OpenAI said Stargate Korea, its planned AI infrastructure initiative with Samsung and SK Group, is moving more slowly than initially expected because rapid technological advances have changed the data-center designs under consideration. Kim Kyung-hoon, OpenAI's Korea country lead, said at a briefing on the 9th marking the first anniversary of OpenAI's entry into South Korea that discussions remain at an early stage. Nearly a year after the strategic partnership was announced last October, investment amounts and groundbreaking schedules have not been confirmed. The broader Stargate project was launched in January last year with SoftBank, Oracle and MGX, with a commitment of $500 billion over four years for U.S. AI infrastructure before expanding overseas, including to South Korea. The Korean initiative involves Samsung Electronics, SK Hynix and SK Telecom, with potential roles for Samsung SDS, Samsung C&T, Samsung Heavy Industries and the newly established SK Hyper. OpenAI is also discussing inference residency (processing AI workloads within South Korea) with multiple local infrastructure operators and said its models could eventually run on South Korean cloud infrastructure. The company confirmed substantive talks with Samsung Electronics on in-house AI compute chips while developing AI accelerators with Broadcom and pursuing its Jalapeño LLM-inference chip, unveiled in June. Kim said South Korean government investment in independent AI models should strengthen the wider ecosystem, supply chain and talent pool. Samsung SDS is the domestic pilot partner for Daybreak, OpenAI's cybersecurity program. ChatGPT advertising in South Korea remains limited to a subset of users and has not produced meaningful early results, while OpenAI is preparing for a possible IPO but has not determined whether it would occur this year or next year.