FuriosaAI to supply 8,800 RNGD NPUs for 15MW Stockholm AI data center

South Korean AI chip startup says the Sweden project supports European expansion as it targets 80 billion to 100 billion won in annual revenue and scales Renegade commercialization.

Summary

FuriosaAI said it will supply more than 8,800 units of its second-generation Renegade, or RNGD, AI inference accelerator to a 15MW data center project in Stockholm being developed with I/ONX HPC and Velox, marking what it described as its first full deployment in a large-scale European data center. The first 2MW phase is due to begin operations in early 2027, followed by an additional 8MW expansion in the second half of 2027. Separately, the South Korean fabless chip company said it aims to produce 20,000 Renegade units this year and generate 80 billion to 100 billion won in annual revenue, after securing about 20 billion won in orders within five months of starting mass production. Analysts and industry insiders said additional contracts and repeat orders will be important to validate commercialization in a market still anchored by Nvidia's CUDA software ecosystem.

Terms & Concepts
  • NPU: A neural processing unit, a chip designed to accelerate AI workloads, especially inference tasks.
  • AI inference: The stage where a trained AI model is used to generate outputs for real-world applications, as distinct from model training.
  • CUDA: Nvidia's software platform and development environment for GPU computing, widely used in AI and a key competitive barrier for rival chipmakers.