Can AI be a public utility? World watches Korea's bold experiment

South Korea’s AI for All initiative is drawing global attention as a bold experiment in treating AI as public infrastructure. Under the plan, SK Telecom, KT and Kakao-led consortia will offer general-purpose AI services free and without usage limits, with beta testing set for next month and a full launch in December. The government will provide up to 512 Nvidia B200 GPUs this year and begin covering nationwide operating costs in 2027. Officials say AI access should be a basic competency, while experts and social media users are debating whether unlimited access is sustainable.

The government plans to give participating consortia up to 512 Nvidia B200 graphics processing units this year.

Beta testing is set to begin next month, ahead of a full launch in December.

The government plans to cover nationwide operating costs beginning in 2027.

Participating services must rely on qualifying Korean-developed AI models for at least 50 percent of their AI model usage, with at least another 30 percent allocated to models developed by other Korean companies.

Published Sep 10, 2026 2:49 am KST

Updated Sep 10, 2026 11:06 am KST

SEATTLE — Korea's plan to give everyone free, unlimited access to artificial intelligence (AI) initially landed at home as another government technology initiative. But over the past week, foreign media and commentators have cast it in far more consequential terms: a bold experiment in treating AI as public infrastructure.

The Wall Street Journal described Korea's “AI for All” initiative as the first major state-led effort to treat generative AI as a public utility, while TechRadar, a global technology news and review publication, said other governments would be watching the experiment closely.

“Most governments still talk about artificial intelligence as something to regulate or subsidize. South Korea is trying something considerably bolder,” TechRadar said in an opinion piece this week.

The publication said Korea is treating AI access less like a premium software subscription and more like “a basic piece of national infrastructure,” a view echoed by industry experts.

“South Korea just made AI a public utility, like water and electricity,” Luis Felipe Gutman, managing partner at venture capital firm OW Ventures, wrote in a recent post.

“The gap between people who can afford AI and people who can't is one of the defining inequalities of this decade,” he added.

The reactions have cast a broader spotlight on a project that has largely been discussed in Korea in terms of its technology, cost and implications for the domestic AI industry.

Under the initiative, consortia led by SK Telecom, KT and Kakao will provide general-purpose AI services free of charge and without usage limits. Beta testing is set to begin next month, ahead of a full launch in December.

The government is providing up to 512 Nvidia B200 graphics processing units this year and plans to cover nationwide operating costs beginning in 2027.

But the government's stated ambitions extend beyond simply providing free access to AI. Officials have described the ability to use AI as a basic competency, much like Hangeul, or the Korean alphabet, and arithmetic, and warned that disparities in AI access could develop into broader economic and social divides.

“Korea is positioning itself as Asia's most interesting AI story, especially on diffusion,” Pranab Baruah, an AI and technology policy expert, said as he pointed to the country's large conglomerates, high technology adoption and the AI for All policy.

“This is essentially an industrial policy,” he added, noting how the government requires participating services to rely on qualifying Korean-developed AI models for at least 50 percent of their AI model usage, with at least another 30 percent allocated to models developed by other Korean companies.

The initiative has also drawn significant public interest. A Reddit post about Korea's plan drew more than 13,000 upvotes, with users questioning whether unlimited access can realistically be sustained and raising concerns about computing costs and privacy.

“If this works at scale, it could change how quickly people adopt AI,” Saurabh Aggarwal, a principal data scientist at Autodesk and guest professor of AI at San Jose State University, wrote in a public post.

“Making access easy is one thing, but building the right infrastructure and guardrails will be just as important,” he added.

The experiment could ultimately test a question other governments may soon face: whether increasingly powerful AI remains a service consumers buy or becomes something governments seek to make universally accessible.

韩国的“全民人工智能”（AI for All）计划因其大胆尝试将人工智能作为公共基础设施而备受全球关注。根据该计划，由SK Telecom、KT和Kakao牵头的联盟将免费提供通用人工智能服务，且不设使用限制。该计划将于下个月进行beta测试，并于12月全面上线。韩国政府今年将提供多达512块英伟达B200 GPU，并于2027年开始承担全国范围内的运营成本。官方表示，人工智能的使用应该成为一项基本技能，而专家和社交媒体用户则在争论这种无限制的使用模式是否可持续。

今年，政府计划向参与的联盟提供多达 512 个英伟达 B200 图形处理单元。

Beta 测试将于下个月开始，全面发布将于 12 月进行。

政府计划从 2027 年开始承担全国范围内的运营成本。

参与服务的 AI 模型使用量中，至少 50% 必须使用符合条件的韩国自主研发的 AI 模型，另外至少 30% 必须使用其他韩国公司开发的模型。

发布于2026年9月10日凌晨2:49（韩国标准时间）

更新于2026年9月10日上午11:06（韩国标准时间）

西雅图——韩国向全民免费、无限制地提供人工智能（AI）服务的计划，最初在国内被视为又一项政府科技举措。但过去一周，外国媒体和评论员却以更为深远的意义来解读这项计划：这是一项将人工智能作为公共基础设施的大胆尝试。

《华尔街日报》将韩国的“AI for All”计划描述为首个由国家主导的、将生成式人工智能视为公共事业的重大举措，而全球科技新闻和评论出版物 TechRadar 则表示，其他国家政府将密切关注这项实验。

“大多数政府仍然把人工智能视为需要监管或补贴的领域。而韩国正在尝试一些更为大胆的做法，”TechRadar本周在一篇评论文章中写道。

该出版物称，韩国对待人工智能的态度与其说是将其视为高级软件订阅，不如说是将其视为“国家基础设施的一部分”，这一观点也得到了业内专家的认同。

“韩国刚刚将人工智能变成了像水和电一样的公共事业，”风险投资公司 OW Ventures 的管理合伙人 Luis Felipe Gutman 在最近的一篇文章中写道。

他补充说：“能够负担得起人工智能的人和负担不起人工智能的人之间的差距，是本世纪最显著的不平等现象之一。”

这些反应使人们对这个项目有了更广泛的关注，此前韩国国内主要从技术、成本以及对国内人工智能产业的影响等方面对此进行了讨论。

根据这项计划，由SK Telecom、KT和Kakao牵头的联盟将免费提供通用人工智能服务，且不限制使用量。测试版将于下个月开始，全面上线时间为12月。

今年政府将提供多达 512 个英伟达 B200 图形处理器，并计划从 2027 年开始承担全国范围内的运营成本。

但政府的雄心壮志远不止于提供免费使用人工智能。官员们将使用人工智能的能力描述为一项基本技能，就像韩文（韩语字母）和算术一样，并警告说，人工智能获取方面的差距可能会演变成更广泛的经济和社会鸿沟。

“韩国正在将自己定位为亚洲最有趣的人工智能故事，尤其是在普及方面，”人工智能和技术政策专家普拉纳布·巴鲁阿说道，他指出韩国拥有大型企业集团、高科技应用以及“全民人工智能”政策。

“这本质上是一项产业政策，”他补充道，并指出政府要求参与的服务机构至少 50% 的人工智能模型使用量必须依赖于合格的韩国自主研发的人工智能模型，另外至少 30% 的人工智能模型使用量必须分配给其他韩国公司开发的模型。

这项计划也引起了公众的广泛关注。Reddit 上一篇关于韩国这项计划的帖子获得了超过 13000 个赞，用户们质疑无限制的访问是否真的能够持续，并对计算成本和隐私问题表示担忧。

“如果这项技术能够大规模应用，它可能会改变人们接受人工智能的速度，”Autodesk首席数据科学家、圣何塞州立大学人工智能客座教授Saurabh Aggarwal在一篇公开文章中写道。

“让出行变得容易是一回事，但建设合适的基础设施和防护措施同样重要，”他补充道。

这项实验最终可能会检验其他政府可能很快面临的一个问题：日益强大的人工智能究竟是消费者购买的服务，还是政府寻求让其普遍可用的技术。