Introduction
In July 2026, South Korea issued a tender for a service no G20 government had attempted before: providing every resident with free, unlimited access to an AI chatbot built entirely on domestically developed language models. The initiative, officially named "AI for Everyone," marks a turning point in how governments conceive their role in the era of generative AI.
An Unprecedented RFP
South Korea's Ministry of Science and ICT opened a call for candidates on July 14, 2026, to select two to three private operators responsible for designing and operating this public AI service — with a submission deadline of August 11. The shortlisted candidates include Kakao, Naver, SK Telecom, and LG, the country's dominant digital players.
The contract runs through 2031, and winning operators must build exclusively on language models developed within South Korea's borders. In exchange, the government provides up to 256 Nvidia B200 GPUs — with operators required to match that commitment with their own resources. This public-private co-investment structure sets the project apart from a conventional sector subsidy: the state is putting hardware on the table, not just budget lines.
Two Tiers of Service
The platform is designed around two complementary layers. First, a general-purpose chatbot available free of charge to all residents, with a beta launch expected by end of September 2026. Second, an agentic layer enabling citizens to interact directly with public services — online administrative procedures, eligibility checks for government benefits, and routine request processing. The full service is targeted for year-end and will be government-funded through 2028.
Digital Sovereignty as Core Architecture
The project is grounded in an explicit sovereignty agenda. One of the Korean government's stated motivations is to shield its citizens from dependency on foreign AI services — access to which could be subject to geopolitical decisions made abroad. South Korea has already launched a national program to develop foundation models, involving several consortia — Naver, LG, SK Telecom, Upstage, and NC AI — structured around a vertically integrated industry model.
The infrastructure relies on a fleet of 13,000 GPUs already acquired by the state, with a target of 50,000 units by 2030, operated within a National AI Computing Center co-managed by Naver Cloud, NHN Cloud, and Kakao. This deliberate buildout is backed by dedicated legislation: the AI Basic Act, which entered into force on January 22, 2026, making South Korea the second country in the world — after the European Union — to have a comprehensive legal framework for artificial intelligence.
What This Means for Organizations
Beyond the Korean context, this initiative raises structural questions for both public and private sector organizations. Treating AI as a universal service — on a par with education or public transportation — requires architectural, legal, and budgetary decisions that look very different from those driving most enterprise AI deployments today.
Concretely, it means prioritizing sovereign models over foreign-hosted LLMs, building owned or shared compute infrastructure rather than relying entirely on hyperscalers, and establishing AI governance frameworks before deploying capabilities at scale.
Technology and digital leaders tracking public AI strategies may find in this less a directly replicable blueprint than a clear directional signal: the era when AI was an optional tool reserved for innovation teams is likely over. The question is no longer "whether" — it's "with what architecture, and on whose terms."