Samsung and SK Hynix pledge ~$2.07 trillion in AI chip investments

The scale is staggering: Samsung and SK Hynix, the world's two biggest memory chipmakers, together pledged 3,200 trillion won (~$2.07 trillion), encompassing a new 800 trillion won chip cluster in the country's southwest plus accelerated fab construction in the existing Yongin cluster, shortening 7-12 year timelines to bring capacity online sooner. President Lee Jae Myung praised the companies — with a deep bow — for backing the government's semiconductor push, a departure from the memory industry's traditionally restrained, boom-bust-shaped approach to capacity.

The government's goal is to double South Korea's memory chip production capacity within five years, a political win for Lee and a strategic bet that AI-driven memory demand (HBM for GPUs) is durable.

The analyst caveat is prominent: aggressively boosting capex over the next decade raises oversupply risk, and a memory downturn would threaten execution. The tension is between SK's chief seeing a prolonged chip shortage and analysts fearing a painful reckoning if AI spending cools — the same demand-durability question hanging over the entire compute buildout, from NVIDIA to Meta's cloud pivot. Separately, Samsung and Japan's KDDI reported gains from an AI-powered 5G downlink trial using Samsung's RAN Speed Optimizer. Watch memory pricing and HBM demand as the leading indicators.