Kolon Benit said on the 15th that it unveiled an enterprise AI platform that provides high-efficiency artificial intelligence (AI) inference based on heterogeneous semiconductors in collaboration with Red Hat and Rebellions.
Kolon Benit held a seminar on the 11th with Red Hat and Rebellions to introduce an enterprise AI platform. The solution supports corporations in adopting and scaling AI services more quickly and reliably through a heterogeneous semiconductor-based AI inference environment. By combining AI model optimization technology with a Neural Processing Unit (NPU)-based high-speed inference architecture, it secured the performance, expense efficiency, and stability required for enterprise AI transformation.
The event also featured a session sharing the latest AI inference architecture for the new product and open-source-based operations technology. Kolon Benit presented a field-validated industrial case that applied Red Hat OpenShift AI, an enterprise-grade AI platform, and a high-performance inference platform based on the Rebellions ATOM NPU, while attendees examined the full AI life cycle—from AI infrastructure build-out to model serving, operations, and scaling—and the technical challenges in real-world operations.
The first presentation session addressed the shortage of specialists that corporations face as they ramp up AI businesses, the limitations of an open-source-centered environment, and the need for an enterprise AI operating framework to compensate. Ahn Gi-seok, a senior executive at Kolon Benit, said, "Although AI adoption entails multiple hurdles across technology, talent, and operations, Kolon Benit supports customers' AI adoption reliably through enterprise AI platforms such as Red Hat OpenShift AI and a team of specialized engineers."
Next, Choi Moon-su, a principal at Kolon Benit, introduced a case of building an NPU-based Multimodal AI safety control system carried out with the Kolon GLOBAL Smart Construction Team. In the Rebellions ATOM NPU environment, Kolon Benit implemented a Vision Intelligence system that determines hazardous situations at construction sites in real time by combining CCTV video object detection with situation descriptions and alert text generation. Choi said, "By using a heterogeneous semiconductor-based infrastructure that combines GPUs and NPUs and a multimodal model, we simultaneously improved expense savings, inference speed, and control accuracy compared with a conventional GPU-only environment."
Executive Director Choi Sang-moon of Kolon Benit said, "AI infrastructure has now moved beyond new technology to become the foundation that determines the core operational competitiveness of corporations," adding, "Through our collaboration with Red Hat and Rebellions, we have introduced an enterprise AI platform with efficiency and stability, and we will continue to deliver solutions that accelerate AI business innovation across industries."