Samsung SDS Launches Subscription-Based NPUaaS Powered by FuriosaAI's Renegade

by Shin Hye An Posted : July 21, 2026, 09:24Updated : July 21, 2026, 09:24

Samsung SDS announced on July 20 that it has launched a subscription-based NPU as a Service (NPUaaS) integrated with FuriosaAI's second-generation neural processing unit (NPU) Renegade (RNGD) on the Samsung Cloud Platform (SCP).


The second-generation NPU Renegade from FuriosaAI is a domestically developed NPU specialized for artificial intelligence (AI) inference. It is characterized by its high power efficiency and cost-effectiveness compared to graphics processing units (GPUs) when applying pre-trained AI models to real-world services, such as generating responses, analyzing documents, and identifying images.


As a result, customers using Samsung SDS's NPUaaS can access NPU servers through a subscription model without the need to purchase or set up their own data centers. They can choose to utilize NPUs in increments of one, two, four, or eight, depending on their business needs and data scale, making it suitable for both startups and large enterprises.


Additionally, the NPUaaS can be integrated with various SCP offerings, including high-performance storage, computing, and high-speed networking.


Notably, Samsung SDS is providing this NPUaaS within a sovereign cloud environment on SCP, ensuring that public sector clients subject to strict security regulations can use the service with confidence.


With the launch of NPUaaS, Samsung SDS aims to expand its offerings beyond existing GPUaaS to include NPU-based computational resources, enhancing customer choices for AI infrastructure and marking a significant turning point for the proliferation of domestic NPU-based AI infrastructure and the development of sovereign AI.


As the AI industry shifts from model development to applying inference in real services, the importance of NPUs is increasing. The rise in generative AI users leads to greater demand for inference operations, as well as increased power and operational costs for businesses. NPUs, optimized for AI inference, provide high power efficiency and cost competitiveness for specific inference tasks while reducing service response delays.


To alleviate the initial infrastructure setup burden for customers and flexibly respond to the growing demand for AI inference, Samsung SDS has introduced NPU as a subscription-based cloud service.


Lee Ho-jun, Vice President and Head of the Cloud Services Division at Samsung SDS, stated, “The launch of NPUaaS is significant not just as a new cloud product, but because it allows customers to utilize high-performance AI technology more flexibly and cost-effectively. Samsung SDS will continue to diversify SCP-based products to meet customer needs and lead the cloud AI technology ecosystem.”





* This article has been translated by AI.