Intel 擴大在南韓的實體 AI 合作

John Healy, Intel vice president, listens to reporters’ questions. (Photo: THE ELEC)

Intel plans to significantly expand cooperation with South Korean companies and the government, positioning the country as a hub for physical artificial intelligence (AI) innovation.

John Healy, Intel vice president overseeing industrial robotics in the Client Computing and Physical AI Group, met with reporters ahead of the Intel Robotics Workshop at JW Marriott Hotel Seoul on Oct. 14.

“Robotics market opportunities are growing rapidly across our Korean customer and partner ecosystem,” Healy said. “We are working with Korean companies in manufacturing, logistics, retail and other industries, and we are very positive about the products under development.” He declined to identify the companies or products.

Intel is pursuing multiple physical AI projects with major Korean manufacturers to support the transition to software-defined manufacturing (SDM) and process automation. The company believes South Korea is driving physical AI innovation with global implications, supported by its precision manufacturing capabilities and advanced technology.

Cooperation with the Korean government on physical AI industries, including robotics, is also taking shape. “Intel strongly supports Korean government-led initiatives,” Healy said. “Yesterday, we shared Intel’s perspective and discussed ways to help ensure the success of those initiatives.” He did not disclose details of the discussions but reaffirmed Intel’s commitment to supporting technological and industrial development.

The previous day, Healy spoke at the Physical AI Alliance event hosted by the Ministry of Science and ICT, describing South Korea as an ideal ecosystem for implementing physical AI.

Intel is supporting robotics markets in South Korea and abroad with its Core Ultra Series 3 processors. The system-on-chip (SoC) integrates a central processing unit (CPU), graphics processing unit (GPU) and neural processing unit (NPU) on a single chip. Compared with computing architectures that use separate chipsets for each function, the integrated design reduces hardware and software complexity, significantly lowering the total cost of ownership (TCO) for robotics products.

“Physical components such as actuators account for the largest share of humanoid robot manufacturing costs, but Intel’s SoC computing system can also contribute significantly to cost reductions,” Healy said. He explained that the approach reduces technical complexity and the burden of managing multiple subsystems while allowing performance improvements through continuous software updates. It also reduces the power required to operate robots.

Intel is emphasizing its open ecosystem as a competitive advantage over Nvidia. The company does not restrict initial AI model training or simulation to a particular vendor or platform, allowing developers to work in their preferred environments.

When developers optimize completed AI models for Intel-powered robots and deploy them, Intel provides software support through PyTorch XPU and the OpenVINO Physical AI Framework. This enables rapid deployment with minimal code changes.

“Intel has contributed to the Robot Operating System (ROS) community for more than 15 years,” Healy said. “We will actively support the development of physical AI technology standards that prevent vendor lock-in and ensure interoperability and developer choice.”