On July 20, the city of Pohang in North Gyeongsang Province held a groundbreaking ceremony for the 'Global AI Data Center' at Gwangmyeong General Industrial Complex, beginning construction of the largest data center cluster in Korea. A total of KRW 2 trillion in private investmentwill fund a facility capable of housing 20,000 high-performance GPUs, with an initial capacity of 40MW to be built first. According to the city of Pohang, once the second phase—adding roughly 260MW—is complete, the site will become the largest AI data center cluster in the country. The investor is Neo AI Cloud, and Hyundai Engineering & Construction is handling construction.
Why Is a Steel City Building a Data Center?
What stands out is that this cluster is being pursued not in the greater Seoul area or a typical IT hub, but in a traditional manufacturing city. Pohang is home to research institutions including the 4th-generation synchrotron radiation accelerator, a strong manufacturing base, and accumulated industrial data. The investment reflects a judgment that success in the AI race hinges less on the models themselves and more on securing computing infrastructure, industrial data, testbeds, and skilled talent. Industry-specific AI applications—such as real-time process data analysis or predictive maintenance—only become practical options when large-scale computing resources are located close to the field.
Not Just Infrastructure
Alongside the data center, Pohang is also laying the groundwork for real-world validation. The city has begun building a steel industry AI platform and plans to cultivate company-tailored graduate-level talent through POSTECH, which has been designated an 'AI-centered university.' It is also pursuing special zone designation to streamline power and water infrastructure development. What's notable about this project is that infrastructure, data, validation, and talent have all been designed as a single package.
Infrastructure is a necessary condition for AX, but not a sufficient one. As data centers move closer, organizational readiness to translate those resources into actual operational innovation becomes even more critical.
TECH2030's Perspective
Expanding regional AI infrastructure can be an opportunity particularly for small and mid-sized manufacturers, as it creates access to high-performance computing resources without requiring massive capital investment of their own. But seizing that opportunity requires the right sequence. Companies must first understand what process data they are accumulating and in what form, then prioritize which problems to address with AI. The gap between companies that can immediately capitalize once computing resources become available and those that only begin organizing their data at that point comes down to this preparation.