OpenAI chief Sam Altman meets about 50 Samsung executives in South Korea over memory supply

SemiconductorsBrief

Summary

OpenAI CEO Sam Altman traveled to South Korea and sat down with about 50 senior Samsung executives as OpenAI looks to grow its AI and memory capacity, Tom's Hardware reported, citing Bloomberg. The headline frames the trip as an effort to secure more RAM for data centers during severe shortages and to discuss future AI data centers. The source does not say whether any supply agreement, volume commitment or pricing was reached.

The visit came shortly after Altman spoke publicly about how scarce global memory output should be divided. On the Chick Pink podcast this week, he said DRAM installed in personal computers frequently goes underused and called that situation "quite terrible". He argued that data centers could get more out of the same memory by pooling computing resources across AI workloads.

Those remarks drew broad criticism, since OpenAI's own services, ChatGPT among them, depend heavily on centralized computing infrastructure. The report says the meeting was held at a Samsung facility, but the available text is cut off before giving further details of the location or the agenda.

Why it matters

Memory has become a gating input for AI data center buildouts, and a direct approach by a major AI developer to a top Korean memory maker shows that buyers are now negotiating supply at the executive level rather than through routine procurement. For operators and hyperscale customers, competition for DRAM allocation can shape server delivery schedules and capacity plans. For global players in Asia-Pacific, the visit underlines how central Korean suppliers are to AI capacity worldwide, and how public debate over who deserves scarce memory may influence where that supply flows.

Expert Lens

A meeting of this size with Samsung leadership suggests OpenAI is trying to lock in memory allocation ahead of rival buyers, which could tighten supply further for smaller cloud and colocation customers. If large AI labs secure priority access, server lead times for other tenants may stretch, pushing ready-for-service dates on new halls even where power and shells are in place. Altman's argument that pooled data center memory beats idle PC memory also likely invites pushback from consumer device makers competing for the same DRAM output. The next signal to watch is whether Samsung or OpenAI discloses a formal supply arrangement with stated volumes or timing.

Sources

This brief was drafted with AI assistance from public sources; every number is attributed to the sources above. See the AI-assisted Disclosure and Methodology.