Overview
- OpenAI confirmed at a Seoul press briefing on Sept. 9 that it has expanded joint research and production plans with Samsung Electronics for next‑generation AI processors while withholding technical details and volumes.
- Company officials said Samsung Foundry could manufacture at least some future processors, which would complement existing production at Taiwan’s TSMC and aim to raise supply capacity and resilience.
- Analysts note sourcing from both foundries could reflect very large wafer needs for inference accelerators, though whether exact chips will be double‑sourced or which designs are involved remains unconfirmed.
- OpenAI’s push follows rapid local adoption after the Sept. 4 launch of GPT‑6 Astra, with ChatGPT Enterprise users in South Korea up about 28‑fold year‑over‑year and weekly active users rising roughly 35 percent.
- The move builds on earlier steps: OpenAI unveiled its first custom inference chip, Jalapeño, with Broadcom and TSMC in June and has letters of intent with Samsung and SK hynix to supply high‑bandwidth memory for its Stargate data‑center plans.