GoodVision AI and ATTO Research have signed an MOU to jointly develop a network of supply chain-focused AI data centers in South Korea, with an initial joint commitment of at least $50 million. The effort targets phased buildouts near Seoul, Busan, and Daegu, pairing data center development and power sourcing with AI inference compute and “AI Factory” deployment capabilities.

The MOU sets a phased capacity plan: an initial build of 0.75 MW of AI Factory compute capacity, followed by a second phase targeting 5.5 MW by 2027. The longer-term goal is 40 MW of combined capacity across multiple sites.

For data center operators, the practical issue being called out here is schedule mismatch. The companies argue that enterprise customers can procure GPU servers in months, while traditional data center builds can take years, and they’re aiming to deliver modular AI data centers on timelines aligned with GPU procurement cycles. That approach can reduce stranded GPU capex and stalled AI deployments, but execution will hinge on site readiness, power availability, and how “modular” translates into real commissioning timelines.

Under the agreement, ATTO Research is responsible for site and power infrastructure resources, power sourcing, network integration, and local development in South Korea. GoodVision AI is contributing AI inference computing expertise, AI Factory deployment capabilities, intelligent compute scheduling, and edge AI network expansion.

Davy Wang, CEO of GoodVision AI, said, “Demand for purpose-built compute capacity is outpacing what traditional data center development can deliver.” Dr. Jae Woong Chung, CEO of ATTO Research, said, “New data center space takes years. GPU servers arrive in months.”

Source: Business Wire