Norway's 2 petabytes of Huawei flash storage and LLM training
blocksandfiles.com
May 25, 2026
4 min read
🔥🔥🔥🔥🔥
63/100
Summary
Norway's National Library is developing a large language model (LLM) for the Norwegian language using 2 petabytes of Huawei OceanStor Dorado flash storage. Marius Husnes, Head of IT Platform, stated that no commercial LLM provider is currently developing a local Norwegian language model.
Key Takeaways
- Norway's National Library is developing a sovereign large language model (LLM) for the Norwegian language using 2 petabytes of Huawei OceanStor Dorado flash storage.
- The library has the largest digital collection of Norwegian cultural heritage, which includes books, newspapers, and web content, and is legally mandated to preserve this data.
- The LLM training process involves overcoming challenges related to data quality, cleaning, and the integration of different storage systems for efficient data pipeline management.
- The library is creating its own evaluation tools for the LLM due to the lack of standard metrics for assessing a sovereign Norwegian language model.
Community Sentiment
Positives
- The national library's initiative to train a sovereign LLM could enhance the representation of Norwegian culture and language in AI, addressing a significant gap.
- The Olivia system's powerful hardware setup, with 448 GPUs, indicates a serious investment in advancing AI capabilities, even if concerns about its utility remain.
- The existing user interface of the national library is praised for its functionality, suggesting a strong foundation for integrating AI to improve text searchability.
Concerns
- Training a sovereign LLM with limited hardware raises doubts about its effectiveness and usefulness, suggesting a potential waste of resources.
- Concerns about the lack of a clear application for the LLM indicate skepticism about its impact on the Norwegian language and culture.
- The notion that a locally trained LLM might be inferior to existing models raises questions about the rationale behind this initiative.