Editor’s Note: This commentary is the second of a two-part series examining how AI is being operationalized in shipbuilding, defense manufacturing, automotive production, and semiconductor fabrication.
By James Kim, Director, Korea Program
This is the second article in a two-part series, suggesting that South Korea’s approach to AI focuses less on leading foundation models and more on building long-lasting industrial capabilities using AI. The discussion examines how AI is integrating across key sectors, such as shipbuilding and the defense, automotive, and semiconductor industries. The author argues that AI, if leveraged properly, can become an industrial and alliance asset that can prevent bottlenecks and variance while making scale possible in the production process without sacrificing quality.
Case I: Shipbuilding
Shipbuilding is a challenging industry characterized by thousands of parts, long lead times, tight tolerances, and constant workforce constraints. In understanding how AI is being integrated into this process in South Korea, we should look at the choke points — designing, planning, and scheduling; welding and automation; and inspection and quality control. South Korean shipbuilding engineers are using AI-powered simulated hydrodynamics and digital twin technology to predict vessel performance in different scenarios, helping them refine hull designs and reduce fuel consumption. Integration of AI- powered robotics in precision welding and predictive maintenance can increase productivity by 30%. AI-powered supply chain management solutions can also reduce bottlenecking on procurement. HD Hyundai has announced that it is building a smart shipyard using these technologies, and that AI is being applied not only to vessel operation but also to shipbuilding processes.
This is an example of a concrete application in manufacturing where a shipyard can anticipate bottlenecks, detect defects earlier, and coordinate work sequences more efficiently. It is not just about cost saving. It allows companies to dictate tempo and increase scale, which are important advantages in the world of contested supply chains and maritime competition.
Case II: Defense Industry
South Korea aspires to be the fourth-largest defense producer by 2030. According to SIPRI, it is the second-largest defense supplier in Asia behind China. South Korean defense exports to Poland illustrate what this ambition looks like in practice. Within five months since signing the award, Poland received an initial shipment of its order that included 10 K2 tanks and 24 K9 howitzers — a strong signal about the South Korean defense industry’s production capacity during a moment when Europe’s demand curve is steepening.
For South Korean defense contractors, AI will be an important contributor to increasing surge capacity. Like shipbuilding, defense manufacturing is a system problem: Bottlenecks hide in sub-tier suppliers, quality variance compounds late in the process, and sustainment capacity becomes a limiting factor faster than most planners expect. AI will not replace vital components or labor. What it can do is reduce friction and variance across the workflow. Predictive maintenance, AI-assisted planning, and faster inspection cycles mean not sacrificing quality for scale and speed.
In a geopolitical environment that is growing increasingly uncertain, the implication for allies is straightforward: The South Korean defense industry is an alliance asset if the U.S. and ROK can figure out how to integrate AI into their manufacturing systems by harmonizing standards (e.g., inspection evidence and traceability), pairing scale with secure operational technology (OT), and building pathways for shared readiness.
Case III: Automobiles and Robotics
The South Korean automotive industry highlights the use of “physical AI” in manufacturing, where technologies like digital twins, robotics, the Internet of Things (IoT), and artificial intelligence are already integrated to enhance flexibility and resilience in production systems. Hyundai Motor Group’s Innovation Hub, for instance, uses these tools to link real-time factory data to virtual simulation so processes can be tested and improved before changes hit the assembly line. The key to this approach is the feedback loop. When AI is embedded into sequencing, inspection, and maintenance, it generates high-fidelity data on failure modes, rework patterns, cycle-time variance, and supplier-driven defects. That data becomes a capability. It improves throughput today, and it trains the next generation of industrial tools tomorrow.
Use of robotics allows Hyundai Motors to take this process to the next level. South Korea’s major industrial players are openly framing humanoids and advanced robots as the next frontier of factory work — particularly for repetitive and high-risk tasks. Hyundai announced plans to deploy Boston Dynamics’ Atlas humanoid robots in manufacturing starting in 2028, explicitly positioning this platform as part of a broader “physical AI” push. Meanwhile, South Korea’s manufacturing AI policy is also pointing toward industrial robotics as a target domain, including government plans tied to manufacturing AI transformation and industrial-use humanoids. The key concern is how South Korea can test scalable physical AI deployment while addressing concerns surrounding safety standards, data-sharing, and workforce transition. The implementation of the AI Basic Act in January of this year is the first important step in this regard.
Case IV: Semiconductor and AI Optimized Fabs
Semiconductors sit at the center of Korea’s AI strategy in two ways: Chips are the critical input for AI, and fabs are one of the most complex manufacturing environments where AI can contribute in meaningful ways. Samsung and NVIDIA plan on building fab-scale digital twins using NVIDIA Omniverse to support anomaly detection, predictive maintenance, and operational optimization before changes hit the physical line. This matters because fab performance is defined by variance management in this sector and affects the bottom line. Tiny shifts in the manufacturing process can cascade into yield loss, rework, and schedule disruption. AI-optimized fabs aim to compress that variance by improving equipment performance, tightening process control, and accelerating learning cycles.
In short, AI helps build chips more efficiently, and those chips help scale AI. If South Korea can couple fab-level AI optimization with secure, auditable production systems, it strengthens not just national competitiveness but allied supply chain resilience, especially as semiconductors become an important part of defense and industrial capacity.
The Data Dividend
The International Federation of Robotics reports that Korea is the world’s top adopter by robot density, at 1,012 robots per 10,000 employees. But robot density becomes a data advantage only if factories instrument and govern the data, and if SMEs can participate without crushing integration overhead. When properly implemented, AI deployments in inspection, routing, maintenance, or process control generate high-fidelity operational data. Production line managers can quickly identify what failed, what passed, why, and under what conditions. Over time, that data becomes an important national asset: It captures process parameters, routing decisions, quality evidence, and automation, and converts that information into repeatable and auditable practice. Eventually, those standardized records can also help establish “tacit” expertise. In effect, scaling and inspection will become foundational features for physical AI.
In a world where AI iterates in days, hardware and industrial capacity will have to catch up. Although South Korea might not have the advantages to become a frontrunner in developing foundation models, it is important for the country to focus on applying AI to improve its manufacturing capacity. In doing so, it can establish its reputation as a secure, traceable, and timely production node in the global supply chain. South Korea does not need to win the model race; it needs to lead in the next race towards model application. That is the path to relevance measured in generations, not quarters.
Revolutionizing the Industrial Base: South Korea’s AI Integration in Manufacturing
By Sean (Hyuk Hyun) Kwon
Korean Peninsula
While the global tech race focuses on who can build the smartest chatbots, South Korea is placing a bet on “Physical AI” to revolutionize manufacturing. By embedding intelligence directly into shipyards, tank factories, and semiconductor plants, the nation is shifting the competition from digital models to tangible industrial dominance. The move could prove to be the next phase of the AI revolution.
Editor’s Note: This commentary is the second of a two-part series examining how AI is being operationalized in shipbuilding, defense manufacturing, automotive production, and semiconductor fabrication.
By James Kim, Director, Korea Program
This is the second article in a two-part series, suggesting that South Korea’s approach to AI focuses less on leading foundation models and more on building long-lasting industrial capabilities using AI. The discussion examines how AI is integrating across key sectors, such as shipbuilding and the defense, automotive, and semiconductor industries. The author argues that AI, if leveraged properly, can become an industrial and alliance asset that can prevent bottlenecks and variance while making scale possible in the production process without sacrificing quality.
Case I: Shipbuilding
Shipbuilding is a challenging industry characterized by thousands of parts, long lead times, tight tolerances, and constant workforce constraints. In understanding how AI is being integrated into this process in South Korea, we should look at the choke points — designing, planning, and scheduling; welding and automation; and inspection and quality control. South Korean shipbuilding engineers are using AI-powered simulated hydrodynamics and digital twin technology to predict vessel performance in different scenarios, helping them refine hull designs and reduce fuel consumption. Integration of AI- powered robotics in precision welding and predictive maintenance can increase productivity by 30%. AI-powered supply chain management solutions can also reduce bottlenecking on procurement. HD Hyundai has announced that it is building a smart shipyard using these technologies, and that AI is being applied not only to vessel operation but also to shipbuilding processes.
This is an example of a concrete application in manufacturing where a shipyard can anticipate bottlenecks, detect defects earlier, and coordinate work sequences more efficiently. It is not just about cost saving. It allows companies to dictate tempo and increase scale, which are important advantages in the world of contested supply chains and maritime competition.
Case II: Defense Industry
South Korea aspires to be the fourth-largest defense producer by 2030. According to SIPRI, it is the second-largest defense supplier in Asia behind China. South Korean defense exports to Poland illustrate what this ambition looks like in practice. Within five months since signing the award, Poland received an initial shipment of its order that included 10 K2 tanks and 24 K9 howitzers — a strong signal about the South Korean defense industry’s production capacity during a moment when Europe’s demand curve is steepening.
For South Korean defense contractors, AI will be an important contributor to increasing surge capacity. Like shipbuilding, defense manufacturing is a system problem: Bottlenecks hide in sub-tier suppliers, quality variance compounds late in the process, and sustainment capacity becomes a limiting factor faster than most planners expect. AI will not replace vital components or labor. What it can do is reduce friction and variance across the workflow. Predictive maintenance, AI-assisted planning, and faster inspection cycles mean not sacrificing quality for scale and speed.
In a geopolitical environment that is growing increasingly uncertain, the implication for allies is straightforward: The South Korean defense industry is an alliance asset if the U.S. and ROK can figure out how to integrate AI into their manufacturing systems by harmonizing standards (e.g., inspection evidence and traceability), pairing scale with secure operational technology (OT), and building pathways for shared readiness.
Case III: Automobiles and Robotics
The South Korean automotive industry highlights the use of “physical AI” in manufacturing, where technologies like digital twins, robotics, the Internet of Things (IoT), and artificial intelligence are already integrated to enhance flexibility and resilience in production systems. Hyundai Motor Group’s Innovation Hub, for instance, uses these tools to link real-time factory data to virtual simulation so processes can be tested and improved before changes hit the assembly line. The key to this approach is the feedback loop. When AI is embedded into sequencing, inspection, and maintenance, it generates high-fidelity data on failure modes, rework patterns, cycle-time variance, and supplier-driven defects. That data becomes a capability. It improves throughput today, and it trains the next generation of industrial tools tomorrow.
Use of robotics allows Hyundai Motors to take this process to the next level. South Korea’s major industrial players are openly framing humanoids and advanced robots as the next frontier of factory work — particularly for repetitive and high-risk tasks. Hyundai announced plans to deploy Boston Dynamics’ Atlas humanoid robots in manufacturing starting in 2028, explicitly positioning this platform as part of a broader “physical AI” push. Meanwhile, South Korea’s manufacturing AI policy is also pointing toward industrial robotics as a target domain, including government plans tied to manufacturing AI transformation and industrial-use humanoids. The key concern is how South Korea can test scalable physical AI deployment while addressing concerns surrounding safety standards, data-sharing, and workforce transition. The implementation of the AI Basic Act in January of this year is the first important step in this regard.
Case IV: Semiconductor and AI Optimized Fabs
Semiconductors sit at the center of Korea’s AI strategy in two ways: Chips are the critical input for AI, and fabs are one of the most complex manufacturing environments where AI can contribute in meaningful ways. Samsung and NVIDIA plan on building fab-scale digital twins using NVIDIA Omniverse to support anomaly detection, predictive maintenance, and operational optimization before changes hit the physical line. This matters because fab performance is defined by variance management in this sector and affects the bottom line. Tiny shifts in the manufacturing process can cascade into yield loss, rework, and schedule disruption. AI-optimized fabs aim to compress that variance by improving equipment performance, tightening process control, and accelerating learning cycles.
In short, AI helps build chips more efficiently, and those chips help scale AI. If South Korea can couple fab-level AI optimization with secure, auditable production systems, it strengthens not just national competitiveness but allied supply chain resilience, especially as semiconductors become an important part of defense and industrial capacity.
The Data Dividend
The International Federation of Robotics reports that Korea is the world’s top adopter by robot density, at 1,012 robots per 10,000 employees. But robot density becomes a data advantage only if factories instrument and govern the data, and if SMEs can participate without crushing integration overhead. When properly implemented, AI deployments in inspection, routing, maintenance, or process control generate high-fidelity operational data. Production line managers can quickly identify what failed, what passed, why, and under what conditions. Over time, that data becomes an important national asset: It captures process parameters, routing decisions, quality evidence, and automation, and converts that information into repeatable and auditable practice. Eventually, those standardized records can also help establish “tacit” expertise. In effect, scaling and inspection will become foundational features for physical AI.
In a world where AI iterates in days, hardware and industrial capacity will have to catch up. Although South Korea might not have the advantages to become a frontrunner in developing foundation models, it is important for the country to focus on applying AI to improve its manufacturing capacity. In doing so, it can establish its reputation as a secure, traceable, and timely production node in the global supply chain. South Korea does not need to win the model race; it needs to lead in the next race towards model application. That is the path to relevance measured in generations, not quarters.
Recent & Related