Sam Altman has returned to South Korea without prior announcement, on a visit focused on the ties OpenAI is building with the country's major industrial groups. On Saturday in Yongin, the US company's CEO met with around 50 Samsung executives from various sectors, spanning electronics, finance, and biotechnology. According to an account released by Samsung, the discussions addressed the evolution of artificial intelligence, collaboration opportunities, and the technology's impact on how enterprises make decisions and organize their operations.
The trip comes as OpenAI seeks to turn Stargate, its AI infrastructure expansion program, into a supply and computing capacity network distributed on an international scale. Achieving this requires more than models, software, and funding: it demands facilities, power, servers, and, above all, high-performance memory components. South Korea is one of the most critical nodes in this chain, driven by the role of Samsung Electronics and SK Hynix in manufacturing advanced memory.
No new agreements or financial details tied to the meetings have been made public. For now, the significance of the visit should be interpreted as continuity with the partnerships already announced between OpenAI, Samsung, and SK Group, rather than as a verifiable prelude to a specific contract. However, Altman's personal presence signals that securing access to South Korean industrial capacity remains a direct priority for OpenAI.
Memory has become a strategic resource for AI
In the infrastructure used to train and run generative models, accelerator availability is only part of the equation. Systems must transfer vast amounts of data rapidly between processors and memory. In this context, High Bandwidth Memory, or HBM, is a crucial component: designed to deliver very high bandwidth, it is deployed in computing systems built for AI workloads.
Samsung Electronics and SK Hynix are two central manufacturers in this sector. The partnerships announced in October 2025 under Stargate specifically envisioned an increase in advanced memory chip production and the exploration of AI data centers in South Korea. OpenAI had cited the country's AI readiness and the public and private investments aimed at national competitiveness among the reasons for establishing a more structured presence in the Korean market.
Altman's visit therefore comes at a time when AI software is increasingly dependent on upstream industrial decisions. Frontier model developers must secure reliable long-term supplies; semiconductor and infrastructure manufacturers want to lock in customers capable of sustaining highly capital-intensive investments. These two needs converge, but they do not eliminate the complexity of a supply chain subject to surging demand, production constraints, and heavy geopolitical pressure.
Samsung, SK Group, and an ecosystem broader than chips alone
The meeting with Samsung did not exclusively involve its electronics division. The participation of executives also active in finance and biotechnology suggests a wider discussion on potential enterprise AI applications, although Samsung has not announced any concrete projects or operational outcomes from the meeting. While this is consistent with OpenAI's interest in extending the reach of its tools beyond the tech sector, it does not warrant inferring commercial partnerships in those fields.
Separately, South Korean media reported a lunch between Altman and SK Group chairman Chey Tae-won, attended by executives from SK Hynix and SK Telecom. Local outlets cited cooperation on AI chips, HBM supply, and data center development among the likely topics. However, SK Hynix, contacted by CNBC, did not confirm the reported meeting. It is an important distinction: neither the meeting nor its contents were directly communicated by the group, nor have any formal commitments emerged.
If confirmed, the involvement of SK Telecom would complement SK Hynix's silicon expertise with networking and operational infrastructure capabilities. For OpenAI, building compute capacity requires a diverse group of players: memory makers, data center operators, energy providers, networking firms, and local partners capable of managing permits, sites, and connectivity. Stargate is presented as a global infrastructure initiative; its execution inevitably depends on deals struck country by country.
A relationship cultivated for over a year
This weekend's stop is not the first high-level contact between Altman and the Korean industry. In October 2025, the OpenAI CEO met with South Korean President Lee Jae Myung, Samsung Electronics Executive Chairman Jay Y. Lee, and Chey Tae-won. On that occasion, OpenAI announced partnerships with Samsung and SK Group linked to Stargate.
More recently, in July, Altman hosted Jay Y. Lee at OpenAI's San Francisco headquarters, according to disclosures from Samsung. The progression highlights a relationship managed at the highest levels of the companies involved, at a time when hardware availability can hardly be considered a mere procurement detail. Investments in AI data centers are planned years in advance and require close coordination between projected demand, manufacturing lines, and facility construction.
Altman left South Korea on Saturday, according to local media. Neither OpenAI nor Samsung has outlined a roadmap for the next phase of the partnerships. However, the previously defined goal remains on the table: expanding advanced memory capacity and evaluating AI data centers in the country.
For South Korea, the relationship with OpenAI could strengthen its companies' standing in the global race for compute infrastructure, a field where demand for HBM and high-density systems is reshaping manufacturing priorities and investments. For OpenAI, meanwhile, the potential upside lies in curbing uncertainty around one of the most bottlenecked resources required to support ever-larger models and widely deployed services.
The limitation, for now, is that the visit does not allow anything to be quantified: chip volumes, data center timelines, locations, investments, and commercial terms remain unknown. The next concrete indicator will be any publication of new agreements or industrial plans by the companies. Until then, the episode primarily confirms a direction: OpenAI is trying to make its physical infrastructure for AI more stable, and the Korean groups are among the essential partners in that strategy.
