SAN FRANCISCO, July 26 — South Korean and American technology giants unveiled a sweeping $950 billion semiconductor partnership on Thursday, marking one of the largest industry alliances ever forged, as global demand for advanced chips outpaces supply and reshapes the geopolitics of artificial intelligence.
The agreements, announced during a landmark AI summit hosted by South Korean President Lee Jae Myung in San Francisco, bind the two countries’ technology sectors at a scale that analysts described as unprecedented. The gathering brought together the leaders of Samsung Electronics, SK Group, Hyundai Motor Group, and Naver alongside CEOs from Nvidia, OpenAI, Anthropic, and Broadcom.
Samsung Electronics signed a memorandum of understanding with Broadcom covering up to $200 billion across memory chips, foundry services, and advanced semiconductor packaging over the next five years. The pact positions Samsung to supply Broadcom with next-generation memory semiconductors while opening its fabrication capacity for the production of AI accelerators — customized chips that cloud providers use to power their data centers.
SK Group, the conglomerate that controls memory leader SK Hynix, announced agreements valued at $750 billion, including a landmark initiative with Nvidia exceeding $500 billion. The Nvidia partnership spans the construction of large-scale AI data centers, long-term supply agreements for next-generation memory chips, and joint development of high-bandwidth memory — the ultra-fast chips essential for training the largest AI models.
“Today we have a vision for the future of AI, and we propose global cooperation to make it happen,” President Lee told the summit, as he outlined the “San Francisco AI Declaration” — a framework for deepening technological cooperation between the two nations.
The scale of the announcements reflects the intensifying race to secure semiconductor supply chains. With AI model training consuming exponentially more computing power, the bottleneck has shifted from software to hardware: there simply are not enough advanced chips to meet demand. South Korea, home to the world’s two largest memory chip makers, has become an indispensable partner for any company building AI infrastructure.
SK Telecom, a unit of SK Group, confirmed plans to build a two-gigawatt data center powered by Nvidia’s next-generation Vera Rubin chips and SK Hynix’s HBM4 memory. The first facility is expected to come online in 2027. Nvidia CEO Jensen Huang said the project would help meet what he called “insatiable global compute demand.”
Anthropic separately signed supply agreements with both Samsung and SK Hynix, and a strategic partnership with Samsung SDS to train AI engineers and explore new AI markets. OpenAI CEO Sam Altman, who also attended, did not announce specific deals but called the summit “a defining moment for AI supply chain cooperation.”
The agreements come weeks after South Korea’s government announced an $880 billion national investment plan to cement the country’s leadership in artificial intelligence, backed by the same conglomerates that gathered in San Francisco. Thursday’s deals turn that ambition into concrete commitments from American partners who need Korean chips to build their AI future.
Industry analysts noted that the partnerships serve a dual purpose: they lock in supply for U.S. firms facing a global chip shortage, while giving Korean manufacturers long-term revenue visibility at a scale that justifies massive capital expenditure. Samsung and SK Hynix have been investing tens of billions of dollars annually to expand fabrication capacity, and these multi-year orders help de-risk those outlays.
The geopolitical dimension is equally significant. The alliances deepen a technology axis between Seoul and Silicon Valley at a moment when governments around the world are racing to secure domestic chip production. The U.S. CHIPS Act has poured subsidies into American semiconductor manufacturing, but the San Francisco summit demonstrated that partnership — not self-sufficiency — remains the operative model for the most advanced chips.
Hyundai Motor Group also joined the collaboration, announcing a joint “robot reference platform” with Nvidia to standardize the development and testing of AI-powered robots. The automaker further revealed plans for an autonomous vehicle foundry in San Francisco, where Alphabet’s Waymo already operates a commercial driverless taxi service.
Naver, South Korea’s largest internet company, signed a strategic investment agreement with Nvidia and, through an infrastructure contract with global investment firm Brookfield, committed to building a $10 billion AI factory for global markets.
Not all chips stocks celebrated the news on Friday. SK Hynix’s American depositary receipts fell more than eight percent in a broad tech selloff, though analysts characterized the decline as a sector-wide rotation rather than a judgment on the deals themselves. The company’s high-bandwidth memory production capacity remains fully booked through the end of the year.
The San Francisco summit underscored a simple reality of the AI era: whoever controls the chips controls the future. On Thursday, South Korea and the United States bet $950 billion that their future is better built together.