KT puts Korean AI chips, language model into one enterprise server

KT has introduced an integrated enterprise server that combines a Korean-made artificial intelligence processor, language model, and operating platform into a single device. This new offering, called the KT NPU LLM Station, aims to enable companies to run generative AI entirely within their own networks, providing greater control and security over sensitive data.

The system is built around Rebellions’ ATOM-MAX neural processing unit, which is specifically designed for inference tasks. KT's own Mi:dm K 2.5 Pro large language model, optimized for complex reasoning and agent-based applications, is also integrated into the appliance. These components work together seamlessly, allowing companies to operate the hardware, model, and related services as a single package.

One of the key features of the KT NPU LLM Station is its application programming interface platform, which simplifies the process of managing and utilizing the hardware, model, and services as a cohesive unit. By processing data and AI computations internally within the customer's facilities, the system addresses the security and network-separation requirements of industries with stringent regulations.

The ATOM-MAX neural processing unit has demonstrated superior processing performance and power efficiency compared to comparable graphics processing units in certified testing, making it a compelling choice for companies seeking high-performance AI capabilities. With the KT NPU LLM Station, Korean enterprises can now leverage advanced AI technologies while maintaining control over their data and infrastructure.