South Korea proposes 4.7 trillion won allocation for sovereign AI model
President Lee’s 2027 budget proposal sets aside 4.7 trillion won for a sovereign AI model alongside broader spending on chips and data centres.
By David L. Chen · Senior Columnist
· 3 min read
South Korea’s sovereign AI budget proposal includes a separate 4.7 trillion won allocation to develop a domestic AI model, President Lee Jae Myung said, with a stated aim of reaching the “world’s highest performance.” The allocation sits within a proposed 2027 national budget of 820.9 trillion won, up 12.8% from 2026, which Lee presented to parliament on November 2.
The proposed budget still needs parliamentary approval. Reuters reported that it would devote 21.3 trillion won to AI and three megaprojects: semiconductors, physical AI and AI data centres. CNBC separately reported the 4.7 trillion won allocation for the sovereign model, meaning the wider 21.3 trillion won package should not be read as funding for the model alone.
What is in South Korea’s sovereign AI budget?
The government has identified two distinct layers of technology spending. The larger 21.3 trillion won package covers AI alongside semiconductors, physical AI and AI data centres, and is nearly twice the current year’s amount, according to Reuters. The 4.7 trillion won allocation is specifically for the sovereign AI model, CNBC reported.
Lee’s description of the intended model’s performance is a policy target rather than evidence of an existing system or independently measured benchmark result. The reports do not set out a timetable, technical design or a breakdown of the model allocation.
Physical AI is one of the three named megaprojects, alongside chipmaking and AI data centres. The budget proposal does not provide further detail in the reporting on individual programmes within that category.
Why is South Korea increasing AI and chip spending?
Lee presented the proposed outlays as part of an effort to raise South Korea’s growth potential amid stronger demand for AI-related computing. Reuters reported that demand for advanced chips used in AI has supported exports and the earnings of memory-chip makers Samsung Electronics and SK Hynix.
Exports totalled $930 billion between January and October 2026, Reuters reported, and Lee said annual exports of $1 trillion were within reach. He also said the gains from growth had not been shared evenly, calling for a move away from benefits concentrated among a limited number of regions, groups and companies.
The proposal also includes a talent component. CNBC reported that the government plans to train 2,000 specialists in semiconductor design and regional semiconductor industries, while expanding AI education for students, job seekers and existing workers.
How would the proposal affect public finances?
Lee said the fiscal deficit would narrow to 0.1% of gross domestic product in 2027 from 3.9% this year, while the debt-to-GDP ratio would fall to 48.3% from 51.6%, according to Reuters. Those figures are the president’s statements accompanying the proposal, rather than an independent assessment in the reporting.
The budget also includes a proposed 162.3 trillion won Future Response Fund. Reuters reported that it would be financed largely with additional tax revenue linked to the semiconductor boom and used for growth industries, young people, regional development and education, as well as a buffer against future revenue shortfalls.
This story draws on original reporting from CNBC.
