[비즈한국] Samsung Electronics, SK, Naver, Hyundai Motor, and other Korean companies, along with US firms such as NVIDIA, Broadcom, and Anthropic, have agreed to $950 billion (approximately 1,390 trillion KRW) in AI cooperation following the ‘AI Summit’ held in San Francisco, USA, on the 24th (local time). The focus is on the specific goals behind how US big tech companies are partnering with their Korean counterparts in their own unique ways. Broadcom appears to be utilizing Samsung Electronics as a contract manufacturing base, while NVIDIA and Anthropic are positioning SK and Naver as partners for building AI infrastructure and as demand centers for semiconductors.
Samsung-Broadcom: Long-term cooperation bundling memory and foundry
Samsung Electronics signed a memorandum of understanding (MOU) with Broadcom, a leading fabless company that supplies semiconductors for AI networks and data centers, for $200 billion (approximately 293 trillion KRW) over the next five years, covering advanced memory semiconductor supply and AI chip foundry (semiconductor contract manufacturing) cooperation. The agreement includes the supply of 6th generation HBM4 and 7th generation HBM4E High Bandwidth Memory, as well as the production of Broadcom’s next-generation communication semiconductors using Samsung’s cutting-edge sub-2-nanometer foundry process. Broadcom’s next-generation AI chips are expected to be mass-produced at Samsung Electronics’ Pyeongtaek Campus.
Analysts suggest that US fabless companies, facing rising cost burdens due to TSMC’s price hikes combined with improvements in Samsung’s 2nm yields, have turned to diversification. Taiwanese media even commented that this cooperation “poses a threat to the foundry leader, TSMC.” Samsung Electronics Chairman Lee Jae-yong said in his summit speech, “If the strengths of South Korea and the United States are combined, we will be able to innovate faster, grow bigger, and solve more difficult challenges.”
Dario Amodei, CEO of Anthropic, the developer of the AI model ‘Claude’, announced on the same day that the company had also signed semiconductor supply contracts with Samsung Electronics and SK Hynix. Although the specific items and scale were not disclosed, it appears to be a move to secure HBM and server DRAM necessary for the training and service expansion of Claude. This signals that AI model companies are beginning to enter the AI semiconductor procurement market previously led by NVIDIA and cloud operators, effectively opening a new customer base for domestic semiconductor firms that directly connects to infrastructure investments by AI service companies.
Samsung SDS signed an enterprise AI partnership with Anthropic based on Claude. Lee Joon-hee, CEO of Samsung SDS, stated, “This is a starting point to lead the growth of the domestic AI market by combining Anthropic’s AI technological prowess with Samsung SDS’s experience in digital innovation across industries.” SK Telecom also signed a comprehensive MOU with Anthropic to build a gigawatt (GW)-scale AI data center (AIDC) in Korea.
Solidifying the SK-NVIDIA alliance… Naver collaborates on AI data center construction
Cooperation on AI data center construction also continued. SK Hynix signed a $500 billion Letter of Intent (LOI) to supply next-generation AI memory to NVIDIA long-term and to build an AI factory.
SK Group Chairman Chey Tae-won, describing his series of meetings with leaders including NVIDIA CEO Jensen Huang, said, “The figures announced today are not empty promises, but are based on reality.” SK Telecom is accelerating plans to build a data center with a capacity of up to 2GW with NVIDIA, and has secured priority allocation for the latest ‘Rubin’ graphics processing unit (GPU) systems.
Naver attracted a $1 billion investment from NVIDIA, becoming the company’s third-largest shareholder after the National Pension Service and BlackRock. Combined with up to $9 billion in infrastructure financing support from global alternative asset manager Brookfield, the deal is valued at $10 billion, effectively solving the challenges of GPU supply and infrastructure funding at once.
This is seen as a result of the alignment between NVIDIA’s interests and Naver’s full-stack capabilities, as Naver has been directly operating data centers, GPU clusters, and AI models. Naver’s Global Investment Officer Lee Hae-jin stated, “The capital and global networks of the two companies will be a great opportunity.” Naver CEO Choi Soo-yeon also added, “Starting with this investment, we are highly encouraged to have created an opportunity to quickly enter the global AI infrastructure market through Naver’s technology and infrastructure.” Hyundai Motor Group is expanding its reach into physical AI by jointly building a robot reference platform with NVIDIA.
President Lee Jae-myung said at the summit, “Based on the world’s best memory semiconductor competitiveness and production capabilities, the Republic of Korea will become a reliable AI semiconductor production base and supply chain partner.”