As global competition in artificial intelligence (AI) intensifies, AI has become a crucial driving force for nations. 'Sovereign AI' has emerged as a vital requirement for safeguarding national sovereignty, economic independence, and unique cultural assets. The South Korean government is mobilizing national capabilities to elevate the country to the status of one of the 'Global AI Three Powers (G3),' following the United States and China. Achieving this goal requires self-sufficiency in five key areas of sovereign AI: generative AI, physical AI, AI data centers, AI semiconductor design and production, and AI-demand industries. Urban planning responses are essential to implement these strategies.
First, there is an urgent need to establish an 'AI-exclusive business district' that incorporates 'AI self-sufficiency foundations and generative AI ecosystems' into urban planning. The essence of sovereign AI lies in securing a unique foundation model that understands the Korean language and our cultural context while establishing data sovereignty. To foster synergy among domestic generative AI anchor companies and innovative startups, aging buildings in city centers should be remodeled for AI use, and high-density mixed developments near transit hubs and public lands (such as vehicle depots) should be pursued to create large-scale AI business complexes.
Additionally, a physical AI ecosystem hub, such as an 'AI robotics and manufacturing autonomy special zone,' is necessary. Pittsburgh, USA, has transformed aging industrial sites into physical AI hubs, with the area around Carnegie Mellon University operating as a high-density mixed R&D zone that accommodates research institutes, startups, and testbed factories. Similarly, we should develop zones for robots, autonomous vehicles, and drones in our cities, designating the entire city as a 'physical AI living lab' and incorporating regulatory sandboxes, real-life data sharing, and dedicated buffer zones and infrastructure for physical AI into urban planning.
The establishment of a 'semiconductor-AI convergence cluster' to support the production and localization of domestic AI semiconductors (GPU/NPU) is also crucial. Relying on imported GPUs from companies like NVIDIA makes self-sufficiency impossible. We must expedite the development of a domestic NPU ecosystem designed by local fabless companies. Like Taiwan's TSMC in Hsinchu Science Park, we need a cluster in the semiconductor belt, such as Yongin, that integrates AI semiconductor design and foundry production. An open innovation platform (OIP) should be expanded as an urban planning facility to allow small fabless companies to share expensive overseas intellectual property and electronic design automation (EDA) tools at lower costs. Furthermore, we need to create an infrastructure ecosystem that connects AI semiconductor prototype production (MPW) to public foundry production.
Expanding the localization of 'learning and inference AI data centers' is essential. Currently, foreign data centers dominate the domestic market, making data sovereignty unattainable. Integrating a 'green distributed energy infrastructure plan' is also necessary. Since ultra-large data center infrastructure consumes significant power and requires heat management, these facilities should be located in areas with stable power supplies, while urban areas must mandate renewable energy distributed power plans in their urban planning. We need to attract edge computing facilities based on domestic NPU to urban knowledge industry centers, mixed-use complexes, and large buildings, creating spaces where buildings themselves serve as AI assistants.
Finally, we must realize the self-sufficiency of competitive industries through 'urban living labs' that connect AI R&D with demand companies. To nurture sovereign AI demand industries and companies, the city itself needs to serve as a testbed. Designating mixed-use complexes (such as Magok and Jamsil Sports Complex) as 'AI living lab districts' will allow for the sharing of living data with companies, creating a virtuous cycle where businesses can conduct technology experiments and generate revenue on-site. Additionally, we should develop 'integrated work-live-play villages' to cultivate AI talent and attract overseas professionals. By building communities where employees can work, live, and enjoy leisure within walking distance, the effectiveness of these initiatives will be enhanced.
For South Korea to ascend to the status of AI G3, the five key elements of sovereign AI must be explicitly reflected in urban planning. Only by redefining urban spaces as innovative ecosystems centered around AI can we achieve technological sovereignty. It is time for the government and urban planners to collaborate and create bold spatial innovations.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.