Around a San Francisco AI summit in the United States, the government unveiled a series of cooperation plans between global AI big tech firms including Nvidia, OpenAI and Broadcom and South Korean companies including Samsung Electronics, SK Group, Hyundai Motor Group and Naver. The plans go beyond a transactional relationship of supplying AI memory, upgrading ties to partnership-level cooperation such as building AI infrastructure together.

On July 24 (local time), Nvidia Chief Executive Jensen Huang (젠슨 황) told President Lee Jae-myung (이재명) during a meeting that a true golden age had begun in South Korea. Lee replied that he hoped it was only the beginning. Around the day the two met, Nvidia and SK Group disclosed a partnership worth $500 billion, and Samsung Electronics and Broadcom disclosed a memorandum of understanding worth $200 billion.

The meeting took place at the Fairmont Hotel in San Francisco ahead of the AI summit. According to Nvidia, Lee began the conversation by saying he had seen reports that Huang enjoyed chicken and beer in Seoul, and Huang responded that the enthusiasm he felt across South Korea, from chicken restaurants to pork belly places, was truly surprising. Huang referred to a partnership of more than 25 years, from PC games to the AI revolution, and said the two sides would expand cooperation across AI infrastructure and semiconductors, physical AI and research overall.

The stage then moved to the AI summit held at The Midway in San Francisco. In the 'San Francisco AI Declaration', Lee said South Korea would leap to become a core country in an irreplaceable global AI supply chain. He said it would become a reliable AI semiconductor production base and a supply chain partner based on world-leading memory semiconductor competitiveness and production capability.

As grounds, he cited an AI project investment of 4,700 trillion won announced in June and a plan for a multipolar production base spanning the Seoul metropolitan area and regional areas. Lee likened artificial intelligence to humanity discovering fire anew and presented a plan for trillions of dollars of investment in AI semiconductors, hyperscale data centres and physical AI.

Cooperation among companies took its most concrete shape in memory. Huang said at the summit, "If HBM memory had not been invented in Korea, Nvidia would not have been able to develop the supercomputers that drive AI today."

Nvidia and SK Group signed a letter of intent for a partnership worth more than $500 billion. SK Telecom will use Nvidia's DSX platform to build an AI cloud of up to 2 gigawatts and introduce Nvidia Vera Rubin accelerated computing based on SK hynix HBM4. The first AI factory is targeted to start operating in 2027.

SK hynix will form a partnership for long-term supply and joint development of next-generation AI memory including HBM. The buildout will introduce Nvidia Vera Rubin accelerated computing based on SK hynix HBM4. Chey Tae-won (최태원), chairman of SK Group, said competitiveness in the AI era depends not only on how effectively AI is used but also on how much intelligence can be produced.

Samsung Electronics signed an MOU at the same event with Broadcom for memory and foundry cooperation worth more than $200 billion through 2030. According to Samsung Electronics, it began mass production shipments in February of HBM4 applying 1c DRAM and a 4-nanometre base die, and supplied HBM4E samples in May. Jun Young-hyun (전영현), head of Samsung Electronics' DS division, said that in the AI era, semiconductor solutions integrating memory, logic and advanced packaging closely are important. Charlie Kawwas (찰리 카와스), president of Broadcom's Semiconductor Solutions Group, said close cooperation across the overall semiconductor ecosystem is becoming important as AI infrastructure expands rapidly.

◆Naver raises $10 billion from Nvidia, Brookfield

The scope of cooperation went beyond semiconductors. Nvidia and alternative asset manager Brookfield will invest $10 billion (about 14.6 trillion won) in Naver. Naver board chairman Lee Hae-jin (이해진) disclosed this at the summit and said it would be a catalyst for a big change as Naver makes a new leap. He said not only the capital provided by the two companies, but also their global brands and networks would be a major opportunity to expand Naver's know-how. Lee said Naver had built up technical and operational know-how by directly designing and building data centres, but faced limits in capital and global business networks in expanding to global markets.

The three companies will expand Nvidia DSX AI factories at Naver's Sejong data centre to a scale of 200 megawatts. The plan calls for about 100,000 GPUs based on the Vera Rubin platform, more than tripling the initial plan announced in June. Huang told Lee during the meeting that Naver is South Korea's leading cloud company, and said the partnership would support not only expansion in Korea but also global expansion. Lee said that for the world's diversity to be preserved, multiple AIs for each country, group and individual must be able to emerge and operate.

Hyundai Motor Group unveiled a physical AI strategy centred on a robot reference platform co-developed with Nvidia and will expand the buildout of infrastructure based on 50,000 existing Blackwell GPUs. Drive Hyperion, an autonomous driving development platform, will also be integrated into Hyundai Motor Group's vehicle platform. Nvidia is also pushing to set up AI technology centres (NVAITC) with the mechanical engineering department at KAIST and with Seoul National University, respectively.

◆President Lee Jae-myung: 'Leap to become a core country in an irreplaceable global AI supply chain'

On the demand side, OpenAI expanded points of contact. CEO Sam Altman said that without South Korea, the remarkable AI revolution being experienced now would not have existed, and that bold investment in AI infrastructure is needed. Samsung Electronics decided to introduce ChatGPT and Codex for all employees. Altman is set to meet Samsung Electronics Chairman Jay Y. Lee (이재용) on July 25, following Chey on July 24, to discuss cooperation in HBM, DRAM and advanced foundry. Anthropic CEO Dario Amodei (다리오 아모데이) said South Korea is playing a role across the full stack, from semiconductors to infrastructure and an environment densely packed with outstanding talent.

Contacts outside the official stage also continued. Nvidia said on July 25 that Jay Y. Lee, Hyundai Motor Group Chairman Chung Eui-sun (정의선), and Naver board chairman Lee visited its headquarters in Santa Clara, California on July 24. The content of the meetings and agenda were not disclosed. The industry is seeing expectations that memory and foundry were likely key topics, given Samsung Electronics supplies Nvidia with high-bandwidth memory (HBM) and other high-performance memory. Chung was reported to have discussed cooperation plans in autonomous driving, robots and manufacturing AI.

In closing remarks, Lee said that if South Korea had so far chased advanced countries, it now aims to stand one step ahead and take a path that becomes an example for all. Chey said that every time he has dinner with Huang, he says "give me more chips", and predicted that whatever volume was mentioned that day, Huang would ask for more chips at the next meeting. Many of the disclosed cooperation plans are at the MOU and letter-of-intent stage. SK Telecom's first AI factory is targeted to start operating in 2027, and Samsung Electronics and Broadcom's cooperation period runs through 2030.