Local Chorus
Local news from local sources, read in your language.
Developing
Verified

KT Cloud starts building 50 MW AI data center in Ansan, due for completion in 2029

🇰🇷 South Korea, Ansan 12:19 Business Business3 Official updated 1 h ago first reported by 뉴시스

In short

KT Cloud said on 22 October that it is starting construction of a 50-megawatt AI data center in Ansan, Gyeonggi Province, targeted for completion in 2029. Yonhap, citing an order disclosure, reported a construction cost of 315 billion won for the contract won by the construction division of CJ Daehan Tongun (CJ대한통운).

Read the full story 2 min read

KT Cloud said on 22 October that it is starting construction of a 50-megawatt AI data center in Ansan, Gyeonggi Province. Chosun Ilbo reported that the facility is being built on land the company owns, and that completion is targeted for 2029. The project is part of the company's medium- to long-term strategy of securing 1 gigawatt of AI data center supply capacity by 2031. [ 1 , 2 , 3 ]

According to Chosun Ilbo and Newsis, the building will have two floors below ground and nine above, with a floor area of about 36,878 square metres and total power receiving capacity of 50 MW. Yonhap gave the floor area as about 36,879 square metres. Yonhap also reported an IT load of 32 MW and described the project as a hyperscale-class data center in the Banwol National Industrial Complex in Ansan. It said KT Cloud plans to use the data center itself. [ 1 , 3 , 4 ]

The power, cooling and structural design is aimed at high-density AI servers that consume a lot of power and give off a lot of heat. All server rooms are designed to support liquid cooling. The company said liquid is used to handle the heat that rises as server computing performance increases. The Ansan site follows the Gasan AI data center that opened last year. [ 1 , 3 ]

The company said the floors and structure are designed to bear heavy loads, in view of the possible installation of high-performance equipment such as Nvidia's next-generation Vera Rubin GPU. This is meant to allow for denser servers or changes in equipment specifications. A free-cooling system is also planned, with cooling coils on the upper exterior walls linked to closed cooling towers, so that cold outside air can be used in transitional seasons and winter. KT Cloud said this would reduce the load on cooling tower fans and chillers. [ 1 , 3 ]

Yonhap reported that the construction division of CJ Daehan Tongun (CJ대한통운) won the construction order. Citing the order disclosure, it gave a construction cost of 315 billion won and said the work would take about 38 months. The division will handle civil, architectural, landscaping and mechanical work, as well as commissioning up to level 3. CJ Daehan Tongun said it will use drone stations, AI-based process management tools and 4D building information modelling on site. [ 4 ]

KT Cloud chief executive Kim Bong-gyun (김봉균) said the Ansan AI data center is a project reflecting power and cooling infrastructure optimised for high-density AI servers and next-generation design technology. He said the company will provide stable and efficient AI infrastructure based on its accumulated experience in building and operating data centers. [ 1 , 3 ]

Why it matters

The project is part of KT Cloud's stated aim of reaching 1 gigawatt of AI data center supply capacity by 2031. Chosun Ilbo and Newsis describe it as a way to expand AI infrastructure capacity in the Seoul metropolitan area as demand for high-performance AI servers grows.

Key facts

  • KT Cloud announced on 22 October that it is starting construction of an AI data center in Ansan, Gyeonggi Province, with 50 MW of total power receiving capacity. [ 1 , 2 , 3 ]
  • The building will have two floors below ground and nine above, with a floor area of about 36,878 square metres according to Chosun Ilbo and Newsis. [ 1 , 3 ]
  • Completion is targeted for 2029. [ 1 , 3 , 4 ]
  • All server rooms are designed to support liquid cooling. [ 1 , 3 ]
  • The project is part of KT Cloud's strategy to secure 1 GW of AI data center capacity by 2031. [ 1 , 3 ]
  • CJ Daehan Tongun's construction division won the building contract; Yonhap reported a construction period of about 38 months and a cost of 315 billion won, citing the order disclosure. [ 4 ]

Confirmed by several sources

  • KT Cloud announced on 22 October that it is starting construction of a 50 MW AI data center in Ansan, with completion targeted for 2029. [ 1 , 3 , 4 ]
  • The building has two floors below ground and nine above, and all server rooms are designed for liquid cooling. [ 1 , 3 , 4 ]
  • The project is part of KT Cloud's plan to reach 1 GW of AI data center capacity by 2031. [ 1 , 3 ]
  • CJ Daehan Tongun's construction division said it won the order to build the data center. [ 4 ]

Still unclear

  • The exact floor area of the building. Chosun Ilbo and Newsis give about 36,878 square metres, while Yonhap gives about 36,879 square metres.
  • The construction cost of 315 billion won and the 38-month construction period. Only Yonhap reports these figures, citing the order disclosure.
  • The IT load of 32 MW and the location in the Banwol National Industrial Complex. Only Yonhap reports these details.

What local media are saying

Business mediaChosun Ilbo and Newsis relayed KT Cloud's announcement, stressing the 50 MW capacity, the 2029 target, liquid cooling in all server rooms, a structure designed for heavy next-generation GPUs and a free-cooling system, and the 1 GW goal for 2031. [ 1 , 2 , 3 ]
Official sourcesYonhap focused on the construction contract won by CJ Daehan Tongun's construction division, giving the cost, the construction period, the 32 MW IT load, the site and the digital tools to be used on site. [ 4 ]

Timeline, local time

  1. Chosun Ilbo published the report on KT Cloud's announcement of the Ansan AI data center construction start. [ 1 , 2 ]
  2. Newsis published its report on the construction start and the 2029 completion target. [ 3 ]
  3. Yonhap reported that CJ Daehan Tongun's construction division had won the construction order. [ 4 ]