LG CNS said on the 5th that it is teaming up with AI Semiconductor FuriosaAI to develop Neural Processing Unit (NPU)-based AI services and target the public AX market.
LG CNS signed a memorandum of understanding (MOU) with FuriosaAI to expand cooperation in the AI infrastructure business at its headquarters in LG Science Park in Magok, Seoul. The signing ceremony was attended by Vice President Kim Tae-hoon of LG CNS AI Cloud Business Unit and FuriosaAI CEO Baek Jun-ho, among other senior executives.
FuriosaAI is an AI Semiconductor that designs and develops NPUs, semiconductors specialized for AI computation, and with its second-generation NPU RNGD, it possesses technology that meets the high-performance requirements for large-scale AI services while significantly reducing power consumption and operating expense compared with GPUs. FuriosaAI began full-fledged global market expansion after succeeding in mass production, taking delivery of 4,000 RNGD units in January.
The two sides will expand cooperation on the independent AI foundation model project, in which they are participating as part of the LG AI Research consortium. Based on K-EXAONE applying FuriosaAI RNGD, LG CNS will optimize AI service performance and push for commercialization, while FuriosaAI will handle stable NPU supply and technical support. Through this, they plan to strengthen a sovereign AI ecosystem composed of domestic AI models and services, infrastructure, and AI Semiconductor, and provide services optimized for the public AX market.
As a first step in the collaboration, LG CNS will conduct technology verification by applying FuriosaAI NPUs to the operating infrastructure of its in-house, corporations-focused agentic AI platform AgenticWorks. Because agentic AI, by nature, sets its own goals and performs complex tasks, high-performance, high-efficiency infrastructure is essential; LG CNS plans to use NPU-based infrastructure to boost both power and operational efficiency.
The two sides will also conduct a proof of concept for performance optimization technology for NPU-based GPUaaS (GPU as a Service). GPUaaS virtualizes GPUs so users can leverage high-performance computing environments without purchasing hardware directly, drawing attention as a core infrastructure model in the AI era. They will optimize infrastructure by applying NPUs across AI training, service operations, and inference to improve power efficiency and expense competitiveness.
Kim Tae-hoon, vice president of the LG CNS AI Cloud Business Unit, said, "Through cooperation with FuriosaAI, we will secure NPU-based AI infrastructure technology and expert talent so customers can use agentic AI more efficiently," adding, "We will support the advancement of the national flagship AI model through collaboration with LG AI Research and contribute to the development of the domestic AI industry."