Nordic AI Cloud Capacity to Hit One Gigawatt by End of 2027, NVIDIA Says

NVIDIA announced that AI cloud capacity in the Nordic countries is projected to reach one gigawatt by the end of 2027. The company also highlighted the region's leadership in unicorn density per capita in Europe. Cloud providers operating in the Nordic AI cloud include Bitdeer, CoreWeave, Crusoe, Hyperstack, Nscale, Verda, and Volta.

Shared Supercomputer and Enterprise Deployments

A Swedish consortium, anchored by Ericsson, AstraZeneca, Saab, SEB, Wallenberg Investments, and SKF, launched Sferical One, a shared supercomputer built with two NVIDIA DGX SuperPODs, 16 racks of DGX GB300, and 1,152 GPUs. Sferical One will deliver services built on NVIDIA Nemotron open models.

Spotify runs AI-powered podcast recommendations with NVIDIA TensorRT-LLM, lowering cost per query. ABB Robotics created RobotStudio HyperReality using NVIDIA Omniverse, achieving up to 99% simulation accuracy and up to 80% faster commissioning. Elisa beta-launches Moontalk AIRI, a real-time enterprise assistant built on NVIDIA Nemotron open models.

Model Customization and Edge AI

Stilta post-trains NVIDIA Nemotron models to scale patent-portfolio analysis across thousands of patents. ConfidentialMind's SampoTron, based on NVIDIA Nemotron-3-Embed, improves cross-lingual retrieval accuracy by more than 20%. Agaton uses NVIDIA Nemotron 3.5 Lightning to cut response latency in live-call Revenue Mining. Klang fine-tuned NVIDIA Parakeet v3 into Pianissimo, a Swedish transcription model 75% more accurate and 64 times faster, released on Hugging Face.

Embedl and Space Cargo Unlimited unveiled a protein-crystallization monitor on a single NVIDIA Jetson Orin NX module, operating without continuous ground link. Sweden AI Factory, one of 19 EuroHPC AI factories, partners with NVIDIA to give Inception startups access to infrastructure and expertise.

Autonomous Systems and Quantum Computing

Einride builds its next-generation Driver on NVIDIA Hyperion, using NVIDIA Halos safety system and Cosmos open-world models for autonomous freight trucks. Toyota Material Handling Europe adopts a simulation-first autonomous material-handling approach with NVIDIA and SoftServe, halving deployment time globally. Norrsken Foundation and NVIDIA will open the Stockholm AI Lab in 2027 for AI-native founders tackling global challenges. FirstQFM used NVIDIA accelerated computing to train a model that cuts quantum-processor execution time by 33% without performance loss, validated on a QuantWare unit.

NVIDIA founder and CEO Jensen Huang will keynote GTC Berlin on 21 Oct 2026 at 11 a.m. CEST.