This week saw the world’s artificial intelligence landscape change in a major way, with South Korea announcing a massive $950 billion in new AI projects with the world’s largest tech companies, including Samsung Electronics, SK Group and even big American tech giants. The announcements were part of a South Korean president’s “AI summit” in San Francisco that highlighted the global scramble to gain control of the next generation of semiconductors. The agreements underscore the need to accelerate the development of new artificial intelligence applications and resolve the critical shortage of AI chips that are faster and more efficient.
The size of these agreements highlights the importance and increasingly pivotal role played by South Korean semiconductor companies in the global AI ecosystem. This time it was SK Group, with an estimated $750 billion worth of deals. A large part of the totality is SK Hynix’s increased ties with Nvidia, which are said to be worth more than $500 billion by the South Korean presidential Blue House. This partnership is aimed at ensuring Nvidia’s future memory needs and to jointly create high bandwidth memory tailored to AI training, AI agents and physical AI uses. As Nvidia continues to enhance the processor’s capabilities, SK Hynix’s partnership with the company stands as a testament to the growing need for specialized memory solutions that can support the rapidly evolving needs of the AI industry.
Samsung Electronics also contributed significantly to the efforts, signing an MoU with Broadcom worth up to $200 billion across various technologies like memory chips, foundry services, advanced packaging and more. This collaboration puts Samsung in a good position to collaborate closely with Broadcom on developing next-generation AI accelerators using Samsung’s high-bandwidth memory technology. Furthermore, Samsung will give Broadcom access to its sub-2-nanometer foundry process along with advanced packaging solutions that showcase the company’s full-spectrum semiconductor capabilities from design to manufacturing.
Such huge offers come at a moment when the AI sector is undergoing its most radical evolution and expansion ever. AI companies, in a bid to secure chips in high demand, have found themselves in competition with each other, with the technology firms engaging in closer ties with chipmakers. This need for chips and computing infrastructure has consistently far exceeded supply, amplifying this urgency and the strategic significance of South Korea’s semiconductor industry to the global AI landscape.
President Lee Jae Myung made a statement at the summit, outlining South Korea’s vision to advance the world into a new era of artificial intelligence. He highlighted the country’s dedication to making its “dynamic AI ecosystem” accessible to enhance the international market for industrial and personal AI products and services. The president announced the “San Francisco AI Declaration” that details South Korea’s strong ambitions for AI and its platform for cooperation in the technology with the United States. The statement marks a formal pledge to enhance cross-border collaboration and make South Korea a crucial force in the international AI arena.
The summit featured an impressive lineup of technology luminaries, including Nvidia CEO Jensen Huang, OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei and Broadcom CEO Hock Tan. They were accompanied by leaders of South Korea’s major industrial conglomerates like Samsung Electronics executive chairman Jay Lee, the SK Group chairman Chey Tae-won, the Hyundai Motor Group chairman Euisun Chung and Lee Hae-jin, the founder of Naver. The gathering of big names on both sides of the Pacific underscored the interdependence of the global AI industry and the vital collaborations between American innovation and South Korea’s manufacturing prowess.
In regard to the Nvidia partnership, SK Group Chairman Chey Tae-won said that AI customers were asking for a lot more memory than they had originally thought. He mentioned recent talks with Nvidia, Anthropic, OpenAI and Broadcom, indicating that Nvidia’s estimated memory demand over the next five years may soon rise above its current estimates. “If those numbers were too low, then the next time I see (Huang), he’ll tell me ‘I want more,’” Chey said, capturing the dynamism of the demands of AI computing.
Chey also revealed that Broadcom CEO Hock Tan had voiced his worries that there might be an eventual shortage of memory usage for the firm. He also said Anthropic was keeping a close eye on the future availability of chips as it transitions from being an organization that creates AI models to one that creates computing capacity. Chey said OpenAI would need to consume massive amounts of computing power and had asked SK to provide the firm with its own chips for use in ChatGPT. The SK Group chairman also noted that Anthropic was developing its own “computing power,” which could be a reference to its plan to investigate the design of its own chips as previously reported. An Anthropic spokesperson said the startup had always been looking to expand its portfolios of chips and operate with a variety of chips, a diversified approach to computing.
In an interview, Anthropic CEO Dario Amodei said he was unaware of any plans to create chips of its own but did confirm that the company has already inked supply contracts with South Korean chipmakers Samsung Electronics and SK Hynix, but did not detail them. This follows Anthropic’s last funding round in May, which saw it close a $965 billion financing round, with Samsung Electronics, SK Hynix, and Micron as partners whose technologies are essential for supply of memory, storage and logic chips.
