Sam Altman met roughly 50 Samsung executives in Yongin, South Korea, on 31 October. Samsung’s account describes a roundtable about how AI will change businesses over the next decade, attended by leaders from its electronics, finance and biotechnology companies. It was part of Samsung’s executive education programme, rather than an announcement of a new chip supply contract.

There is an infrastructure relationship behind the visit. In October 2025, OpenAI announced partnerships with Samsung and SK to expand advanced memory production and explore data centres in Korea. Samsung Electronics and SK hynix planned a production ramp targeting 900,000 DRAM wafer starts a month. That was a stated expansion target, not a report of chips already delivered. The partnerships also covered using OpenAI’s tools inside the companies, which helps explain why the relationship extends beyond semiconductor executives.

Seoul Economic Daily reports that Richard Ho, who leads OpenAI’s in-house chip development, accompanied Altman. His presence prompted speculation about cooperation on AI accelerators. The newspaper also cites OpenAI Korea’s earlier distinction: memory supply was the most concrete area of cooperation, while contract chip manufacturing remained a possibility. The visit alone does not establish that a new manufacturing agreement has been signed.

A separate interview explains why memory allocation is drawing attention. The Verge reported on 30 October that, discussing local AI inference on the Cheeky Pint podcast, Altman argued that scarce memory would be better utilised in cloud servers than in lightly used personal computers. Those remarks concern where computing should happen; they were not a statement of what Samsung agreed at the following day’s meeting.