GoodVision AI and ATTO Research have signed a memorandum of understanding (MOU) to jointly develop a network of AI data centers (AIDCs) across South Korea. The agreement, signed on April 16, 2026, includes a minimum investment commitment of US$50 million for data center construction, GPU infrastructure, and financing solutions for AI and digital infrastructure assets.
The project will focus on locations near Seoul, Busan, and Daegu. Development will take place in phases. The first phase will deliver 0.75MW of AI Factory compute capacity. A second phase aims to expand capacity to 5.5MW by 2027. The long-term target is 40MW of combined capacity across multiple sites.
The partnership seeks to address a growing challenge in South Korea’s AI sector. Enterprises are able to procure GPU servers within months, while conventional data center projects often take years to complete. This gap has contributed to delays in AI deployment and infrastructure availability.
According to the companies, the planned facilities will use a modular approach designed to align data center delivery timelines with GPU procurement cycles. The objective is to provide enterprises with faster access to AI computing resources.
Under the agreement, ATTO Research will provide site and power infrastructure resources, support for power sourcing, network integration expertise, and local development experience. GoodVision AI will contribute AI inference computing capabilities, AI Factory deployment expertise, intelligent compute scheduling technologies, and edge AI network expansion capabilities.
Davy Wang, CEO of GoodVision AI, said demand for AI-focused compute infrastructure in South Korea is growing faster than traditional data center development can accommodate. He added that the partnership provides a foundation for multi-site expansion through 2027.
Dr. Jae Woong Chung, CEO of ATTO Research, Professor at KAIST, and Head of the Task Force at the Presidential Council on National Artificial Intelligence Strategy, said South Korean enterprises are increasingly ready to deploy AI at scale, but infrastructure development has not kept pace. He said the partnership aims to help close that gap through faster deployment of AI-ready facilities.
