Major companies in Sweden, Finland and other Nordic nations are using NVIDIA systems to run large-scale AI models and advanced simulations.
Key points
- A consortium including Ericsson launched Sferical One, a supercomputer with NVIDIA DGX GB300 systems.
- Spotify uses NVIDIA software to lower the cost of AI-powered podcast recommendations.
- ABB Robotics achieves high simulation accuracy using NVIDIA Omniverse libraries.
Nordic technology companies are expanding their computing infrastructure to support advanced artificial intelligence and industrial simulations. These organizations rely heavily on hardware and software from NVIDIA to process data efficiently.
What happened
Nordic nations are building a major technology ecosystem supported by abundant renewable energy and strict data rules. Cloud capacity in the region is projected to reach one gigawatt by late 2027, with providers like Bitdeer and CoreWeave expanding operations. In Sweden, a group anchored by Ericsson and AstraZeneca launched Sferical One, a shared supercomputer. This system runs on two NVIDIA DGX SuperPODs containing 1,152 GPUs. The consortium plans to build services using NVIDIA Nemotron open models.
Spotify uses NVIDIA TensorRT-LLM inference software to deliver AI recommendations for podcasts and music. This setup lowers the cost per query while serving a global audience. ABB Robotics developed RobotStudio HyperReality using NVIDIA Omniverse libraries to simulate robot performance. The solution allows engineers to test robots virtually with high accuracy before deployment. Finnish telecom company Elisa is launching Moontalk AIRI, an assistant built on Nemotron models for real-time meeting insights.
Startups are also adopting these tools. Stilta uses Nemotron to analyze patent portfolios at scale. ConfidentialMind built SampoTron on Nemotron-3-Embed to improve cross-lingual retrieval accuracy. Agaton uses Nemotron 3.5 Lightning to speed up response times during live sales calls.
Why it matters
The fix
- Monitor vendor announcements for software updates related to AI inference engines.
- Review access controls for systems processing large volumes of proprietary data.
- Ensure simulation environments are isolated from production networks.
- Verify that model training data complies with regional privacy regulations.
Why it matters
Teams usually separate development environments from production systems to prevent accidental data leaks. When using large language models, organizations should check whether data used for training or inference leaves their control. It is important to verify that simulation tools do not expose sensitive operational parameters. Companies often audit third-party software to ensure it meets internal security standards before deployment.
What to do and how to stay safe: NVIDIA
- Identify which AI services process sensitive internal data or customer information.
- Confirm that vendors provide clear documentation on how model weights are stored.
- Test simulation outputs to ensure they do not reveal proprietary engineering details.
- Review cloud configurations to prevent unauthorized access to computing resources.
Step-by-step guide: What is password policies and how they protect systems
General security guidance from the Crypt Weekly newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
Which supercomputer did the Swedish consortium launch?
The consortium launched Sferical One, which runs on two NVIDIA DGX SuperPODs with 1,152 GPUs.
How is Spotify using NVIDIA technology?
Spotify uses NVIDIA TensorRT-LLM inference software to run AI-powered podcast recommendations more efficiently.
What benefit does ABB Robotics get from its new solution?
ABB Robotics reduces commissioning time by up to 80% and achieves high simulation accuracy using NVIDIA Omniverse.
