發生了什麼事
Hyundai Engineering was awarded the first‑phase construction contract for the Gunsan SGC AI Data Center, a 60 MW facility on a 115,702‑square‑meter site in Gunsen, North Jeolla Province. The contract, valued at roughly ₩870 billion ($641 million), was announced on 28 May 2026. Hyundai will handle architectural, civil, mechanical, landscaping, and telecommunications work, while its MEP scope includes cooling towers, chillers, switchgear, and other power‑and‑cooling equipment. kt cloud will supply UPS, batteries, and automated control systems. The center will deploy Nvidia’s next‑generation “Vera Rubin” AI chip, and Vertiv’s modular skid solutions will be used for pre‑fabricated power rooms and cooling equipment. The project is the first phase of a planned 300 MW AI‑data‑center cluster, with Hyundai expecting priority negotiation rights for the remaining 240 MW.
On 28 May 2026, Hyundai Engineering announced that it had been selected by SGC Group to construct the first phase of the Gunsan SGC AI Data Center. The project covers a 115,702‑square‑meter site in Bieungdo‑dong, Gunsan City, and will deliver a 60 MW AI‑focused facility. Construction is slated to begin in October 2026.
Hyundai’s responsibilities span architectural, civil, mechanical, landscaping, and telecommunications work. Its mechanical, electrical, and plumbing (MEP) scope includes installing cooling towers, chillers, switchgear, and other essential power‑and‑cooling equipment. kt cloud will provide uninterruptible power supplies (UPS), battery systems, and automated control systems.
The data center will host Nvidia’s next‑generation AI processor, dubbed “Vera Rubin,” which is expected to increase power density and heat output per server rack. To accommodate these demands, the project will employ Vertiv’s modular skid solutions, pre‑fabricating power rooms and cooling equipment off‑site before on‑site assembly.
為什麼這很重要
The contract signals Hyundai Engineering’s rapid entry into the AI‑infrastructure market, a sector traditionally dominated by specialist data‑center builders. By leveraging its experience in large‑scale power and mechanical construction—such as EV battery plants and shipyards—Hyundai aims to meet the heightened power‑density and cooling demands of next‑generation AI chips like Nvidia’s Vera Rubin. The deal also illustrates South Korea’s broader push to expand AI‑ capacity, aligning with national strategies to attract AI workloads and reduce reliance on foreign data‑center services. If Hyundai secures the subsequent phases, its AI‑related order backlog could reach about ₩1.9 trillion ($1.4 billion) this year, underscoring the growing commercial appetite for high‑performance AI compute infrastructure.
Hyundai Engineering’s entry into AI‑data‑center construction reflects a strategic diversification beyond its traditional heavy‑industry portfolio. The company’s claim that this is its first major AI‑data‑center order since declaring the sector a growth business in March 2026 highlights the rapid acceleration of its market positioning.
The contract’s scale—₩870 billion—places Hyundai among the few Korean firms capable of delivering large‑scale AI‑infrastructure projects, a capability that could attract further domestic and foreign AI‑ workloads seeking proximity to South Korean renewable energy sources.
By integrating renewable‑energy concepts such as solar, small modular reactors (SMRs), and hydrogen into its data‑center power strategy, Hyundai aims to address the carbon‑reduction pressures facing the global data‑center industry, potentially setting a for sustainable AI‑ facilities.
互動機制：它實際上是如何運作的
以互動方式探索這項發展背後的基礎技術。
In AI, what are a model's "parameters"?
接下來看什麼
Future developments include Hyundai’s ability to win the remaining 240 MW phases, the timeline and cost performance of the modular construction approach, and how the center’s power mix integrates renewable sources or emerging technologies such as SMRs and hydrogen. Observers should also monitor whether other Korean construction firms follow Hyundai’s model, potentially reshaping the domestic AI‑data‑center supply chain.
Whether Hyundai secures the priority negotiation rights for the remaining 240 MW phases and converts them into signed contracts, which would significantly expand its AI‑data‑center order backlog.
The effectiveness of the modular skid construction method in meeting project timelines and cost targets, especially given the tight power‑density requirements of Nvidia’s Vera Rubin chips.
The extent to which Hyundai can integrate renewable energy sources or emerging power technologies (e.g., SMRs, hydrogen) into the data‑center’s supply chain, influencing broader industry standards for sustainable AI .