Along with the launch of the M.AX Alliance last September, the Electronic Times began its special series, 'Seize the Golden Time for Manufacturing AX (M.AX)'. The series has sequentially examined the possibilities and limitations our manufacturing industry faces on the threshold of Artificial Intelligence Transformation (AX) across sectors including semiconductors, robotics/humanoids, autonomous driving, ai factories, shipbuilding, home appliances, defense/drones, and retail.
The M.AX roundtable discussion was held on the 19th at the Korea Evaluation Institute of Industrial Technology in Jung-gu, Seoul. From left: Byung-hoon Song, Head of the Autonomous Manufacturing Research Center at the Korea Electronics Technology Institute; byeng dong youn, CEO of onepredict; Min-woo Lee, Director General for Industrial Policy at MOTIE; and Su-ho Lee, Executive Vice President of Ecopro. Photo by Dong-geun Lee foto@etnews.com
To conclude the series, a roundtable was organized, bringing together key stakeholders from the government, research institutes, and both supply and demand companies. The discussion assessed the current level and future tasks of M.AX and deliberated on the direction for the next phase.
During the roundtable, a consensus was reached that manufacturing AX is not merely an issue of technology adoption, but a structural transformation that requires the simultaneous redesign of data, workforce, regulations, and governance. While successful cases of AX settling in are emerging in some leading companies and processes, participants agreed that in order to spread this across the entire industry, tasks such as revamping data utilization structures, improving systems focusing on empirical demonstrations, and enhancing the AI utilization capabilities of the field workforce still remain.
The roundtable was attended by Min-woo Lee, Director General for Industrial Policy at the Ministry of Trade, Industry and Energy (MOTIE); Byung-hoon Song, Head of the Autonomous Manufacturing Research Center at the Korea Electronics Technology Institute (KETI); byeng dong youn, CEO of onepredict (Professor at the Department of Mechanical Engineering, Seoul National University); and Su-ho Lee, Executive Vice President of Ecopro. The discussion was moderated by Young-guk Ahn, Deputy Head of the Political Policy Department at the Electronic Times.
From left: Byung-hoon Song, Head of KETI Center; byeng dong youn, CEO of onepredict (Professor at SNU); Min-woo Lee, Director General for Industrial Policy at MOTIE; and Su-ho Lee, Executive Vice President of Ecopro.
〈Attendees (in alphabetical order)〉
△ Byung-hoon Song, Head of Center at KETI
△ byeng dong youn, CEO of onepredict (Professor at SNU)
△ Min-woo Lee, Director General for Industrial Policy at MOTIE
△ Su-ho Lee, Executive Vice President of Ecopro
The M.AX roundtable discussion was held on the 19th at the Korea Evaluation Institute of Industrial Technology in Jung-gu, Seoul. Clockwise from left: Min-woo Lee, Director General for Industrial Policy at MOTIE; Byung-hoon Song, Head of the Autonomous Manufacturing Research Center at KETI; byeng dong youn, CEO of onepredict; and Su-ho Lee, Executive Vice President of Ecopro. Photo by Dong-geun Lee foto@etnews.com
◇ Moderator (Young-guk Ahn, Deputy Head at Electronic Times) = Why did the Ministry of Trade, Industry and Energy set 'M.AX' as a core strategy among various policies this year?
◇ Min-woo Lee, Director General for Industrial Policy at MOTIE = Because it is not an option, but a necessity. Our manufacturing industry has allowed rapid pursuit by competitor nations, and in some sectors, we are seeing situations where our competitiveness falls behind emerging countries like China. Conversely, in the fields of fine chemicals, advanced materials, and precision machinery, the reality is that we still cannot catch up with Japan and Germany. It is an expression of desperation amidst this situation. The government believes that it is impossible for the manufacturing industry to survive and leap forward again without AI transformation. In fact, during a recent presidential briefing, the AI transformation of manufacturing was reported as the top priority for the Ministry, and I stated that we would mobilize all our capabilities to actively support it.
The reason we set M.AX as a core strategy is because it is a concept that bundles various policies and strategies for manufacturing AI transformation into a single value. The intent is to bring together over 1,000 companies, institutions, academia, and research circles to pool our strength and wisdom for manufacturing AI transformation. Until now, manufacturing companies did not know who to ask for help even if they wanted to pursue AI transformation, and AI supply companies found it difficult to establish touchpoints with the manufacturing sector. To bridge this gap, M.AX was promoted to bring everyone together, expand the manufacturing AI ecosystem, and create new business and win-win models. The core of M.AX lies in various players collaborating as a single team to create new added value.
◇ Moderator = Compared to competing countries, do you view the speed of our manufacturing AX as fast?
◇ Byung-hoon Song, Head of Center at KETI = The relationship with the US, Germany, and Japan—with whom we have compared manufacturing competitiveness—is changing. Chinese products, which in the past had an image of being low-cost, have now significantly improved their price-performance ratio, reaching a stage where they disrupt the market. Germany has strengths in completeness and standards centered on facilities and equipment, while the US, though its manufacturing base is weak, is leading in software and AI technology. Concerns follow as to whether Korea can independently design AI from scratch like the US, or push equipment completeness to the extreme like Germany. Therefore, I believe the strategy Korea should adopt is to rapidly mass-produce intelligent products by combining AI with our well-proven 'fast commercialization.'
Since the past, Korea has swiftly attempted IoT home appliances and convergent products. The same goes for AI. It is important to quickly bring products equipped with AI, such as refrigerators, washing machines, and hair dryers, to the market. Even if we cannot lead in all cutting-edge AI technologies, a method that maintains product completeness while rapidly applying AI is suitable for Korea. In that respect, the M.AX strategy is one that maintains completeness while accelerating, and currently, I believe we are not late and are accelerating well.
◇ Moderator = From the perspective of a supply company, do you view now as the 'golden time'?
◇ byeng dong youn, CEO of AI transformation cannot start without any foundation. Fortunately, Korea has a solid base of manufacturing companies, a skilled workforce with 20 to 30 years of field experience, and data accumulated by pushing digital transformation over the past 20 years. Right now, when these three elements exist simultaneously, is the timing. If we miss this period, the specialized field workforce might disappear, or factories might relocate overseas. Now is the exact moment we must capitalize on our strengths as assets.
◇ Moderator = In actual manufacturing sites, where do you think it is effective to start AX?
◇ Su-ho Lee, Executive Vice President of Ecopro = Ultimately, the core is data. We have promoted smart factories and equipment automation, but the perspective of data integration was lacking. AI's fuel is data, and if data is fragmented, AI cannot operate properly either. In the past, there were many cases where AI adoption was halted if partial implementation failed, but now it is a matter of survival.
We approached it centering on quality. The secondary battery cathode material industry is a materials industry where quality measurement is difficult, so predicting it with AI was urgent. With support from the Ministry of Trade, Industry and Energy, we created a quality prediction model for the firing process up to an 80% level in the first year, and we are aiming for over 90% this year. The important point is that you cannot do it alone. By combining our digital transformation (DX) experience with AI, systematizing the data, and then integrating the AI model, we saw potential.
◇ Moderator = What do you think is practically needed for AX to spread to small and medium-sized enterprises (SMEs)?
◇ Min-woo Lee, Director General for Industrial Policy = Expanding to SMEs is a very difficult task. During the past DX transition period, I visited the field often. While large corporations achieved results by applying AI to their entire processes, even just visiting primary subcontractors, there were many cases where they hesitated to invest due to the uncertainty of Return on Investment (ROI).
The major premise of AI is data, but SMEs lack the personnel and technology for data preprocessing, collection standards, and standardization. So, we intend to approach this in two directions. First is a cooperative model where large corporations and SMEs jointly develop AI models. Based on high-quality manufacturing data held by large corporations, industry-specific AI models are created and shared collaboratively by SME partners. Second is AI support tailored for SMEs. We plan to support data preprocessing and low-code-based AI model development, and provide consulting through regional AI cooperation centers.
◇ Moderator = In what areas is AX being commercialized most rapidly, and what is the opposite case?
◇ byeng dong youn, CEO of at commercialization from a product perspective, AI agents are spreading the fastest. Quality inspection, quality prediction, predictive maintenance, energy optimization, and management are mostly being implemented in the form of AI agents. However, currently, the majority are problem-specific AI models. A model created for a specific company or a specific problem is difficult to apply directly to another company. To overcome this, foundation models specialized for industries and tasks are needed.
The biggest obstacle is data. Foundation models require data at the level of tens of billions of parameters. I believe we need attempts to gather data at the value chain level or industrial complex level, while acknowledging the data silos between companies and factories.
The issue of data sharing has been continuously discussed since the early days of smart factories. Data is extremely complex and abstract. Therefore, clear goals like a purpose-driven foundation model are needed. Our strength is that we have a skilled workforce and field data. If we rapidly structure this data and connect it with AI, I believe we have a chance to win.
◇ Min-woo Lee, Director General for Industrial Policy = We will actively reflect this in policy. From January 1st next year, the Ministry will establish a new policy bureau related to AI. It is an organizational restructuring to integrate scattered AI tasks and launch manufacturing AI transformation in earnest. We are currently operating 10 alliances, focusing on creating a vertically integrated data ecosystem. We will also expand budget and personnel support with the goal of building 500 ai factories by 2030.
◇ Su-ho Lee, Executive Vice President = About 30 terabytes (TB) of data is accumulated monthly. Management, quality, and equipment data were fragmented, but we are currently building an integrated data lake. We are approaching this with a 'Mother Factory' strategy, piloting one line to create a success case and then expanding it. What is important is not just accumulating data, but creating a system that can interpret meaningful data. Furthermore, security and safety regulations at manufacturing sites are also factors that make AI application difficult. Depending on the importance of data, flexible governance and special regulatory exemptions exclusive for AX are necessary.
◇ Moderator = Do you have any proposals for the success of manufacturing AX?
◇ Byung-hoon Song, Head of Center = There is a concept called the 'Golden Line Project' that Apple talks about these days. Apple manages the baseline of their production lines—the so-called 'Golden Baseline'—almost obsessively at sites including Foxconn. Beyond simple equipment levels, they scrutinize the internal automation level and process optimization down to code units and simulation stages, including software and AI. The data accumulated in that way leads to AI.
The core of the golden project is to meticulously build one line and then spread it to the entire line. In our 'ppalli-ppalli' (hurry-hurry) culture, we have often glossed over things even if digital data was missing, but now I believe we need to spread such standards starting from specific industries and support them at the policy level.
◇ Su-ho Lee, Executive Vice President = We must push forward by bundling AI technology and the data industry, and that is why M.AX was created. Now, the competitors for domestic companies are not other domestic companies. They are global companies overseas. A mother line is completed when production, quality, equipment, and operations management are all equipped, and AI infrastructure and solutions are layered on top.
The key is how far we can open and share data. Areas that can be shared should be put on a platform, and areas of super-gap should be taken by each company to create synergy.
The core of a manufacturing company is ultimately cost competitiveness. We must combine AI-based manufacturing competitiveness with raw material competitiveness to create a technological super-gap.
◇ byeng dong youn, CEO of concept of a golden line has existed for a long time, but it is gaining attention again as AI is combined with it. In manufacturing, AI can play a role when DX has reached a certain level. AI is not everything; it is a tool. Taking this opportunity, I hope we review the entire manufacturing structure including AI, create standards like the golden project, and connect them to collaborative businesses. This is not an issue of individual companies, but a national agenda.
Particularly in the manufacturing sector, there is no data governance system at the national level yet. Standards on what data is important, and how to manage and utilize it according to that importance, have not been established. Even if standards and security are handled at the corporate level, the design of governance at the government level is necessary, at least regarding the rules and structures of data utilization. I hope we leverage this opportunity well so that Korea can stand in a much more unrivaled position as a manufacturing powerhouse around 2035 than it is now.
Arranged by Sung-woo Cho
...
...
...
...
Along with the launch of the M.AX Alliance last September, the Electronic Times began its special series, 'Seize the Golden Time for Manufacturing AX (M.AX)'. The series has sequentially examined the possibilities and limitations our manufacturing industry faces on the threshold of Artificial Intelligence Transformation (AX) across sectors including semiconductors, robotics/humanoids, autonomous driving, ai factories, shipbuilding, home appliances, defense/drones, and retail.
The M.AX roundtable discussion was held on the 19th at the Korea Evaluation Institute of Industrial Technology in Jung-gu, Seoul. From left: Byung-hoon Song, Head of the Autonomous Manufacturing Research Center at the Korea Electronics Technology Institute; byeng dong youn, CEO of onepredict; Min-woo Lee, Director General for Industrial Policy at MOTIE; and Su-ho Lee, Executive Vice President of Ecopro. Photo by Dong-geun Lee foto@etnews.com
To conclude the series, a roundtable was organized, bringing together key stakeholders from the government, research institutes, and both supply and demand companies. The discussion assessed the current level and future tasks of M.AX and deliberated on the direction for the next phase.
During the roundtable, a consensus was reached that manufacturing AX is not merely an issue of technology adoption, but a structural transformation that requires the simultaneous redesign of data, workforce, regulations, and governance. While successful cases of AX settling in are emerging in some leading companies and processes, participants agreed that in order to spread this across the entire industry, tasks such as revamping data utilization structures, improving systems focusing on empirical demonstrations, and enhancing the AI utilization capabilities of the field workforce still remain.
The roundtable was attended by Min-woo Lee, Director General for Industrial Policy at the Ministry of Trade, Industry and Energy (MOTIE); Byung-hoon Song, Head of the Autonomous Manufacturing Research Center at the Korea Electronics Technology Institute (KETI); byeng dong youn, CEO of onepredict (Professor at the Department of Mechanical Engineering, Seoul National University); and Su-ho Lee, Executive Vice President of Ecopro. The discussion was moderated by Young-guk Ahn, Deputy Head of the Political Policy Department at the Electronic Times.
From left: Byung-hoon Song, Head of KETI Center; byeng dong youn, CEO of onepredict (Professor at SNU); Min-woo Lee, Director General for Industrial Policy at MOTIE; and Su-ho Lee, Executive Vice President of Ecopro.
〈Attendees (in alphabetical order)〉
△ Byung-hoon Song, Head of Center at KETI
△ byeng dong youn, CEO of onepredict (Professor at SNU)
△ Min-woo Lee, Director General for Industrial Policy at MOTIE
△ Su-ho Lee, Executive Vice President of Ecopro
The M.AX roundtable discussion was held on the 19th at the Korea Evaluation Institute of Industrial Technology in Jung-gu, Seoul. Clockwise from left: Min-woo Lee, Director General for Industrial Policy at MOTIE; Byung-hoon Song, Head of the Autonomous Manufacturing Research Center at KETI; byeng dong youn, CEO of onepredict; and Su-ho Lee, Executive Vice President of Ecopro. Photo by Dong-geun Lee foto@etnews.com
◇ Moderator (Young-guk Ahn, Deputy Head at Electronic Times) = Why did the Ministry of Trade, Industry and Energy set 'M.AX' as a core strategy among various policies this year?
◇ Min-woo Lee, Director General for Industrial Policy at MOTIE = Because it is not an option, but a necessity. Our manufacturing industry has allowed rapid pursuit by competitor nations, and in some sectors, we are seeing situations where our competitiveness falls behind emerging countries like China. Conversely, in the fields of fine chemicals, advanced materials, and precision machinery, the reality is that we still cannot catch up with Japan and Germany. It is an expression of desperation amidst this situation. The government believes that it is impossible for the manufacturing industry to survive and leap forward again without AI transformation. In fact, during a recent presidential briefing, the AI transformation of manufacturing was reported as the top priority for the Ministry, and I stated that we would mobilize all our capabilities to actively support it.
The reason we set M.AX as a core strategy is because it is a concept that bundles various policies and strategies for manufacturing AI transformation into a single value. The intent is to bring together over 1,000 companies, institutions, academia, and research circles to pool our strength and wisdom for manufacturing AI transformation. Until now, manufacturing companies did not know who to ask for help even if they wanted to pursue AI transformation, and AI supply companies found it difficult to establish touchpoints with the manufacturing sector. To bridge this gap, M.AX was promoted to bring everyone together, expand the manufacturing AI ecosystem, and create new business and win-win models. The core of M.AX lies in various players collaborating as a single team to create new added value.
◇ Moderator = Compared to competing countries, do you view the speed of our manufacturing AX as fast?
◇ Byung-hoon Song, Head of Center at KETI = The relationship with the US, Germany, and Japan—with whom we have compared manufacturing competitiveness—is changing. Chinese products, which in the past had an image of being low-cost, have now significantly improved their price-performance ratio, reaching a stage where they disrupt the market. Germany has strengths in completeness and standards centered on facilities and equipment, while the US, though its manufacturing base is weak, is leading in software and AI technology. Concerns follow as to whether Korea can independently design AI from scratch like the US, or push equipment completeness to the extreme like Germany. Therefore, I believe the strategy Korea should adopt is to rapidly mass-produce intelligent products by combining AI with our well-proven 'fast commercialization.'
Since the past, Korea has swiftly attempted IoT home appliances and convergent products. The same goes for AI. It is important to quickly bring products equipped with AI, such as refrigerators, washing machines, and hair dryers, to the market. Even if we cannot lead in all cutting-edge AI technologies, a method that maintains product completeness while rapidly applying AI is suitable for Korea. In that respect, the M.AX strategy is one that maintains completeness while accelerating, and currently, I believe we are not late and are accelerating well.
◇ Moderator = From the perspective of a supply company, do you view now as the 'golden time'?
◇ byeng dong youn, CEO of AI transformation cannot start without any foundation. Fortunately, Korea has a solid base of manufacturing companies, a skilled workforce with 20 to 30 years of field experience, and data accumulated by pushing digital transformation over the past 20 years. Right now, when these three elements exist simultaneously, is the timing. If we miss this period, the specialized field workforce might disappear, or factories might relocate overseas. Now is the exact moment we must capitalize on our strengths as assets.
◇ Moderator = In actual manufacturing sites, where do you think it is effective to start AX?
◇ Su-ho Lee, Executive Vice President of Ecopro = Ultimately, the core is data. We have promoted smart factories and equipment automation, but the perspective of data integration was lacking. AI's fuel is data, and if data is fragmented, AI cannot operate properly either. In the past, there were many cases where AI adoption was halted if partial implementation failed, but now it is a matter of survival.
We approached it centering on quality. The secondary battery cathode material industry is a materials industry where quality measurement is difficult, so predicting it with AI was urgent. With support from the Ministry of Trade, Industry and Energy, we created a quality prediction model for the firing process up to an 80% level in the first year, and we are aiming for over 90% this year. The important point is that you cannot do it alone. By combining our digital transformation (DX) experience with AI, systematizing the data, and then integrating the AI model, we saw potential.
◇ Moderator = What do you think is practically needed for AX to spread to small and medium-sized enterprises (SMEs)?
◇ Min-woo Lee, Director General for Industrial Policy = Expanding to SMEs is a very difficult task. During the past DX transition period, I visited the field often. While large corporations achieved results by applying AI to their entire processes, even just visiting primary subcontractors, there were many cases where they hesitated to invest due to the uncertainty of Return on Investment (ROI).
The major premise of AI is data, but SMEs lack the personnel and technology for data preprocessing, collection standards, and standardization. So, we intend to approach this in two directions. First is a cooperative model where large corporations and SMEs jointly develop AI models. Based on high-quality manufacturing data held by large corporations, industry-specific AI models are created and shared collaboratively by SME partners. Second is AI support tailored for SMEs. We plan to support data preprocessing and low-code-based AI model development, and provide consulting through regional AI cooperation centers.
◇ Moderator = In what areas is AX being commercialized most rapidly, and what is the opposite case?
◇ byeng dong youn, CEO of at commercialization from a product perspective, AI agents are spreading the fastest. Quality inspection, quality prediction, predictive maintenance, energy optimization, and management are mostly being implemented in the form of AI agents. However, currently, the majority are problem-specific AI models. A model created for a specific company or a specific problem is difficult to apply directly to another company. To overcome this, foundation models specialized for industries and tasks are needed.
The biggest obstacle is data. Foundation models require data at the level of tens of billions of parameters. I believe we need attempts to gather data at the value chain level or industrial complex level, while acknowledging the data silos between companies and factories.
The issue of data sharing has been continuously discussed since the early days of smart factories. Data is extremely complex and abstract. Therefore, clear goals like a purpose-driven foundation model are needed. Our strength is that we have a skilled workforce and field data. If we rapidly structure this data and connect it with AI, I believe we have a chance to win.
◇ Min-woo Lee, Director General for Industrial Policy = We will actively reflect this in policy. From January 1st next year, the Ministry will establish a new policy bureau related to AI. It is an organizational restructuring to integrate scattered AI tasks and launch manufacturing AI transformation in earnest. We are currently operating 10 alliances, focusing on creating a vertically integrated data ecosystem. We will also expand budget and personnel support with the goal of building 500 ai factories by 2030.
◇ Su-ho Lee, Executive Vice President = About 30 terabytes (TB) of data is accumulated monthly. Management, quality, and equipment data were fragmented, but we are currently building an integrated data lake. We are approaching this with a 'Mother Factory' strategy, piloting one line to create a success case and then expanding it. What is important is not just accumulating data, but creating a system that can interpret meaningful data. Furthermore, security and safety regulations at manufacturing sites are also factors that make AI application difficult. Depending on the importance of data, flexible governance and special regulatory exemptions exclusive for AX are necessary.
◇ Moderator = Do you have any proposals for the success of manufacturing AX?
◇ Byung-hoon Song, Head of Center = There is a concept called the 'Golden Line Project' that Apple talks about these days. Apple manages the baseline of their production lines—the so-called 'Golden Baseline'—almost obsessively at sites including Foxconn. Beyond simple equipment levels, they scrutinize the internal automation level and process optimization down to code units and simulation stages, including software and AI. The data accumulated in that way leads to AI.
The core of the golden project is to meticulously build one line and then spread it to the entire line. In our 'ppalli-ppalli' (hurry-hurry) culture, we have often glossed over things even if digital data was missing, but now I believe we need to spread such standards starting from specific industries and support them at the policy level.
◇ Su-ho Lee, Executive Vice President = We must push forward by bundling AI technology and the data industry, and that is why M.AX was created. Now, the competitors for domestic companies are not other domestic companies. They are global companies overseas. A mother line is completed when production, quality, equipment, and operations management are all equipped, and AI infrastructure and solutions are layered on top.
The key is how far we can open and share data. Areas that can be shared should be put on a platform, and areas of super-gap should be taken by each company to create synergy.
The core of a manufacturing company is ultimately cost competitiveness. We must combine AI-based manufacturing competitiveness with raw material competitiveness to create a technological super-gap.
◇ byeng dong youn, CEO of concept of a golden line has existed for a long time, but it is gaining attention again as AI is combined with it. In manufacturing, AI can play a role when DX has reached a certain level. AI is not everything; it is a tool. Taking this opportunity, I hope we review the entire manufacturing structure including AI, create standards like the golden project, and connect them to collaborative businesses. This is not an issue of individual companies, but a national agenda.
Particularly in the manufacturing sector, there is no data governance system at the national level yet. Standards on what data is important, and how to manage and utilize it according to that importance, have not been established. Even if standards and security are handled at the corporate level, the design of governance at the government level is necessary, at least regarding the rules and structures of data utilization. I hope we leverage this opportunity well so that Korea can stand in a much more unrivaled position as a manufacturing powerhouse around 2035 than it is now.
Arranged by Sung-woo Cho