SK Telecom plans to introduce the first version of its digital twin-based robot training platform in the second half of this year and begin full-scale commercialization in 2027. The company aims to build an enterprise platform that continuously trains a wide range of robots using customer-specific manufacturing data, with a long-term vision of expanding the service through its AI Data Center (AIDC) infrastructure.
As of July 13, SK Telecom is conducting Physical AI pilot programs with several manufacturing affiliates within the SK Group, as well as external companies.
The pilots focus on training robots to perform tasks traditionally carried out by human workers--such as transporting materials and delivering components in manufacturing and logistics facilities--and validating their performance in real-world industrial environments.
“We expect to complete the first version of our digital twin-based robot training platform in the second half of this year and introduce it to enterprise customers,” said Cho Ik-hwan, head of Physical AI at SK Telecom. “Starting next year, we plan to move into full-scale commercialization based on the platform.”
The platform is designed to integrate the entire robot development workflow, from defining tasks and collecting training data to robot training, simulation, and real-world validation. Its ultimate goal is to enable enterprises to continuously improve robot performance by leveraging their own operational data.
SK Telecom plans to expand the business in phases. It will first build reference cases by deploying digital twin and robotics technologies in highly complex manufacturing environments, such as semiconductor fabrication plants. The company will then evolve the platform into an enterprise solution that allows customers to continuously refine both robots and AI models using proprietary data.
A key example is SK hynix's semiconductor fabrication facilities. Using NVIDIA Omniverse, SK Telecom has recreated semiconductor manufacturing environments in a digital twin and optimized the platform for large-scale industrial operations. The company plans to broaden deployments based on the experience gained in these demanding manufacturing settings.
Over the longer term, SK Telecom intends to expand the platform into an AIDC-based service that provides robot training and simulation capabilities through its AI data centers. Companies without dedicated AI infrastructure would be able to use SK Telecom's computing resources and platform to train robots and validate their performance before deployment.
“Our long-term goal is to offer the platform on top of our AIDC much like a public cloud service,” Cho said. “Companies without their own AI infrastructure will be able to use SK Telecom's infrastructure and services to train robots as part of our phased business strategy.”
SK Telecom is also designing the platform to be hardware-agnostic, allowing it to support robots from various manufacturers, including collaborative robots and humanoids. It will also be compatible with robot foundation models developed by both domestic and international companies.
The company aims to make Physical AI a major revenue driver for its B2B AI business within the next three years. “Physical AI is still an emerging market,” Cho said. “From a business perspective, our goal is for it to account for a double-digit share of SK Telecom's B2B AI revenue within three years. Right now, we're focused on validating the technology and building the platform.”
