OpenAI will design its next-generation AI accelerators with the help of South Korean foundry giant Samsung Electronics. This move could help the ChatGPT maker with memory, process tech, or other key things it needs to succeed.
The announcement came from OpenAI Korea general manager Harrison Kim during a press conference Wednesday. This followed weeks after the company showed its first-generation Jalapeño accelerators besting Nvidia Blackwell GPUs across a slew of inference workloads.
Kim stated that joint production and research on next-generation chips is an area of significant progress and recognition with Samsung Electronics. Following publication, OpenAI clarified that Kim referenced an existing relationship announced last year rather than a new deal.
An OpenAI spokesperson stated that the reference to expanding the collaboration with Samsung was made in the context of the existing relationship. This includes Samsung Electronics adopting ChatGPT and signing a letter of intent last October. The spokesperson added that the company had nothing new to announce and did not announce plans to manufacture chips through Samsung.
OpenAI still contends with the realities of being a fabless chip designer. The vast majority of leading-edge silicon made today comes from TSMC. Securing meaningful capacity requires competing with AMD, Nvidia, and every other chip business and hyperscaler designing custom chips.
TSMC is the biggest wafer fab, but smaller options include Samsung Electronics and Intel Foundry. High-bandwidth memory supply is similarly constrained with just three major manufacturers capable of volume production. Those manufacturers are SK Hynix, Micron, and Samsung Electronics.
OpenAI showcased its Jalapeño AI inference chip at Hot Chips, utilizing HBM4 technology. The next chip will likely use faster HBM4E technology. Securing ample supply during ongoing shortages requires forging relationships early.
Next-generation chips might take advantage of custom base dies to move memory controllers or logic from compute silicon to the HBM stack itself. Fostering a close relationship with Samsung benefits OpenAI even if compute chiplets are manufactured by TSMC.
OpenAI could also fab its next chip on Samsung 2nm process technology to secure an alternative supplier. Samsung is the second largest leading-edge foundry operator in the world. As of 2026, Samsung hosts AI accelerators from designers including Rebellions, Tenstorrent, and Groq.
Rebellions utilizes a multi-die architecture and HBM3E, proving Samsung capable of producing advanced accelerators. Doing so could open the door to larger allocations than available from TSMC alone.
OpenAI maintains its existing relationship with Samsung Electronics while continuing the development of next-generation AI accelerators.
