"Beyond HBM, Expanding AI Infrastructure"... SK hynix and Nvidia to Jointly Develop Memory for AI Factories - 아시아경제

SK Hynix and NVIDIA have announced a joint effort to develop next-generation memory technologies for global AI factories, deepening their long-term partnership. Beyond enhancing memory performance, th... Read original →

The SK hynix–NVIDIA alliance signals a strategic pivot from raw memory performance to intelligent co-design. Technically, their Omniverse-powered digital twins will disrupt legacy EDA/TCAD workflows, pressuring Synopsys and Cadence to embed AI-native capabilities. Integrating CUDA-X into 3nm EUV process optimization could dismantle the traditional memory-compute separation paradigm. Geopolitically, this U.S.-ROK tech entente raises IP access barriers for non-aligned players, especially mainland Chinese foundries. Competitors like Samsung and Micron will likely accelerate custom AI memory deals with AMD and Intel, possibly rallying around CXL to counter HBM dominance. Within 18 months, this partnership will catalyze an 'AI Factory-as-a-Service' model, shifting semiconductor manufacturing toward software-defined control—where mastery of AI-driven physical simulation dictates next-gen capacity pricing.