An investor group and AI developer Voltai plan to use AI to design, build, and run a 3GW data center in South Korea, set to cost up to $35B and open in 2028
Jiyoung Sohn / Wall Street Journal :
Context & Ripple Effects
Voltai's proposal is an early large-scale entry in South Korea's AI-infrastructure push. It sits alongside a reported 250MW project involving Reflection AI and Shinsegae, showing development activity spanning very different project sizes.
Later coverage of a national data-center target of 8.4GW initially and 18.4GW by 2035 places this 3GW plan within a broader domestic capacity buildout. The scale also echoes wider concerns over the limits and returns of the global AI data-center buildout.
First-order effects
- Voltai and its investor group take on a proposed up-to-$35B, 3GW development whose differentiator is using AI across design, construction, and operations; delivery credibility will hinge on executing that integrated model by the stated 2028 opening.
- The project would add a very large prospective source of AI compute capacity in South Korea, while concentrating substantial capital and execution exposure in one development.
Second-order effects
- A 3GW proposal raises the competitive benchmark for South Korean developers and corporate partners, including smaller projects such as the reported 250MW Reflection AI-Shinsegae development.
- The plan intensifies competition for the financing, power access, construction capacity, and hardware supply needed for large AI facilities, making those inputs more consequential to project timing and economics.
Third-order effects
- If projects of this scale advance, AI data centers increasingly become long-duration infrastructure assets rather than incremental enterprise IT deployments, drawing investor-led financing alongside technology operators.
- The buildout could make execution discipline—not announced capacity—the key differentiator: large national targets and individual projects will be tested by whether capital, power, and operational demand arrive on compatible timelines.
The trend: AI infrastructure is shifting toward utility-scale, investor-backed campuses, with national AI ambitions increasingly expressed in gigawatts of planned compute capacity.