional advanced-packaging orders through the Texas fab, building a US-based "memory + packaging" closed loop.