OpenAI CEO Sam Altman met around 50 senior Samsung Group executives in South Korea on 31 October as demand for memory chips used in artificial intelligence infrastructure continues to put pressure on semiconductor supplies. The meeting took place at Samsung’s Human Resources Development Centre in Yongin, south of Seoul. According to Samsung’s official announcement , discussions covered the development of AI technology, emerging business opportunities and the effect of AI on corporate decision-making. Executives from Samsung’s electronics, financial and biotechnology businesses attended. Although the visit came amid tight supplies of DRAM and high-bandwidth memory (HBM), Samsung did not announce a new memory supply contract or disclose additional purchasing commitments. OpenAI-Samsung talks build on Stargate agreement The OpenAI-Samsung talks follow a strategic partnership established in October 2025, when Samsung Electronics and OpenAI signed a letter of intent covering memory semiconductors and AI infrastructure for the Stargate programme. At the time, Samsung identified projected OpenAI demand of up to 900,000 DRAM wafers per month. That figure represented potential demand rather than a confirmed monthly supply order. The agreement also involved Samsung SDS in data centre development and enterprise AI services, while Samsung C&T and Samsung Heavy Industries agreed to investigate floating data centres. The companies have subsequently broadened their discussions beyond memory. In September, eeNews Europe reported that OpenAI and Samsung were exploring next-generation processor development , raising the possibility of foundry cooperation in addition to their existing memory relationship. No manufacturing contract for OpenAI processors was disclosed at that time. Memory shortages complicate AI infrastructure expansion The pressure on DRAM supplies reflects competition between AI data centres, enterprise servers and conventional electronics manufacturers. AI accelerators require HBM to provide high-bandwidth access to model data, while the servers surrounding those accelerators also consume substantial quantities of conventional DRAM. Samsung has been expanding its advanced memory portfolio to address this demand. In February 2026, the company announced commercial shipments of HBM4 , specifying transfer speeds of 11.7 Gbit/s per pin and capacities from 24 to 36 GB for its 12-layer devices. Altman also met SK Group chairman Chey Tae-won during his South Korean visit, according to Electronic Times . SK hynix is another major supplier of HBM and an existing OpenAI infrastructure partner. Beyond infrastructure procurement, Samsung is expanding its internal use of AI. The company said it had provided AI training to approximately 2,300 executives and intends to extend training to all employees by February 2027. The latest OpenAI-Samsung talks formed part of that programme, rather than a publicly announced expansion of the existing memory supply agreement. The post OpenAI-Samsung talks amid tightening AI memory supply appeared first on eeNews Europe .