SK Builds AI Data Center in Japan with NVIDIA
SK Group Chairman Chey Tae-won announced plans to build Japan's first overseas AI Factory targeting 2028-2029, in remarks at the Nikkei Forum in Tokyo on June 9, as reported by MK and Chosun. The announcement follows a June 7-8 series of Seoul meetings where Chey and NVIDIA CEO Jensen Huang elevated the SK-NVIDIA alliance to the group level. NVIDIA and SK Telecom confirmed a Korea AI Factory coming online in 2027 using the NVIDIA DSX platform, scaling to gigawatt-class capacity over time. MK and Chosun report the Japan site would combine SK hynix high-bandwidth memory with NVIDIA GPUs for large-scale AI training and inference, and requires gigawatt-scale power and large land areas; site selection and local partner discussions are ongoing. The Korea Times notes the Japan move comes amid South Korean government pressure for domestic investment.
What happened
SK Group Chairman Chey Tae-won disclosed plans to build an AI-dedicated data center, branded an "AI Factory," in Japan, targeting 2028-2029. MK and Chosun report the statement came from Chey's interview with the Nikkei and remarks at the Nikkei Forum's Korea-Japan Special Session in Tokyo on June 9. The Korea Times describes the Japan site as SK's first overseas AI Factory, and notes the announcement came amid South Korean government pressure for domestic investment. The Japan AI Factory remains stated intent from a forum speech - site selection, local partners, and power agreements have not been finalized.
Korea AI Factory - confirmed deal
The Japan announcement builds on a confirmed Korea deal. NVIDIA and SK Telecom announced on June 7 (NVIDIA newsroom) that SK Telecom will build a gigawatt-scale AI Cloud in Korea using the NVIDIA DSX platform, with the first AI factory coming online in 2027. On June 8, Chey and Jensen Huang met in Seoul and elevated the SK-NVIDIA relationship from memory supply to full SK Group level, per The Asia Business Daily. "We will work with NVIDIA to tackle GPU, memory and energy challenges and become a leading AI factory player shaping Asia's AI ecosystem," Chey said, per the NVIDIA press release. NVIDIA and SK Group also announced joint research on next-generation AI factory architecture covering accelerated computing, memory technologies, and data center operations.
Technical structure
MK and Chosun describe the Japan AI Factory as an AI training and inference facility combining SK hynix high-bandwidth memory (HBM) with NVIDIA GPUs to improve power efficiency. Site requirements include gigawatt-class power supply and large land areas. The Korea AI Factory uses the NVIDIA DSX full-stack architecture - hardware, software, and operations - targeting lowest token cost at maximum energy efficiency per megawatt.
Korea-Japan strategic framing
At the Nikkei Forum, per The Elec (June 10), Chey argued Korea and Japan could generate $1 trillion in economic synergies and named joint AI factory projects as a priority. He called for a permanent bilateral platform to institutionalize cooperation. SK Group's Japan operations head Park Sang-kyu said: "We will work to materialize Korea-Japan economic cooperation following the launch of the first Korea-Japan Special Session this year."
What to watch
The Japan AI Factory has not advanced to site selection, signed power agreements, or named local partners. Watch for: Japanese partner announcements, grid capacity commitments, and formal deal documentation. The Korea 2027 deployment - first factory at 55 MW scaling to 200 MW by 2028 before reaching gigawatt scale - is the operational precedent.
Scoring Rationale
SK Group's Japan AI Factory announcement is a notable regional infrastructure strategy signal from a major conglomerate, backed by a confirmed Korea 2027 deal with NVIDIA. The Japan-specific 2028-2029 timeline is stated intent from a forum speech rather than a signed contract, which caps the score below a fully confirmed mega-deal. Solid notable-tier coverage for AI/DS practitioners tracking Asian AI infrastructure deployment.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
