Sam Altman, CEO of OpenAI, visited South Korea for a brief two-day trip, meeting with executives from Samsung Electronics and SK Group. As key partners in OpenAI's massive AI infrastructure project, Stargate, the discussions likely focused on semiconductor supply and the development of AI data centers.
According to industry sources, Altman had a lunch meeting with SK Group Chairman Chey Tae-won on October 31, followed by a session with over 50 executives from Samsung affiliates at the Samsung Human Resources Development Center in Yongin. Richard Ho, OpenAI's Vice President of Hardware, accompanied Altman during the visit. While specific details of the discussions were not disclosed, it is believed that expanding collaboration on AI infrastructure was a major agenda item.
Altman's visit is significant as a follow-up to the letter of intent (LOI) signed last October for the Stargate collaboration. At that time, Samsung Electronics and SK Hynix agreed to supply memory semiconductors to OpenAI. Notably, SK Hynix proposed plans to establish a production system capable of responding to requests for high-bandwidth memory (HBM) supply of up to 900,000 wafers per month based on DRAM.
One of the key areas of collaboration is the construction of AI data centers. SK Telecom is pursuing a joint project with OpenAI to build an AI data center in the southwestern region, while Samsung SDS is collaborating on data center construction and enterprise AI services. Samsung C&T and Samsung Heavy Industries have also agreed to participate in the joint development of offshore floating data centers, indicating a broadening of cooperation beyond semiconductor supply to encompass overall AI infrastructure.
High-level exchanges between the two sides have continued. Lee Jae-yong, Vice Chairman of Samsung Electronics, visited OpenAI's headquarters in San Francisco in July, and Chey Tae-won met with Altman at an AI summit held locally that same month to discuss collaboration strategies.
Additionally, Samsung is expanding its collaboration with OpenAI to encompass the group's overall AI transformation (AX). On October 31, Altman discussed future management strategies with over 50 Samsung affiliate executives under the theme 'The Next 10 Years of AI: Choices for Businesses.' This follows an initial AI-focused training session held for executives in June, with this additional training occurring four months later. He emphasized that simply using AI will not secure a sustainable competitive advantage, highlighting the importance of enhancing execution and product competitiveness through AI utilization.
Since June, Samsung has been implementing an 'AI transformation' initiative across all areas, including research and development, production, and marketing. Following training for approximately 2,300 executives, the company plans to expand training to all employees by February next year. During the visit, Altman presented Samsung executives with a special small keyboard for the AI coding agent 'Codex,' showcasing the changing work environment due to AI.
As OpenAI expands its investments in AI infrastructure, Samsung Electronics and SK Hynix are also seeing increased opportunities for memory supply. With OpenAI accelerating its development of proprietary AI chips, the supply chain for HBM, currently dominated by Nvidia, is likely to diversify. It is believed that Samsung's HBM is integrated into OpenAI's self-developed AI chip, 'Jalapeño.'
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.