[Gangnam Perspective] Government AI Infrastructure Project Still Lacks Momentum

The new administration moved early to launch an artificial intelligence (AI) infrastructure drive, yet the project still appears to be stuck in low gear. This refers to the National AI Computing Center initiative, which has taken its first step but remains at a standstill. For individual companies to build advanced AI services, training and experimentation using expensive equipment such as graphics processing units (GPUs) are essential. Large corporations can secure such infrastructure on their own and establish a base for research and deployment, but for most companies, including capital‐constrained startups, competing on infrastructure is virtually impossible.
The National AI Computing Center, together with the AI data center, is expected to become one of the key strategic facilities that will shape the country’s future. The concept is this: the state builds large‐scale AI computing infrastructure that is too costly for the private sector to handle alone, and then allows companies and research institutes to rent it at low cost and with stable access as a national AI infrastructure. In the past, companies mainly needed cloud support; now, the focus is on providing GPU‐based AI computing power via the cloud. Through the National AI Computing Center, the government plans to secure more than 15,000 GPUs by 2028 and expand the system to as many as 30,000 GPUs by 2030. It is a large‐scale project that will inject over 2 trillion won by combining government investment with policy finance loans to the private sector. Beyond simply installing equipment, the blueprint is to build a "public AI infrastructure hub" where domestic AI companies and researchers can experiment with large‐scale models and commercialize them.
The project has already gone through growing pains, including two failed bidding rounds. At a government briefing last year, more than 100 companies took part, yet not a single bidder came forward for the first tender, which was consequently scrapped. The reason was clear: the design of the special purpose vehicle (SPV) and the equity investment structure was heavily skewed against private participants. Critics pointed out that while the private sector would shoulder the burden of massive upfront investment and operational risk, profitability and management autonomy would be tightly constrained. When the second tender also failed with no change in conditions, the government revised the unfavorable equity structure for private firms and, on its third attempt, finally secured a participant.
Even so, concerns persist that the project is still being delayed. Samsung SDS formed a consortium with Samsung Electronics, KT, NAVER Cloud, Kakao and others and became the sole project operator, but the government’s financial screening, which was scheduled for the end of last year, has been pushed back. The Samsung SDS consortium needs to move ahead with establishing a Special Purpose Company (SPC) and breaking ground at the site, yet the SPC is reportedly still only at the level of having a Task Force (TF) or dedicated team in place.
Although this is a public‐private joint project, the Samsung SDS consortium is increasingly anxious as delays drag on. Recently, Samsung SDS officials personally visited the Solaseado site in Haenam County, South Jeolla Province, and Junehee Lee, the head of Samsung SDS, even laid out a concrete project timeline at a CES 2026 (Consumer Electronics Show 2026) media roundtable. Lee briefly stated, "We will break ground on the National AI Computing Center in July" and added, "We are in the process of undergoing financial review." He did not elaborate further, but his remarks hint at the weight of the challenges involved. Once the financial review is completed and the preferred bidder is formally announced, the consortium will have barely five months to finalize all detailed schedules related to the center’s design and construction.
President Lee Jae-myung of South Korea likened AI infrastructure to an "expressway" at the groundbreaking ceremony for the Ulsan AI Data Center in June last year. He stressed, "If the Gyeongbu Expressway was the starting point of Korea’s industrialization, now it is time to lay down an AI expressway." As global competition over AI transformation, driven by AI agents, intensifies, the importance of this AI expressway has only grown. Building such an expressway is not simply a matter of installing more GPUs. It is an area that demands the government’s full administrative capacity, with the state taking on the initial risks that the private sector cannot bear and coordinating finance, site selection, power supply and regulatory frameworks all at once. With both the operator and the site now chosen, execution speed is what matters. At the very least, to secure basic competitiveness, the government needs to accelerate the rollout of this infrastructure.
ksh@fnnews.com Kim Sung-hwan Reporter