SK Group Chairman Chey Tae-won said the merger between SK Innovation and SK E&S, SK Group’s leading energy companies, was conducted to deal with challenges related to the growing demand for artificial intelligence (AI).
“AI requires substantial energy, which cannot be met solely by batteries or traditional energy sources like liquefied natural gas (LNG) and hydrogen,” he said at a Korea Chamber of Commerce and Industry press conference in Jeju on July 19. “But if the two companies merge, SK can effectively tackle the energy challenges related to AI data centers.”
SK Innovation and SK E&S held separate board meetings on July 17 and approved the merger plan. The merger is set to create an energy behemoth with over 100 trillion won ($72 billion) in assets. The following is a Q&A with Chairman Chey.
What are the conditions for maintaining leadership in the AI sector, including high-bandwidth memory (HBM)?
“We need to build a robust AI infrastructure in Korea. If we fall too far behind, we could become dependent on other countries. Even in tough economic times, we must build a lot of infrastructure, including AI data centers.
I consider people who understand AI and open new businesses as “AI warriors.” We need to lay a solid and supportive foundation for these people to thrive. We should establish data centers, attract companies like Microsoft, Open AI, Amazon, and Google, and open AI to citizens and students.
There are concerns that South Korea lacks data for AI learning.
“We must collect more data. AI data centers alone are not enough. Even if we gather all the data in Korea, it is still insufficient. We need to cooperate with other countries like Japan to increase data size and gather specialized data in fields such as manufacturing, bio, and media. This cooperation is essential to develop systems for sharing, converting, and utilizing data effectively.”
What is the future direction of the semiconductor business?
“In the past, we increased the integration density of semiconductors via research and development. The ratio needed for technological evolution was 9 for R&D and 1 for capital expenditure. Recently, this ratio changed to 5 for R&D and 5 for investment. We have reached the limit of increasing density through R&D. The market demands endless [chip] upgrades, which require massive capital investments. Building one factory costs 20 trillion won. For HBM, it costs even more. This means we are in an era that demands tremendous investment.”
What are your expectations for government support, including tax benefits?
“Tax benefits alone are no longer sufficient. In the past, we jumped from 20 nanometers (1 nanometer is one-billionth of a meter) to 12 nanometers, but now it’s almost 2 nanometers, 1 nanometer. This ultimately requires a lot of capital investment. Many factories in the United States and Japan are being built with significant support. We need to follow suit.
Increasing memory capacity is crucial for AI but requires more investment than profits can cover. Government support is vital to sustain this investment, particularly during potential demand fluctuations. Investing in HBM is expensive, and sometime in the future, AI may also experience a chasm (temporary stagnation in demand), similar to what is happening with batteries.”
How can we address labor shortages in science and engineering, including so-called ‘AI warriors’?
“We need to break down the dichotomy between science and humanities. You can’t suddenly create a workforce that doesn’t exist. We need to start planting the seeds now. A specialized professional education system in high schools and universities is necessary. We should build a lot of AI infrastructure and provide training and education from elementary school. Then, in 10 to 15 years, we will have a capable workforce. Attracting skilled workers from India and many other countries is also a possible solution to the workforce shortage problem.”
What are your predictions for the U.S. presidential election?
“Predictions are tricky. Regardless of the outcome, the policy towards China is unlikely to change. There will likely be substantial changes in environmental and industrial policies. If Trump wins, uncertainty may increase.”
