OpenAI CEO Sam Altman has discreetly visited South Korea, meeting high-profile executives from Samsung Electronics and SK Group. The goal behind the visit? To move closer to his goal of bringing AI chip production in-house by raising billions of dollars for creating a chip fabrication plant network. The pivot comes as a response to Nvidia’s long lead times and the worldwide rising demand for AI chips.
Nonetheless, money for building chip fabrication isn’t all; state-of-the-art technology is also needed. Key to performing AI chip architecture is High Bandwidth Memory (HBM); the advanced version, HBM3, is the prized component in NVIDIA’s H200 GPUs. Samsung and SK have the largest global market share in HBM technology, supplying 90% of the 2023 global supply.
During his visit, Altman met Kye Hyun Kyung (CEO of Samsung Electronics Device Solutions), SK Group Chairman Tae-won Chey, and SK Hynix President Kwak Noh-Jung. They were focused on securing HBM and potentially obtaining investment for his chip fabrication project.
Before this, Altman had met with Taiwan Semiconductor Manufacturing Co (TSMC) earlier in the month. The strategy was to discuss a potential chip manufacturing alliance enabling OpenAI’s reduced dependence on NVIDIA.
This dedication to chip production isn’t isolated to Altman—it’s also shared by US authorities wary of the potential risks involved in the US tech sector’s heavy reliance on Taiwanese made chips. This dependency creates a substantial risk given potential geopolitical issues in China.
With enough funding and successful IP sharing partnerships, OpenAI could alternatively become a leading player in not just AI software, but also AI hardware.
