Samsung lands OpenAI deal to build custom AI chip as memory push grows

- Samsung to produce custom AI chip for OpenAI, expanding partnership beyond TSMC - Samsung advances high-bandwidth memory, shipping first HBM4E samples for AI workloads On September 9, 2026, Reuters reported that Samsung Electronics agreed to manufacture a custom AI inference chip for OpenAI, marking a significant expansion of their existing partnership. According to Reuters, the deal includes joint semiconductor research, actual chip production, and integration of OpenAI’s enterprise software into Samsung’s internal operations. The new chip will succeed OpenAI’s Jalapeno chip and give OpenAI a second manufacturing partner, thereby reducing its previous reliance on Taiwan Semiconductor Manufacturing Company (TSMC). As a result, the collaboration is expected to boost both companies’ capabilities in advanced AI hardware and semiconductor technology. In addition, with its development of high-bandwidth memory (HBM4E), Samsung has become the first supplier to deliver samples of this next-generation memory for AI workloads, according to Reuters. Reuters also reported that integration of OpenAI’s enterprise software at Samsung is a formal part of the partnership. Meanwhile, the deployment of ChatGPT in Korean enterprises continues to rise. According to OpenAI figures reported by Reuters, enterprise usage of ChatGPT across South Korea increased 28-fold by August 2026, which highlights the widespread adoption of AI software solutions in the region.