We are excited to share a major milestone in our mission to bring high-performance, energy-efficient AI inference to data centers worldwide. FuriosaAI is partnering with I/ONX HPC and Velox on a flagship AI data center project in Stockholm, Sweden.
Velox recently broke ground on the facility, with the first 2 MW of AI compute capacity expected to come online in early 2027. An additional 8 MW is expected later that year, as the project scales toward a total capacity of 15 MW. Under the project framework, I/ONX plans to deploy 1,800 Furiosa RNGD accelerators in Phase 1, with more than 7,000 RNGD accelerators planned across subsequent expansion phases.
Sweden has become an increasingly attractive location for AI infrastructure, supported by its mature data center ecosystem, reliable energy infrastructure, and strategic position for serving customers across Europe. The project represents an important step in expanding access to AI inference capacity in the region.
Furiosa’s silicon will operate alongside GPUs deployed as part of I/ONX HPC’s high-density Symphony SixtyFour platform. Designed for heterogeneous AI compute, its architecture enables AI workloads to be matched with purpose-builtcompute resources, improving performance per watt while optimizing infrastructure utilization.
Velox is developing, managing, and operating the physical data center facility, which will serve customers under an Infrastructure-as-a-Service (IaaS) model.
I/ONX takes the role of system integrator and platform partner, while Furiosa supplies the RNGD silicon, drivers, software stack, and architectural support to power high-throughput, low-latency inference across the facility.
Solving the inference power wall in Europe
As AI workloads transition toward continuous, stateful agentic AI loops, token-generation volumes are exploding. Running continuous inference on traditional high-wattage hardware presents significant challenges around operational cost, heat generation, and regional power grid constraints.
By pairing the high-performance, power-efficient Furiosa RNGD accelerator (180W TDP) with I/ONX’s Symphony SixtyFour platform inside Velox’s facility, this project provides a sustainable solution to enterprise customers and cloud providers in Europe. By deploying purpose-built AI accelerators optimized for inference, the facility will deliver AI compute with improved power efficiency and a more sustainable infrastructure model compared to traditional GPU-centric approaches.
What’s next
Over the coming months, the teams will continue technical validation, software integration, and supply planning in preparation for the first deployment phase. We are honored to work alongside I/ONX and Velox to help build the foundation for sustainable, high-performance AI compute in Europe.
Written by
The Furiosa Team
.png)
