South Korea has unveiled nearly $950 billion in new artificial intelligence (AI) initiatives, involving major domestic firms Samsung Electronics and SK Group alongside leading U.S. technology companies. The announcement came during an AI summit held in San Francisco on July 27, where South Korean President Lee Jae Myung outlined the country’s strategic vision to expand its role in the global AI market.
SK Group accounted for the largest share of the agreements, with deals valued at approximately $750 billion. Among these, SK Hynix entered into a partnership with Nvidia worth over $500 billion, focused on developing next-generation memory technologies to advance AI capabilities. Additionally, Samsung Electronics signed a memorandum of understanding with Broadcom, valued at up to $200 billion, aimed at collaboration in memory chips, foundry services, and advanced packaging technologies.
The summit brought together key industry leaders including Nvidia CEO Jensen Huang, OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, and Broadcom CEO Hock Tan. South Korean executives attending included Samsung’s Executive Chairman Jay Y. Lee, SK Group Chairman Chey Tae-won, Hyundai Motor Group Executive Chair Euisun Chung, and Naver founder Lee Hae-jin.
As part of its AI ambitions, South Korea announced the “San Francisco AI Declaration,” promoting enhanced technological cooperation with the United States and positioning the nation as a hub for a “dynamic AI ecosystem.” President Lee emphasized the desire to lead a new era of AI development through these global partnerships.
A centerpiece of the announced initiatives is a collaboration between Nvidia and SK Group to construct a 2-gigawatt data center powered by Nvidia’s Vera Rubin chips and SK Hynix’s HBM4 high-bandwidth memory, planned to be operational by 2027. This facility aims to provide essential computing infrastructure for training advanced AI models and supporting physical AI applications.
SK Chairman Chey Tae-won highlighted the rapidly increasing demand for memory capacity from AI developers. He noted that Nvidia’s projected memory needs for the next five years might be underestimated and expects these requirements to rise further. Chey also reported that Broadcom anticipates its memory demand could exceed current supply and that Anthropic is closely monitoring chip availability as it scales beyond AI model development to boosting computing power.
Chey stated that OpenAI is expected to require substantial computing resources and has encouraged SK Group to supply chips directly to the company. Regarding Anthropic, the SK Group chairman mentioned the startup’s efforts to develop its own computing hardware, although details remain limited. An Anthropic representative confirmed the company’s approach includes exploring diverse chip options for its AI infrastructure.
These agreements reflect growing efforts by U.S. tech firms to secure a stable supply of advanced chips amid increasing competition and supply shortages. South Korea’s semiconductor industry, led by Samsung and SK Hynix, is central to meeting these demands, underscoring the country’s pivotal role in driving the next phase of AI technology development worldwide.