Go big or go home: South Korean presidential hopefuls pledge mega AI investments

Straits Times

24-04-2025

South Korean candidates from the rival main political parties on both sides of the aisle are rolling out various AI policies on the campaign trail. PHOTO: REUTERS
Go big or go home: South Korean presidential hopefuls pledge mega AI investments
SEOUL - In South Korea's heated race for the early presidential election on June 3, artificial intelligence has become the latest buzzword.
Candidates are vying to outbid one another with eye-popping investments and grand promises, each casting themselves as the ideal leader for the new era.
From a 'Korean-style ChatGPT' to mandatory military service exemptions for tech talent to ambitious plans to nurture the next Nvidia or Palantir, candidates from the rival main parties on both sides of the aisle are rolling out various policies on the campaign trail.
However, observers warn that many of the plans lack concrete road maps or practical details.
Numbers game
Overall front-runner Representative Lee Jae-myung, former chair of the liberal Democratic Party of Korea kicked off the AI race by pledging a 100 trillion won (S$91.9 billion) investment in AI over five years to strengthen Korea's global competitiveness in AI.
With this massive budget, he vowed to build national AI data clusters, secure more than 50,000 cutting-edge graphic processing units and support the development of AI-critical neural processing units.
He also floated the idea of developing a 'Korean-style ChatGPT', providing it free to the public.
Other plans incudes creating schools specialised in AI at national universities and expanding the range of mandatory military service exemptions for science and tech talent.
Fellow Democratic Party candidates Gyeonggi Governor Kim Dong-yeon and former South Gyeongsang Governor Kim Kyoung-soo also joined the investment spree, each pledging the same 100 trillion won.
Mr Kim Dong-yeon, with a focus on building large-scale infrastructure and ecosystem, pledged to secure 1 million GPUs and nurture 100 top-tier AI professionals.
Mr Kim Kyoung-soo's plan focuses on public-private partnerships to develop a Korean AI foundation model in a bid to secure 'AI sovereignty'.
Attempting to outmatch the liberal bloc, conservative presidential hopeful Han Dong-hoon, former justice minister and leader of People Power Party, raised the stakes to 200 trillion won in what is the largest promise by a contender yet.
Of that, 150 trillion won is earmarked for infrastructure, with the remainder allocated to AI-related fields such as in defense, health care and robotics.
He also proposed to create a new government agency, tentatively named the Ministry of Future and Strategy, and pledged to nurture 10,000 AI experts.
Another People Power Party hopeful, Mr Hong Joon-pyo, the recently resigned mayor of Daegu, pitched a relatively modest 50 trillion won.
His policies include expanding research and development in cutting-edge industries including, AI, quantum computing and superconductors, along with pushing for a 'gate-free' innovation model of reducing barriers and regulatory hurdles for new technologies and businesses.
Mr Kim Moon-soo, former labour minister, promised 100 trillion won to foster 200,000 talented AI professionals and develop an AI-based education system.
Meanwhile, Representative Ahn Cheol-soo of People Power Party, who was once an anti-virus software entrepreneur and medical doctor, touted himself as the most qualified candidate to lead the AI industry.
Mr Ahn laid out broader science and tech agenda, pledging to train 1 million science and tech experts, boosting R&D spending to 5 per cent of GDP and establishing a 20 trillion won start-up fund.
Money will not build AI nation
Despite the parade of lofty promises, observers are raising concerns that many of the proposals lack practical details.
Even more tech-savvy presidential contenders have attacked one another for being out of touch.
Mr Lee Jun-seok, candidate for the minor conservative Reform Party, who studied computer engineering and economics at Harvard University, blasted candidates' policies as 'scam-level', that lacks understanding of the industry.
'The fundamentals of the AI industry come from stimulating private investments,' Mr Lee told reporters on April 16.
'What's important is talking about deregulation, but discussing superficial things like 100 or 200 trillion won only shows a lack of understanding of this industry,' according to him.
Mr Lee Jun-seok further condemned Mr Lee Jae-myung's policy of providing free AI services to the Korean public, as a 'foolish attempt linked to his 'freebie' policy series'.
He also took a swing at Mr Han for simply doubling Mr Lee Jae-myung's proposed investment amount to 200 trillion won.
Mr Ahn echoed a similar view, saying: 'Without a mid- to long-term blueprint for national science and technology development, shouting 100 trillion won without any consideration won't develop AI technology.'
While experts and industry officials welcomed the surge in focus and investment in AI at the national level, two of them voiced concern over the lack of detailed vision and actionable plans.
'The greater the national investment, the better,' said Dr Choi Byung-ho, a professor at Korea University's Human-inspired AI Research Lab.
'But the focus should be on what the budget supports and the overall direction.'
Prof Choi identified three priorities that the policies need to address if South Korea is serious about raising its AI competitiveness - talent, GPUs and data - the trio that are interdependent.
'At a time when the technology is accelerating, we need to invest more on people. Simply creating new AI departments won't solve the problem; that's been done before. But they need to make sure existing ones get the resources, research funding and benefits like military exemptions to retain talent,' said Prof Choi.
The professor stressed that securing GPUs are critical to running data centres and developing South Korea's own foundation model, which also require large amount of data.
'Obtaining data costs lots of money. Public data is still remains largely inaccessible, and so are legal documents and medical data. The question is whether the government will make these data accessible for AI, and if so, how and to what extent.'
Mr Kim Sung-hoon, CEO of Korean AI startup Upstage, also stressed these three pillars.
'I am very happy that all the presidential candidates have placed AI-related agendas at the top of their priorities,' Mr Kim said at a press event last week.
'But what's essential for AI development is support for training data, GPUs and the recruitment of top talent.' THE KOREA HERALD/ASIA NEWS NETWORK
Join ST's Telegram channel and get the latest breaking news delivered to you.