Shinsegae Group is pushing ahead with the construction of the largest AI data center (AIDC) in Korea together with U.S. artificial intelligence (AI) company Reflection AI. Based on this initiative, the group intends to foster AI as a new future growth engine.

Shinsegae Group announced on the 17th that it signed a “Strategic Partnership MOU for the Construction of the Korea Sovereign AI Factory” with Reflection AI in San Francisco, United States, on the 16th (local time). The event was attended by Shinsegae Group Chairman Chung Yong-jin, Reflection AI CEO Misha Laskin, and U.S. Secretary of Commerce Howard Lutnick.

Shinsegae and Reflection AI plan to build an AIDC in Korea with a power capacity of 250MW. The project will proceed in stages, with power capacity increased sequentially. The two companies intend to establish a “full‑stack AI factory” capable of providing not only cloud services but also customized AI solutions for users, all based on a large-scale data center (DC).

The 250MW power capacity far exceeds the scale of existing or planned AIDCs in Korea and represents the largest capacity to date. This has been made possible by securing GPUs, the core equipment. Reflection AI has agreed to procure GPUs from NVIDIA for the AIDC it will build together with Shinsegae. In October last year, Reflection AI achieved a corporate valuation of USD 8 billion (approximately KRW 12 trillion) and attracted an investment of USD 2 billion (approximately KRW 3 trillion) from NVIDIA and others.

The partnership between the two companies is also the first case of technological cooperation under the “AI Export Program” launched by the U.S. government last year. The AI Export Program, initiated by the Department of Commerce following an executive order signed by U.S. President Donald Trump in July last year, aims to transfer an AI ecosystem—including center-based AI services along with AIDCs—to other countries.

Shinsegae plans to provide AI cloud services that both Korean government agencies and companies can use with confidence, in line with the government’s vision to strengthen AI competitiveness and build sovereign AI.

Based on the partnership, Shinsegae plans to grow into an AI cloud provider equipped with technological capabilities and data security competence.

Shinsegae will also actively pursue synergies with its existing retail business. It expects that combining its long‑standing retail customer touchpoint infrastructure and data with AI capabilities will enable the realization of “differentiated AI commerce” that offers customers new experiences and benefits. For example, there are expectations for the advancement of “AI agents” that select products optimized for each customer on online malls and handle everything from payment to delivery.

Beyond AI commerce, Shinsegae plans to develop a full‑stack AI solution applicable across its retail operations to improve management efficiency—including enhanced inventory efficiency—and thereby improve profitability. It also intends to build detailed and rapid delivery logistics suited to the coming era of delivery innovation. Through this, the group aims to usher in the “Emart 2.0” era optimized for the future retail industry.

The two companies plan to establish a joint venture (JV) within this year. They will then work closely with relevant institutions and local governments to advance the project.

Chairman Chung Yong-jin stated, “AI will comprehensively transform all areas, including future industries, the economy, and human life, to the extent that future industries will not survive without AI,” adding, “The data center construction collaboration project with Reflection AI will not only serve as a foundation for Shinsegae’s future growth but also contribute to advancing the AI ecosystem across Korean industries.”

Meanwhile, Reflection AI was founded in February 2024 by a group of AI experts, including current CEO Laskin, a former key developer at Google DeepMind, and current CTO Ioannis Antonoglou, one of the key figures behind the development of AlphaGo. In the United States, it is regarded as a leading company in “open‑weight AI model development.” Unlike closed AI models, open‑weight AI models provide infrastructure that allows users to modify model architectures according to their objectives and are advantageous for securing “data sovereignty,” enabling independent management of information.

Popular News