In September 2026, reports said OpenAI and Samsung Electronics were working together on research and production of next-generation AI chips. The account attributed the description to Harrison Kim, OpenAI’s general manager in Korea, who characterized the work as an area of progress between the companies.
What the reported chip work covers
Kim described joint research and production on next-generation chips. Research and production are both part of the reported cooperation, placing Samsung Electronics alongside OpenAI in the companies’ work on future AI hardware.
The development concerns chip research and production, rather than a publicly named product launch. It extends the picture of OpenAI’s custom-chip effort, which includes its own inference accelerator, Jalapeño.
Where Jalapeño fits
Jalapeño is OpenAI’s inference accelerator, designed for the stage when an AI model generates responses. For the first Jalapeño chip, Broadcom was described as the development partner and TSMC as the planned manufacturer. Those roles belong to that first-chip account.
OpenAI’s custom accelerator effort sits alongside its broader use of Nvidia hardware. The reported Samsung work adds a next-generation chip collaboration to that mix; it does not announce a change to Nvidia’s role.
The separate 2025 Samsung partnership
OpenAI and Samsung announced a letter of intent on October 1, 2025, for cooperation on advanced memory, data centers and enterprise AI services, among other areas. That earlier agreement gives the 2026 chip cooperation a wider infrastructure context.
Samsung Electronics was named as a strategic memory partner. Samsung SDS was identified as a potential partner for data-center development and enterprise services, while Samsung C&T and Samsung Heavy Industries were to explore floating data-center infrastructure.
How the work fits OpenAI’s chip strategy
The reported collaboration puts Samsung Electronics in the story of OpenAI’s next-generation chip research and production. It sits alongside the custom-inference effort represented by Jalapeño and the broader accelerator strategy that includes Nvidia hardware.
