OpenAI Partners with Samsung to Develop Next-Gen Chips, Deepening South Korean Semiconductor Supply Chain Ties
nashnova research
OpenAI is co-developing next-generation AI chips with Samsung Electronics, extending its in-house chip strategy into Korea's semiconductor supply chain — a sign that the AI giant's chip ambitions now stretch well beyond a single product into a multi-partner network.
What are OpenAI and Samsung actually doing together?
OpenAI's Korea general manager Harrison Kim confirmed on Sept 9 in Seoul that the two companies have made "the most progress and received the most recognition" on jointly producing and developing next-generation chips.
This means → Samsung's role is not fabricating an existing chip design but joining at the R&D stage for a future product — a deeper engagement than contract manufacturing.
Neither side disclosed specifics or a mass-production timeline. Samsung did not immediately respond to a request for comment.
How does this fit OpenAI's existing chip lineup?
In June, OpenAI unveiled its first in-house AI chip, Jalapeno — co-designed with Broadcom, manufactured by TSMC, and purpose-built for AI inference.
In plain terms = Broadcom helps draw the blueprint, TSMC builds it, and now Samsung joins on next-gen R&D — three parallel tracks.
This reflects OpenAI's deliberate push to build a diversified chip-partner network: design with Broadcom, fabrication with TSMC, next-gen R&D with Samsung — spreading risk across multiple players.
Why Korea specifically — and is memory a separate track?
Last year, Samsung and SK Hynix both signed letters of intent to supply memory chips for OpenAI's Stargate data-center project.
Harrison Kim said demand for memory chips will keep growing as more advanced AI chips require faster, more complex computation.
This means → Korea's value to OpenAI is twofold: Samsung's chip R&D and manufacturing capability plus the world's strongest memory supply chain (Samsung + SK Hynix). Partnering with Korea is a strategic-level priority.
Is Samsung also an OpenAI software customer?
Harrison Kim called Samsung "one of the largest enterprise deployments of ChatGPT globally," with employees using it across R&D, marketing, and sales.
Samsung announced in June that staff in its Device eXperience division — covering phones, TVs, and appliances — can use ChatGPT, Google Gemini, and Anthropic's Claude internally.
In plain terms = Samsung is both a chip partner and one of OpenAI's biggest enterprise customers — a "supply and buy" relationship that ties the two companies together more tightly.
How important is the Korean market to OpenAI?
By end of August, ChatGPT Enterprise users among Korean companies and institutions had surged roughly 28× year-on-year.
Korea was the market with the most paid ChatGPT subscribers outside the US as of late last year.
OpenAI expects enterprise adoption to accelerate further with the rollout of GPT-6 Astra, its latest model released on Sept 4.
What does the big picture look like?
OpenAI's chip supply chain now runs on three parallel tracks: Broadcom for design collaboration, TSMC for fabrication, Samsung for next-gen R&D.
This means → A diversified supply chain is taking shape, but the exact division of labor and mass-production timelines remain undisclosed — the strategic direction is clear, the execution details are not.
This reflects a broader trend: leading AI companies are no longer content to simply buy off-the-shelf Nvidia chips — they are building their own chip ecosystems from the ground up.
市场有风险，内容仅供研究参考，不构成投资建议。