How South Korea Became OpenAI’s Unlikely Stronghold
From paying subscribers to chip production: South Korea's multiplying role in OpenAI's global AI push

South Korea has become one of OpenAI’s largest markets outside the United States, and its role in the company’s plans is growing at the same time that demand for computing hardware is rising.
OpenAI said the number of corporate users of ChatGPT Enterprise at South Korean companies and institutions was 28 times higher at the end of August than a year earlier. The country also had the largest number of paying ChatGPT subscribers outside the United States late last year.
AI use has spread well beyond technical teams at some Korean companies. Employees are using it for legal work, sales, finance, operations and marketing. Samsung Electronics is among the biggest deployments, with workers using ChatGPT for research, development, marketing and sales in different parts of the company.
Samsung has not restricted its employees to one AI system. In June, it expanded access in its Device eXperience division, which includes products such as Galaxy smartphones, televisions and home appliances, allowing employees to use ChatGPT along with Google’s Gemini and Anthropic’s Claude.
The semiconductor side of the relationship has been less public.
Harrison Kim, general manager of OpenAI Korea, said at a news conference in Seoul that OpenAI and Samsung had made progress in research and production involving next-generation chips. He said the work with Samsung was one of the areas where the two companies had made the most progress.
Kim did not provide technical specifications or say exactly what Samsung’s role in production would be. Samsung said it could not confirm details about individual customer relationships.
OpenAI is also developing its own hardware. The company announced Jalapeño in June, an inference chip developed with Broadcom, with Taiwan Semiconductor Manufacturing Co. expected to manufacture it.
Running those systems requires more than processors. Large AI models also consume substantial amounts of high-bandwidth memory, an area dominated by South Korean manufacturers.
Samsung Electronics and SK hynix are among the world’s leading producers of DRAM and HBM. Both signed non-binding letters of intent last year to supply memory for Stargate, OpenAI’s planned multibillion-dollar AI infrastructure project.
The letters did not establish a final volume of shipments, and OpenAI has not said how much memory either company will ultimately provide. Kim said relationships with South Korean memory manufacturers remain a priority.
Some of the demand behind that effort is coming from the companies already using OpenAI’s software. As AI moves into more departments, systems are being asked to handle more than writing or simple research.
GPT-6 Astra is designed for computer-use tasks, coding, cybersecurity, scientific research and other professional workflows. Those uses can involve longer sequences of work and substantially more computing than a typical chatbot interaction.
Samsung is dealing with both sides of that expansion. Its employees are adopting OpenAI’s products while the company remains one of the largest semiconductor manufacturers in the world.
SK hynix is supplying the other major part of the Korean memory industry, while the custom-chip effort brings Broadcom and TSMC into the hardware picture.
OpenAI’s business in South Korea is growing as all of this develops around it. More companies are using the software, while the infrastructure behind those systems increasingly depends on manufacturers that are already deeply embedded in the country’s semiconductor industry.











