"The United States and China are ahead in artificial intelligence (AI) models, but Korea can seize a big opportunity in the actual deployment of physical AI. (Abhinav Gupta, co-founder of Skild AI)
"It would be good for Korea to be a bit more ambitious in AI and aim for No. 1. It is important to build a sustainable industrial ecosystem." (Kim Tae-soo, corporate vice president for security at Microsoft)
On the 26th, SMARTCLOUD SHOW 2026, the country's largest tech conference, held at the Westin Josun Hotel in Jung-gu, Seoul, concluded with great success. With the theme "AI: Shaping the World," this year's event brought together world-renowned scholars and industry leaders in physical AI, security, Robotics, semiconductors, and space to seek ways for Korea to move toward becoming one of the top three in AI and to present a blueprint for future industries. Hosted by the Ministry of Science and ICT and organized by ChosunBiz and the National IT Industry Promotion Agency (NIPA), the event drew so many attendees that the 300 or so seats prepared on-site were not enough.
Hwang Jeong-a, a lawmaker from the Democratic Party of Korea, said, "As a manufacturing and IT powerhouse, Korea can leverage its industrial ecosystem to become No. 1 in the physical AI era." Ryu Je-myeong, second vice minister of the Ministry of Science and ICT, said, "To secure leadership in AI, we need a stable infrastructure such as large-scale computing and data along with strong models." Oh Se-hoon, the Seoul mayor, said, "We will build an innovation ecosystem where corporations, talent, and technology gather around the 'Seoul Physical AI Belt,' and turn the entire city into a testbed for new technologies."
◇ "As a manufacturing powerhouse, Korea can find opportunities in physical AI"
As the government speeds up efforts to achieve "top three in AI" and "No. 1 in physical AI," experts suggested leveraging Korea's competitiveness in manufacturing as a strength in physical AI.
In a keynote speech, Abhinav Gupta, co-founder of Skild AI and a professor of computer science at Carnegie Mellon University in the United States, said, "There are not many physical AI robots actually deployed in the U.S., and in China we mostly see entertainment robots," adding, "If we combine need and opportunity, Korea can be the first to join physical AI experiments and secure deployment experience that is hard for other countries to accumulate." However, he noted, "We need to ease regulations and speed up procedures until actual field deployment." Founded in 2023, Skild AI is a robot AI company valued at about 20 trillion won, developing as its core technology a robot foundation model (RFM), the "brain of the robot," that can be applied to various types of robots. Lim Hyo-joon, head of the Next-Generation Computing Research Institute at LG Electronics, also said, "Korea has a very strong manufacturing base and an abundance of data generated in factories, so it can find opportunities in physical AI."
Opinions differed on what strategy Korea should take in the AI race. Kim Tae-soo, corporate vice president for security at Microsoft (a professor of computer science at the Georgia Institute of Technology), said, "Among startups developing open-source models, many have teams of 20 to 30 outstanding talents," adding, "We should pursue a strategy to be No. 1, not No. 3." By contrast, Jeong Seok-geun, CEO of SK Hyper, said, "Even if it's third place, we should set a proper strategy and become No. 3 quickly," adding, "Demand for a third player outside the U.S. and China is very strong." As the U.S. leads global AI standards and China builds an independent AI ecosystem, demand for sovereign AI could grow, particularly among countries that find it difficult to rely entirely on the two countries' technologies.
◇ "In the AI era, attacks are bombs, not arrows… We should avoid unconditional adoption"
As AI technology advances faster, threats surrounding corporations' security environments and infrastructure are changing rapidly. Experts said that because AI can both amplify the destructive power of cyberattacks and serve as a means of defense, corporations must fundamentally overhaul their security systems.
Lim Hyo-joon, executive vice president and head of the Next-Generation Computing Research Institute in the CTO institutional sector at LG Electronics, said, "In the AI era, an attack is like a bomb flying in, not an arrow." Park Song, head of the Cybersecurity AI Lab at SK shieldus, said, "The era of blindly building higher walls is over," emphasizing, "We must establish a cyber resilience system that can recover on its own even after an attack." In fact, there was a case where AI solved 74% of high-difficulty red-team tasks that existing AI models could not solve. Tom Scully, director for government and critical infrastructure in the Asia-Pacific and Japan region at Palo Alto Networks, said, "The only way to respond to cyberattacks that use real-time AI is to fight with AI," and added that organizations should establish controls such as granting access only to the systems and data needed by AI agents. According to Unit 42, the threat intelligence organization at Palo Alto Networks, it takes just 15 minutes from when a major vulnerability is disclosed to when attackers launch large-scale scans targeting it.
There was also a warning that AI should not be adopted unconditionally. Park Jun-sik, head of the Korea office at STMicroelectronics, said, "We must not leave all decisions of humanoids to a central AI processor," stressing the importance of a distributed intelligence architecture. Gong Seong-bae, chief AI officer at Megazone Cloud, said, "When corporations adopt AI, they should make the harness the core," adding, "If you recklessly apply AI across the entire organization, you are bound to fail." A harness refers to an execution environment for controlling and managing AI models efficiently and stably.
As AI competition intensifies, securing infrastructure has also emerged as a key task. Lee Il-jun, a director and head of the Technology Business Division at NHN Cloud, said, "AI developers worry about performance, but infrastructure managers are now worrying about power," adding, "AI competition is ultimately a competition to secure data centers and power." Shin Jeong-gyu, CEO of Rebellions, said, "Currently, there are about 20 agent AIs running per person, but that could increase by 100 times next year," forecasting, "The AI infrastructure bottleneck will expand from GPUs to overall computing resources such as CPUs, memory, storage, and power." The spread of AI agents is also expected to lead to changes in the related software ecosystem. Lee Hae-seok, vice president at Polaris Office, said, "The use of software as a service (SaaS) that supports AI agents to work better will further increase."
◇ "Combining satellites with AI extracts meaningful information… We must solve data and security issues together"
In the satellite sector, AI is emerging as a core technology that analyzes satellite imagery and Earth observation data in real time and supports decision-making in diverse areas such as climate, disasters, and defense. Pekka Laurila, co-founder and chief strategy officer (CSO) of ICEYE, a Finnish space company founded in 2015, said, "The use of satellites is expanding beyond disasters to national security and public safety," adding, "By applying AI, we can automatically analyze objects and movements in imagery to extract meaningful information."
However, there was also a point that to combine AI and satellite technology, we must solve data and security issues together. Park Won-gyu, an executive vice president at Satrec Initiative, said, "To connect AI's potential to real industrial competitiveness, we must address data and security together." The explanation was that if Korea is to secure "K-space AI sovereignty," which combines indigenous satellite technology with AI, it must overhaul related systems and security frameworks so that high-quality data can be used for AI training.