Nvidia, SK Telecom plan gigawatt-scale AI cloud in Korea

SK Telecom’s gigawatt-scale AI cloud with NVIDIA marks a strategic pivot: telcos are no longer pipes but sovereign AI infrastructure operators. This move will catalyze Korea’s local stack—especially HBM3E adoption, advanced packaging, and liquid cooling—while exposing supply chain fragility under tightening U.S.-ROK export controls, likely inflating per-petaflop costs by 15–20%. Competitors like NTT and Deutsche Telekom may counter by deepening ties with AMD or custom ASICs to reduce CUDA lock-in. Within 18 months, national telco-AI alliances will proliferate across Japan, Germany, and the Gulf, but the real bottleneck won’t be power—it’ll be high-density interconnects and green energy quotas. That shift redefines entry barriers, sidelining legacy data center players unable to integrate compute, energy, and policy.