AI is emerging in physical form beyond simple software at manufacturing sites. Now AI has entered the 'Physical AI' era where it moves in actual spaces, picks up objects, and converses with workers, going beyond analyzing data.
This trend that began in earnest from 2021 is completely changing the paradigm of manufacturing. Looking at the development status of Tesla's Optimus, Boston Dynamics' Spot, and Figure AI, you can feel how rapidly this change is progressing.
Global manufacturing companies are already moving quickly. Hyundai Motor Group invested 960 billion won to acquire Boston Dynamics, and Samsung Electronics and LG Electronics showcased next-generation robot products such as Ballie and smart home AI agents at CES 2024. According to Goldman Sachs' analysis, 62% of the humanoid market by 2035 is expected to be filled with household use, showing how rapidly this market is growing.
Let's examine three key strategies that manufacturing leaders should pay attention to amid these changes.
Manufacturing has been introducing automation and robot technology for decades, but Physical AI is bringing much more fundamental changes. Understanding the essence and impact of this change accurately is necessary to establish effective response strategies.
The development of AI technology can be broadly divided into four stages. NVIDIA CEO Jensen Huang defined these as Perception AI, Generative AI, Agentic AI, and finally Physical AI.
Early AI focused on recognition functions such as image recognition and voice recognition, then evolved into generative AI like ChatGPT. Currently, we have entered the Agentic AI stage that makes autonomous decisions, and now the Physical AI era where AI acts in the physical world is beginning in earnest.
Physical AI is not simply mounting AI on robots. This is a fundamental paradigm shift where AI perceives, understands, and physically interacts with the real world.
This change has become possible through the convergence of multiple technologies. The advancement of computer vision, natural language processing, sensor technology, robotics, and materials science has combined to lay the foundation for AI to operate in the physical world. Tesla's Optimus developing from walking to dancing level in just one year demonstrates the speed of this technological advancement.
The biggest change that Physical AI brings to manufacturing sites is the dramatic improvement in 'flexibility' and 'adaptability'. Existing industrial robots could only repeat predetermined tasks and required complex reprogramming when changing operations.
However, Physical AI robots can recognize environments in real-time and adjust their actions according to situations.
For example, a Physical AI robot working on an automotive parts assembly line can recognize and appropriately respond even when parts' positions or orientations change. Furthermore, when new work instructions are given in natural language, it can understand and execute them.
This flexibility perfectly aligns with modern manufacturing trends such as high-variety low-volume production and customized manufacturing. In situations where consumer demands change rapidly and product lifecycles shorten, Physical AI provides the ability to quickly reconfigure production lines and switch to new products.
While existing automation systems focused on efficiency, Physical AI provides both efficiency and flexibility simultaneously. This means a fundamental change in manufacturing competitiveness.
Global manufacturing companies are already attempting various strategies for Physical AI adoption. Analyzing their approaches reveals three main strategies.
The first is self-development strategy. Tesla is developing Optimus based on its AI technology and manufacturing know-how. Elon Musk revealed the goal of "mass-producing humanoid robots priced under $20,000," pursuing both manufacturing innovation and cost reduction through Physical AI.
Tesla has a vision to not only automate repetitive and dangerous tasks in factories through Optimus but also expand to household and service areas in the long term.
The second is capability acquisition through M&A. Hyundai Motor Group secured top-level robot technology through the acquisition of Boston Dynamics. This was a strategic decision considering business expansion not only in automobile manufacturing but also future mobility services.
Hyundai utilizes Boston Dynamics' 'Spot' for factory safety inspections while pursuing a 'mobility + robotics' strategy that fuses autonomous driving technology with robot technology.
LG Electronics is also expanding investment in the robot market, emphasizing "robots are a newly focused area. We will watch development directions and keep open possibilities for equity investment and M&A."
The third is open innovation and partnership strategy. Samsung Electronics and LG Electronics are accelerating robot development through cooperation with AI companies. Particularly, LG Electronics is taking an approach to strengthen AI functions by forming strategic partnerships with Microsoft. Such partnerships have advantages of faster market entry and risk distribution compared to self-development.
These global cases show the importance of strategic choice suitable for each company's situation and capabilities in Physical AI adoption. Which strategy is optimal among self-development, M&A, and partnerships can vary depending on the company's technological capabilities, financial strength, and business model.
Successful adoption of Physical AI depends not simply on robot technology itself, but on building systems where humans and robots can collaborate effectively. This means comprehensive redesign of organization, space, and processes beyond technical aspects.
The goal of Physical AI adoption lies in 'augmentation' rather than 'replacement' of human labor. Building models where robots and humans create synergy by utilizing their respective strengths is necessary for successful adoption.
Humans excel at creative problem-solving, complex judgment, and responding to exceptional situations, while robots have strengths in precise and repetitive tasks, work in dangerous environments, and processing large amounts of data. Collaboration models utilizing these complementary characteristics are needed.
BMW's US factory is building a system where Figure AI robots work collaboratively with human workers. Robots handle heavy parts transportation and precise assembly, while human workers handle quality inspection and complex decision-making. Through this collaboration, they achieve effects of simultaneously improving productivity and quality. After introducing Figure 02, BMW achieved 4 times faster work speed and 7 times improved reliability as of November 2024, enabling processing of up to 1,000 tasks per day.
An important point in designing collaboration models is not simple division of work processes, but creating structures where humans and robots can continuously interact, learn together, and develop. This approach enables continuous process improvement by utilizing Physical AI's learning capabilities.
For humans and robots to work in the same space, fundamental redesign of work environments is necessary. While existing industrial robots were isolated with safety fences, Physical AI robots operate in the same space as humans.
Safety is the most basic yet important factor. Worker safety must be ensured through collision prevention sensors, torque limiting mechanisms, and real-time environment recognition functions.
Space arrangement is also an important consideration. Robot work areas, human worker movement paths, and material transportation routes must be optimized to maximize efficiency.
Ergonomic design is also important. Work desk heights, control panel positions, visual feedback systems, etc. must be ergonomically designed so human workers can interact comfortably with robots. This reduces worker fatigue and contributes to long-term productivity improvement.
Physical work environment design must balance safety, efficiency, and ergonomics. This is a task that must be completed before technology adoption.
The ultimate goal of Physical AI adoption lies in innovation of the business model itself beyond simple productivity improvement. This means fundamental changes not only in product manufacturing methods but also in ways of delivering value to customers.
Physical AI plays an important role in manufacturing companies transitioning from product sales-centered to service provision-centered business models. The so-called 'Servitization' strategy is becoming a key element for enhancing manufacturing profitability and sustainability.
A representative example of service-based models is the 'Product-Service System (PSS)'. This is a model that provides various services throughout the product lifecycle rather than ending with product sales. Physical AI enables various services such as remote monitoring, predictive maintenance, and performance optimization.
German industrial equipment manufacturer TRUMPF innovated its business model with a 'Pay-per-Part' service using Physical AI. Instead of purchasing equipment itself, customers pay according to the number of parts produced, while TRUMPF remotely monitors and optimizes equipment through AI robots to increase customer productivity.
Through this model, TRUMPF succeeded in securing fixed revenue while building long-term relationships with customers, raising service revenue proportion to 30% of the total.
The core of service-based models is directly connecting customer performance with manufacturer revenue. This aligns interests between manufacturers and customers, enabling continuous value creation.
Service-based models provide various benefits including stable revenue flow, long-term relationship building with customers, and competitive differentiation. Moreover, data collected during continuous service provision is utilized for product improvement and new service development, leading to additional value creation.
One of the most powerful characteristics of Physical AI is real-time data collection and analysis capability. This fundamentally changes manufacturing companies' decision-making processes and opens new dimensions of data-based operations.
In traditional manufacturing environments, data collection was limited and analysis took time, making reactive decision-making common. However, in Physical AI environments, vast amounts of data are collected and analyzed in real-time from production facilities, robots, workers, and products, enabling predictive and even prescriptive decision-making.
The evolution of data-based decision-making progresses through stages of 'What happened? (descriptive analysis)', 'Why did it happen? (diagnostic analysis)', 'What will happen? (predictive analysis)', and 'What should be done? (prescriptive analysis)'. Physical AI integrates all these stages in real-time to enable autonomous decision-making.
Particularly in core business decisions such as demand forecasting, the combination of Physical AI and data analysis technology can bring remarkable results. Companies like ImpactiveAI's DeepFlow solution analyze over 6 million data points to provide 98.6% high prediction accuracy, and when linked with Physical AI, autonomous production planning and execution beyond prediction becomes possible.
GE's smart factory combined Physical AI and data analysis to predict equipment failures in advance and automatically establish optimal maintenance schedules, improving equipment utilization by over 20%. Such cases show that data-based decision-making can strengthen fundamental manufacturing competitiveness beyond simple efficiency improvement.
Such data-based operations contribute to enhancing visibility and resilience of entire supply chains beyond simple efficiency improvement. This will be an important competitive element for manufacturing companies that recently experienced global supply chain crises.
The changes that Physical AI will bring to manufacturing have already begun. Companies leading this change are innovating business models and organizational capabilities together beyond simple technology adoption.
Now is the time for Korean manufacturing companies to recognize the potential of Physical AI and secure future competitiveness through strategic approaches suitable for their situations.
.svg)
