Key points
- A Swedish consortium including Ericsson and Saab launched Sferical One, a supercomputer using 1,152 NVIDIA GPUs.
- Spotify uses NVIDIA TensorRT-LLM to lower costs for AI-driven podcast recommendations at global scale.
- ABB Robotics achieved 99 percent simulation accuracy using NVIDIA Omniverse libraries for robot testing.
The short version
- Nordic firms are deploying NVIDIA DGX SuperPODs and open models to build regional AI infrastructure.
- Major companies like Spotify, ABB and Elisa are using NVIDIA software for inference, simulation and agents.
- Sweden raised $2.8 billion in the first half of 2026 to support this expanding technology ecosystem.
The details
A consortium anchored by Ericsson, AstraZeneca, Saab, SEB and Wallenberg Investments launched Sferical One in Sweden. The shared supercomputer runs on two NVIDIA DGX SuperPODs containing 16 racks of NVIDIA DGX GB300 systems. The setup includes 1,152 GPUs. SKF serves as a strategic partner for the project. Researchers will use NVIDIA Nemotron open models to customize and deploy models in a secure environment. This launch marks a shift from regional ambition to practical application in advanced computing.
Spotify uses NVIDIA TensorRT-LLM inference software to power AI recommendations for music and podcasts. The deployment allows the company to deliver these recommendations to hundreds of millions of listeners at a lower cost per query. This efficiency case study highlights the ability to scale inference operations globally while maintaining performance. The system supports personalized content delivery for audiobooks and other media formats.
ABB Robotics developed RobotStudio HyperReality using NVIDIA Omniverse libraries. The tool bridges the gap between robot simulations and real-world performance by running the same firmware as physical robots. Engineers can train, test and validate robots virtually. According to ABB Robotics, the solution delivers up to 99 percent simulation accuracy. It also reduces commissioning time by up to 80 percent by using real-world data to improve digital models.
The context
Finnish telecom company Elisa launched a beta of Moontalk AIRI, an enterprise agentic assistant. The tool retrieves information and documents decisions during meetings and calls in real time. It is built on NVIDIA Nemotron open models and uses low-latency networking for context-aware responses. This deployment reflects the broader trend of Nordic firms building proprietary intelligence on NVIDIA infrastructure.
Sweden’s startup ecosystem raised $2.8 billion in the first half of 2026. The region is projected to reach $5 billion for the year. Startups are building fast on open models. Stilta uses NVIDIA Nemotron models to scale patent analysis for intellectual property protection. ConfidentialMind built SampoTron on NVIDIA Nemotron-3-Embed, improving cross-lingual retrieval accuracy by more than 20 percent. Agaton uses NVIDIA Nemotron 3.5 Lightning to cut response latency for live sales call analysis.
What to watch
- AI cloud capacity in the Nordics is projected to reach 1 gigawatts by the end of 2027.
- Providers including Bitdeer, CoreWeave, Crusoe, Hyperstack, Nscale, Verda and Volta are expanding.
- The region holds a structural advantage due to abundant renewable energy and rigorous data governance.
- The Nordics lead Europe in unicorn startup density per capita.
What it means for organisations
Operations and security teams should assess how third-party AI infrastructure impacts their data governance policies. When vendors use shared supercomputing resources or open models, verify that isolation controls meet internal risk standards. Supply chain managers need to understand the dependencies on specific hardware vendors for critical workloads. Leadership should evaluate whether the efficiency gains from centralized AI platforms justify the concentration of risk. Ensure that all partners have clear incident response protocols for shared environments.
Background: Supply chain attacks
A supply chain attack targets the software or hardware you use before it reaches you. By compromising a trusted vendor, attackers insert malicious code that installs automatically on your systems when you update or deploy the product. This bypasses perimeter defenses because the malicious payload arrives with a valid digital signature.
Read the full guide: Supply Chain Attacks: How Trust Becomes Your Biggest Security Risk
What to do and how to stay safe: NVIDIA
- Review data classification policies to ensure sensitive information is not processed on shared infrastructure without encryption.
- Verify that vendor contracts include specific clauses about model isolation and data retention.
- Monitor for unusual network traffic patterns associated with AI inference endpoints.
- Ensure that access controls for AI platforms are strictly limited to authorized personnel.
Step-by-step guide: Secure Home Routers for Small Business Networks
General security guidance from the TechSec Headlines newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
What hardware powers the Sferical One supercomputer?
Sferical One runs on two NVIDIA DGX SuperPODs with 16 racks of NVIDIA DGX GB300 systems and 1,152 GPUs.
How does ABB Robotics benefit from NVIDIA technology?
ABB Robotics uses NVIDIA Omniverse libraries to achieve up to 99 percent simulation accuracy and reduce commissioning time by up to 80 percent.
What is the projected AI cloud capacity in the Nordics by 2027?
The region is projected to deploy 1 gigawatt of AI cloud capacity by the end of 2027.
