OpenAI's Chief Financial Officer, Sarah Friar, recently held a significant meeting in Seoul, which points toward a major strategic partnership in the works with Korean companies.
This wasn't just a routine visit. Ms. Friar met with leaders from SK Networks and the AI startup Upstage, alongside global investors like ICONIQ Capital. The presence of a top financial executive suggests the talks were focused on two critical areas: securing massive funding and building the physical infrastructure needed for next-generation AI. This is happening because OpenAI has enormous ambitions that require equally enormous resources.
So, why Korea, and why now? The context is key. First, OpenAI is in the middle of a massive fundraising effort for its AI supercomputer project, nicknamed 'Stargate'. The scale is immense, with reports suggesting a total compute spend of around $600 billion by 2030. SoftBank recently committed a new $30 billion, confirming that OpenAI is actively seeking huge capital injections. This Seoul meeting appears to be part of that global fundraising tour.
Second, recent events have highlighted the need for reliable partners. There was brief but notable uncertainty around a major data center expansion with Oracle in the U.S. While Oracle quickly clarified that the project was on track, the incident likely reinforced the importance of diversifying infrastructure partners to avoid bottlenecks. This makes Korea, a country moving quickly on data center construction, an attractive alternative.
Finally, Korea holds a powerful position in the AI supply chain. Companies like SK hynix and Samsung are world leaders in producing HBM (High Bandwidth Memory), a type of memory essential for training large AI models. Any company building AI at scale, including OpenAI, needs a stable supply of HBM. By partnering directly with Korean conglomerates, OpenAI can secure not just capital and data center locations, but also a more direct line to these critical components. In short, this visit was a calculated move to align financing, infrastructure, and supply chain needs all at once.
- Capex: Short for Capital Expenditure, these are funds used by a company to acquire, upgrade, and maintain physical assets like property, buildings, or equipment.
- HBM (High Bandwidth Memory): A high-performance type of computer memory used in high-end graphics cards and network devices. It is crucial for the massive data processing required by AI models.
- Stargate: The reported codename for a massive, multi-billion dollar supercomputer project planned by OpenAI and its partners to power future AI advancements.