Nordic firms deploy NVIDIA infrastructure for AI, supercomputing and agentic systems
Swedish and Finnish companies are leveraging NVIDIA hardware and open models to accelerate AI adoption across industries in the region.
Key points
- A Swedish consortium launched Sferical One, a supercomputer powered by NVIDIA DGX SuperPODs.
- Spotify uses NVIDIA TensorRT-LLM software to lower costs for AI-driven podcast recommendations.
- Nordic AI cloud capacity is projected to reach 1 gigawatt by the end of 2027.
Nordic technology companies are expanding their use of NVIDIA computing infrastructure to support artificial intelligence and advanced simulation projects. According to NVIDIA blog posts, the region is leveraging its renewable energy and data governance strengths to build a significant technology ecosystem.
The Nordic countries lead Europe in unicorn startup density per capita. This structural advantage supports a rapid transition from research to application. As Threat Times reported on October 15, 2026, NVIDIA has also committed capital to UK infrastructure, highlighting broader European investments in agentic computing.
Infrastructure and Supercomputing
A consortium including Ericsson, AstraZeneca, Saab, SEB and Wallenberg Investments launched Sferical One in Sweden. SKF acts as a strategic partner. The shared supercomputer operates on two NVIDIA DGX SuperPODs containing 16 racks of DGX GB300 systems.
The system utilizes 1,152 GPUs to enable researchers to customize and deploy models securely. The consortium plans to build services using NVIDIA Nemotron open models. This setup allows for private adaptation of large language models within a controlled environment.
Enterprise AI Deployments
Spotify employs NVIDIA TensorRT-LLM inference software to process AI-powered podcast recommendations for hundreds of millions of listeners. The deployment focuses on inference efficiency, allowing the company to deliver global recommendations at a lower cost per query.
ABB Robotics developed RobotStudio HyperReality using NVIDIA Omniverse libraries. This tool bridges the gap between robot simulations and physical performance. According to ABB Robotics, the solution achieves up to 99 percent simulation accuracy and reduces commissioning time by up to 80 percent.
Startup Ecosystem Growth
Sweden’s startup sector raised $2.8 billion in the first half of 2026 and is projected to reach $5 billion for the full year. Many of these emerging companies are building on open models to scale their operations quickly.
Stilta, a Stockholm-based startup, uses NVIDIA Nemotron open models to analyze entire patent portfolios. This technology allows for continuous monitoring of thousands of patents, a task previously difficult to scale.
ConfidentialMind built its SampoTron tool on NVIDIA Nemotron-3-Embed. The system delivers more than 20 percent improvement in cross-lingual retrieval accuracy for English, Finnish and Swedish languages.
Agaton uses NVIDIA Nemotron 3.5 Lightning to speed up its Revenue Mining solution for live sales calls. By routing knowledge retrieval to this model, Agaton reduces response latency to inform customer buying decisions in real time.
Finnish telecom Elisa is launching a beta of Moontalk AIRI, an enterprise agentic assistant. Built on NVIDIA Nemotron open models and distributed infrastructure, the tool retrieves information and documents decisions during meetings using low-latency networking.
What should you do?
Organizations evaluating AI infrastructure should review their energy consumption and data governance frameworks before scaling. Companies should assess whether open models like Nemotron meet their specific accuracy and security requirements.
Background: Large language models
Large language models are a specific type of neural network trained on text. Generative AI is the broader umbrella term for any system that creates new content, including images and code. Not all generative AI uses large language models, but most modern text generators do.
What to do and how to stay safe: NVIDIA
- Review internal data governance policies to ensure they align with regional regulatory standards before deploying AI models.
- Evaluate the energy requirements of proposed AI workloads to determine if renewable energy sources can sustain operations.
- Test inference efficiency tools like TensorRT-LLM in non-production environments to measure cost savings and latency improvements.
- Monitor vendor documentation for updates on open model compatibility with existing enterprise software stacks.
Step-by-step guide: Mobile Device Security: How to Protect Data Beyond Passwords
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
Which companies are part of the Sferical One supercomputer consortium?
The consortium includes Ericsson, AstraZeneca, Saab, SEB and Wallenberg Investments, with SKF as a strategic partner.
What is the projected AI cloud capacity for the Nordics by 2027?
The region is projected to deploy 1 gigawatt of AI cloud capacity by the end of 2027.
How does ABB Robotics claim its new simulation tool impacts commissioning time?
ABB Robotics states that RobotStudio HyperReality reduces commissioning time by up to 80 percent.
