Godongjin: "Honam Semiconductor Is for Election Purposes... The Market Could Shift by Completion" [AI Rulemakers] ⑨

"Honam fab cannot be operational within Lee's term... a political show for elections"

Positive assessment of the government's independent foundation model policy

TSMC production capacity at saturation... "A golden opportunity for Samsung Fou

Editor's NoteSouth Korea became the second country in the world, following the European Union (EU), to enact a comprehensive Artificial Intelligence (AI) Basic Act. Now, half a year after the law took effect, are our regulations and systems keeping up with the rapid pace of technological and industrial change? Through the [AI Rulemakers] series, The Asia Business Daily interviews key political figures leading AI-related legislation to explore institutional solutions for South Korea to survive in the global AI competition.

Godongjin, a member of the People Power Party, criticized the Honam semiconductor cluster initiative from the Lee Jaemyung administration, calling it an "election-driven plan hastily put together in line with the President's term." He argued that initiating operations within the current presidential term would be difficult, and further pointed out that by the time the project is completed, both the semiconductor market and its main products could have changed.

Representative Go, formerly CEO of Samsung Electronics, said in an interview with The Asia Business Daily on the 16th, "Semiconductor investment decisions are usually reviewed over several years, yet this plan was decided in just a few months—like beans roasted over a lightning fire. This is highly unusual." He added, "It raises questions about whether this is an election-focused political show aimed at forcing companies to show results during the President's term."

Godongjin, a member of the People Power Party, is being interviewed by The Asia Business Daily at the National Assembly. Photo by Hyunmin Kim

He also dismissed the government’s target of beginning operations by June 2030 as “an unrealistic schedule for anyone familiar with the realities of the industry.” He explained, “Power supply, water, transmission networks, as well as production and R&D personnel and partner companies, all need to be secured. Even if everything goes smoothly, it could take eight to ten years before actual operations can begin.”

He continued, “High Bandwidth Memory (HBM) is currently in the spotlight, but who knows if the market will want the same products at that time? Companies’ investment timelines should not be determined by presidential terms, but rather by the pace of the market and technology.”

Regarding Korea’s AI competitiveness, he acknowledged the gap with the United States but said there is ample potential to catch up. Go stated, “AI competitiveness should be assessed in terms of three elements: algorithms, computing power, and data. Domestic proprietary foundation models are currently estimated to handle about 800 billion parameters.”

He was particularly positive about the continued government support for developing proprietary foundation models. “The government’s approach—supporting companies with GPUs and narrowing down participating companies based on evaluations—is very much in line with the direction envisioned by the People Power Party when it was the ruling party. Progress is being made,” he stated.

Using China’s DeepSeek as a case of rapid catch-up, he further emphasized, “We must devote ourselves to developing our own models, by any means necessary.” He proposed that the final models be released as open source, so they can be leveraged as the foundation for developing AI across diverse fields including healthcare and defense.

In the semiconductor industry, he stressed the need to bolster the non-memory ecosystem in addition to memory. Go said, “With TSMC’s production capacity nearing its limit and companies like Tesla and OpenAI expanding partnerships with Samsung, now is the perfect opportunity for Samsung Foundry. We need to establish a virtuous cycle in which domestic foundries manufacture the semiconductors designed by domestic fabless companies.”

He saw a need to support fabless startups that lack financial resources through the use of the National Growth Fund. “Many companies, even those with good technology, struggle to commercialize because of the high cost of building prototypes. The National Growth Fund should be boldly deployed at the prototyping stage, and used to foster connections with domestic foundries to help these companies grow,” he emphasized.

Godongjin, a member of the People Power Party, is being interviewed by The Asia Business Daily at the National Assembly. Photo by Hyunmin Kim

The following is a Q&A with Assemblyman Go.

-How would you assess Korea’s level of AI competitiveness?

▲AI competitiveness should be evaluated based on three elements: algorithms, computing power, and data. Korea’s proprietary foundation models currently process about 800 billion parameters. Although there remains a substantial gap with the leading models in the United States, given the rapid pace of advancement in AI, it likely won’t take a decade to catch up as it did in the past. The infrastructure for computing and data utilization is also being established, so what remains is effective execution.

-Are domestically developed proprietary foundation models progressing properly?

▲The government’s strategy—supporting elite teams with GPUs and narrowing down funding recipients through phased evaluations—was originally envisioned by the People Power Party when it held the majority. Currently, with the final two out of three teams to be selected soon, development should accelerate further. The final models must be released open source, allowing multiple companies to use them as the basis for developing AI across fields such as healthcare and defense. I believe the Lee Jaemyung administration is doing well in continuing the foundation model program.

-What is necessary to strengthen Korea’s semiconductor ecosystem?

▲We should not remain focused only on memory, but develop the broader non-memory sector. With TSMC’s production at capacity and both Tesla and OpenAI expanding cooperation with Samsung, now is an optimal time for Samsung Foundry. Simultaneously, we need to build a virtuous cycle where domestic foundries manufacture semiconductors designed by domestic fabless companies. For fabless startups lacking financial resources, aggressive support should be provided from the prototyping stage, for example using the National Growth Fund.

-How do you evaluate the government’s Honam semiconductor cluster initiative?

▲Barely two months after Euisun Chung, Chairman of SK Group, remarked, "It's true you need to be where the electricity is, but why must it be a semiconductor plant?" the Honam semiconductor investment plan was announced. Approving a semiconductor investment in the time it takes to roast beans over a lightning fire, a process that normally takes years of review, is highly unusual. It is questionable whether this is simply a ‘company-bullying political show’ by the Lee Jaemyung administration. Instead of the government dictating the location and calling on companies to follow, companies should select sites and the government should provide the necessary infrastructure.

-Is the goal of beginning operations in 2030 realistic?

▲Aiming to start up the Honam fab in June 2030, coinciding with President Lee Jaemyung’s term end, is unrealistic from the perspective of industry professionals. Securing power, water, and transmission lines—as well as the fact that "Pyeongtaek is the southern limit for production workers and Pangyo for R&D personnel"—underscores major workforce and ecosystem issues. Even if all conditions are optimal and operations begin in 8–10 years, there is no guarantee the market will still demand HBM as it does now. Corporate schedules should follow the pace of the market and technology, not presidential terms.

-How do you view China’s technological catch-up?

▲Over the past 15 years, Korea has lost ground to China in home appliances, TV, and smartphones. For semiconductors, the gap has been reduced to roughly one year in NAND Flash, two years in memory, and three years in HBM. Had the U.S. not stepped in to restrict exports of advanced equipment, Chinese progress would have been even faster. Korea must remain vigilant since Chinese government officials and civil servants provide robust support to business based on deep industry and technological knowledge.

-What policies are needed to secure AI talent?

▲We need to build an environment in which companies can secure talent and focus on research. Conscripted service exemptions for core AI and semiconductor experts should be expanded, and systems must be reformed so that even large corporations can utilize necessary talent. The 52-hour workweek and compensation structure should also be applied flexibly to reflect the realities of R&D workplaces.

-What needs to change for more businesspeople to enter the National Assembly?

▲The current blind trust regulations for stock holdings act as a barrier for entrepreneurs entering politics. While I agree with the intent to prevent conflicts of interest, rather than forcing unconditional stock sales, alternatives could include prohibiting trading during and for a defined period after the term of office. More reasonable systems must be established so industry experts can leverage their experience in the National Assembly.

-What is the difference between corporate and parliamentary decision-making?

▲Companies revise their strategies the moment market conditions change, whereas the National Assembly can take years to pass or amend laws and regulations. For instance, it took one year and seven months from proposal to passage of the Special Semiconductor Act. When I was in the private sector, it felt like driving in a car at 200 km/h; in the National Assembly, it feels like trailing behind on a broken bicycle. Telling companies to win in the global marketplace despite this is unfair. Politics, too, must keep pace with change and deliver real solutions.

-What must the People Power Party change to regain public trust?

▲The party must move beyond “just being good at political fighting” to become “a party that solves problems.” It is not enough to merely criticize government decisions; proposals must be made on how to build the semiconductor industry over the next decade, or how to secure power supplies for the AI era. By solving problems for the people and for businesses one by one, we must regain the trust that “if you leave the economy and industry to the People Power Party, at least those will be managed well.” The competitiveness of a conservative party lies not in protecting the past, but in preparing for the future.

This content was produced with the assistance of AI translation services.