South Korea Unveils $950 Billion AI Chip Supply Deals with U.S. Tech Giants

South Korean officials disclosed a suite of agreements worth roughly $950 billion that pair the nation’s two leading memory-chip makers—Samsung Electronics and SK Hynix—with U.S. artificial-intelligence leaders. The centerpiece is a partnership between Nvidia and SK Hynix valued at more than $500 billion, covering large-scale AI data-center construction and next-generation high-bandwidth memory (High Bandwidth Memory (HBM)) development. Samsung and Broadcom signed a memorandum of understanding for up to $200 billion in memory-chip supply, foundry services and advanced packaging. The remaining $250 billion of commitments involve broader AI initiatives across the SK Group, Hyundai Motor Group and Naver, as outlined by the South Korean presidential Blue House after an AI summit in San Francisco.

Samsung and SK Hynix to supply chips to Nvidia and other US tech companies in $950 billion deal Express TribuneCNBCReutersSamsung, SK Hynix Prepare Major Chip Supply Deals With US Firms

The announcements come amid a global shortage of high-bandwidth memory needed for AI accelerators. Analysts have noted that demand for AI-focused GPUs and custom chips has outpaced supply, prompting U.S. firms to secure long-term contracts with Asian semiconductor producers. South Korea’s “dynamic AI ecosystem” strategy, articulated by President Lee Jae Myung, seeks to position the country as a reliable supplier of AI infrastructure while expanding domestic AI adoption. Earlier in the year, Seoul unveiled state-backed investment plans exceeding $880 billion to expand chip production, build AI data-center clusters and strengthen physical-AI ecosystems.

Express Tribune

• Total AI-related initiatives announced: $950 billion (South Korean presidential Blue House).

Express Tribune

• SK Group’s aggregate deals: $750 billion, including the $500 billion Nvidia-SK Hynix partnership.

CNBC

• Samsung-Broadcom memorandum: up to $200 billion.

Samsung and SK Hynix to supply chips to Nvidia and other US tech companies in $950 billion deal

• Planned AI data-center capacity: 2 gigawatts of power, slated to come online in 2027.

CNBC

• Expected GPU deployment: “hundreds of thousands” of units, requiring 2 gigawatts of electricity (Nvidia).

CNBC

firms. He emphasized that the nation’s “dynamic AI ecosystem” will expand the global market for both industrial and personal AI applications.

Express Tribune

• “I wouldn't be surprised if, the next time I see (Huang), he'll tell me those figures were too low and say, 'I want more,'” — Group Chairman Chey, SK group chairman

Reuters

• “The expansion will include a co-develop opportunity for us on the next-generation SK Hynix AI memory, and this will help us secure a stable supply of HBM memory,” — Raj Mirpuri, Nvidia enterprise vice president

CNBC

• July 16, 2026 – Nvidia CEO Jensen Huang addressed media in Tokyo following the company’s “Japan AI Ecosystem” reception.

CNBC

• July 24 – South Korea announced the $950 billion AI initiatives at an AI summit in San Francisco.

Express Tribune

• 2027 – SK Telecom’s 2-gigawatt data center, powered by Vera Rubin chips and SK Hynix HBM4 memory, is scheduled to become operational.

Reuters

• Saturday (following the summit) – President Lee is set to attend a Silicon Valley meeting between South Korea’s National Pension Service and U.S. venture-capital firms to promote investment in Korean startups.

Reuters

The announced data centers will begin commissioning in 2027, providing the power and memory capacity required for large-scale AI model training and inference. Subsequent meetings in Silicon Valley aim to translate the announced financing into concrete projects, while South Korean policymakers plan to monitor memory-supply dynamics as AI demand continues to grow.

CNBC