SK, Nvidia expand chip alliance into AI infrastructure - The Korea Herald

SK Group and NVIDIA have expanded their long-standing chip partnership beyond memory chips into the artificial intelligence (AI) infrastructure domain, with plans to build what they call 'AI factories... Read original →

SK Group’s expansion with NVIDIA beyond HBM into AI infrastructure signals a strategic fusion of semiconductor manufacturing and AI compute ecosystems. Technically, this accelerates Korea’s vertical integration from memory to full-stack AI systems, spiking demand for HBM4 and advanced packaging like CoWoS, while pressuring thermal management and EDA tools to evolve. On compliance, tighter U.S.-Korea alignment on AI chip export controls may raise barriers for non-allied nations, forcing SK to diversify supply chains despite deep NVIDIA reliance. Competitively, TSMC will likely deepen ties with AMD and Microsoft, while Samsung may fast-track its in-house AI platform. Over the next 12–24 months, ‘national AI factories’ will proliferate in Japan, India, and the Middle East—but only U.S., Korean, and select Taiwan, China players will command full-stack capabilities, cementing a new geopolitical axis centered on high-bandwidth memory and localized AI data centers.