Domestic IT service corporations once called the "computer rooms of conglomerates" have moved to secure the physical artificial intelligence (AI) market applied to industrial sites. While manufacturers handle the hardware of physical AI, such as robots and large equipment, IT service corporations are seen as strong in the data and robot/AI platforms that serve as the brains of next-generation factories.
According to the related industry on the 3rd, as competition among leading domestic manufacturers such as Samsung, LG, SK, and POSCO intensifies to seize leadership in the future robot market, the IT service corporations under these groups are also making Robotics and physical AI their next growth engines and expanding related investment. In particular, they are accelerating development of a factory operating system (OS) in which robots and equipment at industrial sites make their own work judgments and efficiently share tasks with people, and, based on this, building autonomous "intelligent factories."
In an AI-based "intelligent factory," various robots must be managed and operated within a single integrated system, and robots must be connected to production equipment, factory control systems, and data platforms to optimize process flow. IT service corporations plan to develop an operating system that enables this by leveraging production-site data built on more than 20 years of experience implementing group manufacturing systems.
POSCO DX laid out a blueprint to control key processes at steel mills with AI and build an "intelligent factory" in which large equipment operates on its own without people. The company is focusing on converting the tacit knowledge held by veteran workers nearing retirement into core data for physical AI to automate and autonomize high-difficulty repetitive tasks, and on building a manufacturing AI-based system in which AI handles the judgment and control previously performed by people.
Currently, at the Gwangyang and Pohang steel mills, large equipment is used for each core process, and most tasks are highly complex and dangerous, so the goal is to make them unmanned to boost on-site safety and productivity. POSCO DX is pushing "equipment robotization" that combines physical AI with equipment weighing tens of tons—such as large cranes that move steel products, reclaimers that transfer raw materials in yards (work sites), and ship unloaders for raw materials—so they judge and control themselves. This equipment is now in a trial run phase and could be commercialized as early as the second half of this year for introduction at the steel mills.
SK AX said last month it will ramp up its "Manufacturing RX full-stack service" to shift manufacturing sites to a robot-centric model. By combining a Digital Twin (a technology that implements real-world spaces in virtual spaces) with physical AI, the goal is to realize an autonomous manufacturing factory in which the entire plant makes its own decisions and moves organically. Before introducing robots, the service verifies risk factors in a virtual environment, and in the factory it consolidates robots with the manufacturing execution system (MES) and equipment data in real time to provide integrated control of the entire plant.
An SK AX official said, "We are verifying field data accumulation, related systems, and proof-of-concept models in the semiconductor industry, and will expand this to the shipbuilding industry."
LG CNS in May unveiled PhysicalWorks, a robot transformation (RX) platform that manages the entire cycle from robot training to integrated control. The goal is to quickly train intelligent robots with AI and shorten the on-site deployment time for industrial robots, which used to take several months, to one to two months. It also helps manufacturers operate different robots and various types of robots within a single system.
The company plans to apply PhysicalWorks to domestic manufacturing and logistics sites to advance it, then expand to other fields. It is currently conducting or has completed robot proofs of concept (PoC) with more than 20 customers in key industries, including LX Pantos and Kurly.
Recently, it signed a 189.7 billion won contract with LG Electronics to build physical AI infrastructure and will supply PhysicalWorks and GPU-based AI infrastructure. The two companies plan Korea's first large-scale intelligent robot training infrastructure, the "Robot Data Factory."
An LG CNS official said, "Investment in smart factories and robots is not a short-term expansion of production facilities, but a strategic investment to address rising labor costs, shortages of production workers, and stronger quality and safety standards," adding, "Our goal is to set a new standard for commercialization of physical AI and ultimately realize a robot-centric autonomous operating system."
Samsung SDS is also preparing a platform that controls robots deployed at industrial sites. Earlier, Samsung Electronics announced it would convert all domestic and overseas production plants into "AI autonomous factories" by 2030, fueling expectations that Samsung SDS will play a key role in building the factories' "AI brains."
As part of its strategy to strengthen Robotics and physical AI businesses, Samsung SDS made a strategic investment last month in U.S.-based Walden Robotics through Samsung Venture Investment. On a conference call after last month's earnings release, Samsung SDS said, "Starting with the Walden Robotics investment, we will expand cooperation with global robot corporations and push a 'robot orchestration' business that integrates and controls various robots and links them with the manufacturing execution system."