Korea Aims for Standalone Global No. 3... Senior Secretary Ha Jungwoo: "AI Is Not a Bubble, but an Opportunity We Must Seize" [Q&A]

Interview with Senior Secretary Ha Jungwoo on December 10
Action Plan for Scaling Up to Top Three in AI to Be Unveiled This Year
Alliance of Middle Powers Possible If Standalone Third Place Is Out of Reach

The Lee Jaemyung administration will unveil a national-level "action plan" for achieving "AI Top 3 Advancement" within this year. The plan aims to elevate South Korea’s status from being part of the "third-place group" behind the United States and China to securing an undisputed third place. This comprehensive execution plan covers AI infrastructure, data centers, AI semiconductors, industrial transformation (AX), talent development, and institutional reforms. The government believes that if South Korea firmly secures the third spot in AI competitiveness, the country-which has consistently emphasized building an "AI-based society"-will be able to compete with China even in terms of security and trust.

On December 10, Ha Jungwoo, Chief of AI Future Planning at the Presidential Office, said in an interview with The Asia Business Daily at the Yongsan Presidential Office, "We have already developed an action plan in collaboration with various ministries to strengthen AI competitiveness," adding, "We have decided to present the contents to the public in some form within this year." The plan is expected to proceed in a manner similar to the United States, where a draft is released first to gather feedback from AI-related companies and experts, after which the concrete execution plan is finalized.

The following is a Q&A with Chief Ha.

Is it really possible for South Korea to be among the "AI Top 3"? Are there concerns about a bubble?

While the United States is firmly in first place and China in second, the third spot is being contested by the United Kingdom, France, South Korea, Canada, and Singapore, forming a "third-place group." According to global indicators, South Korea has risen from 7th to 6th to 5th place, and the country shows relatively balanced intensity and scale in areas such as talent, research and development (R&D), infrastructure, and policy. With government strategy, corporate investment, and talent development in place, I believe a leap into the Top 3 is entirely possible.
Rather than seeing AI as a bubble, we view it as an opportunity that must be seized now. While there are concerns due to large initial capital investments, the profit and revenue structure of companies is different from the dot-com bubble era. Although some adjustments may occur, we believe it is highly likely to serve as a springboard for long-term growth.

How should we define the "AI Top 3" nations? What is the roadmap going forward?

It is no longer possible to rank countries solely based on the performance of individual models like ChatGPT or Gemini. Just as evaluations of internet powerhouses consider not only network speed but also the number of unicorns, platform sales, and GDP contributions, AI should be assessed comprehensively, including energy infrastructure, memory, GPUs, data centers, cloud, foundational technologies, and industrial application capabilities. Going forward, indicators such as "How much does AI contribute to GDP?" and "To what extent are South Korea’s AI technologies and infrastructure utilized overseas?" will become increasingly important.
If securing an undisputed third place proves difficult, a "Top 3 Alliance" of middle-power nations is also a possibility. Reflecting this perspective, we have created an AI action plan containing implementation tasks by ministry and sector, with the goal of making it accessible to the public within this year.

Can South Korean AI win in price and technology competition with China?

It is not necessary to surpass China in every field. Many middle-power nations are wary of relying too heavily on China, yet find it difficult to commit solely to the United States. South Korea is a rare country equipped with energy infrastructure, memory semiconductors, GPUs, data centers, cloud, large language models (LLMs), as well as factories and industrial robots. Even if Chinese AI scores 100 points at 100 won, and Korean AI scores 90 points at 110-120 won, AI is directly linked to security, so there is strong motivation to choose a predictable and trustworthy partner. If the performance and price are within an acceptable range, South Korea’s position is sufficient.
Regarding talent, rather than simply increasing the number of pure AI specialists by the tens of thousands, we believe it is more realistic for experts in industries such as automotive and shipbuilding to learn AI, thereby enhancing AI utilization capabilities within each sector.

Is there a plan for the government and public sector to act as the "first customers" for domestic AI semiconductors such as Neural Processing Units (NPUs) and Tensor Processing Units (TPUs)?

Now that meaningful domestic AI semiconductor products are emerging, we agree with the need for the government and public sector to play the role of early adopters. The specific methods and timing are being coordinated within relevant ministries’ policies, and we are considering ways to generate demand by linking with GPU voucher programs and national projects.

How will South Korea participate in the United Arab Emirates (UAE) "Stargate Project"? What are the timeline and scope?

The "Stargate" project led by OpenAI in the United States and the "UAE Stargate" project in the UAE should be seen as separate initiatives. The UAE project, in collaboration with OpenAI and NVIDIA, envisions building an AI data center with a capacity of up to 5GW; 1GW has been outlined, while the remaining 4GW is still in the design phase.
Centered around UAE’s state-owned AI company G42 and its data center subsidiary, the project will involve designing a total energy solution, including power generation, transmission, distribution, solar, nuclear, and gas. Currently, a broad agreement has been reached at the presidential and government levels, and from January next year, working groups from both countries will begin in-depth discussions, with the division of roles and timeline likely to become clearer by the first half of the year.

How will the government-allocated GPUs be distributed? Will there be support for private sector purchases?

The 13,000 GPUs secured through the supplementary budget are being prepared for use by startups and universities from the first half of next year, following server and network installation and testing. Rather than distributing physical GPUs, the plan is to provide credits or vouchers allowing projects to use a certain amount of GPU resources. The specific ratios and recipients are being designed by the Ministry of Science and ICT.
While it is not appropriate for the government to intervene directly in private investments, such as purchases of 50,000 GPUs by companies like NVIDIA, the government can help create domestic demand through voucher programs, which may encourage companies to invest in infrastructure. We are also considering ways to expand the scope of utilization to include not only government-allocated GPUs but also those introduced by the private sector.

There are concerns that the introduction of 260,000 NVIDIA GPUs will lead to a "NVIDIA monopoly." What is your response?

Given the current restrictions on exporting high-end GPUs to China, the interpretation that "leftover stock from China is being diverted to South Korea" is somewhat exaggerated. Globally, GPU supply remains extremely tight, and competition among countries to secure them is intense.
Since it is not feasible to completely replace GPUs with domestic AI semiconductors in the short term, a realistic two-track strategy is needed: securing sufficient GPUs while simultaneously fostering the domestic NPU and TPU ecosystem to reduce dependency. Related policies will be announced sequentially.

What is the vision for an "AI-based society" alongside the "AI Highway" initiative?

An AI-based society envisions integrating AI into essential social safety nets such as welfare, disaster safety, education, finance, and healthcare, so that everyone can enjoy AI-enabled problem-solving capabilities as a basic right. This includes not only bridging the digital divide within South Korea but also narrowing gaps between countries.
South Korea’s proposals at the United Nations, APEC, and G20 summits that "AI opportunities should not be monopolized by a few powerful nations" are in line with this vision. This agenda is not just about principles-it has practical benefits as well. Since enormous investments are required in power, data centers, software, and talent to realize an AI-based society, it is crucial to translate collaborations with BlackRock, the UAE, OpenAI, and NVIDIA into concrete contracts and investments going forward.
AI competition will not be easy, but if the government, business, and academia work as one team and South Korea expands alliances with more countries starting with the UAE, I believe it is entirely achievable.