Mimer - Sweden’s national AI Factory
Speakers: Eric Schnepf (ISC Group) · Gerhard Wellein (Friedrich-Alexander-University Erlangen-Nuremberg (FAU))
Session summary
Petra Dahlunde, head of AI Ecosystem and Policy at Mimer, presents Sweden's national AI Factory, run jointly by RISE (Research Institutes of Sweden) and NAISS, the national academic infrastructure for supercomputing, with Linkoping University leading the supercomputer procurement. The as-yet-unnamed machine will be a liquid-cooled, cloud-based Kubernetes platform on Bull's BullSequana XH3500 architecture with 100 nodes of NVIDIA GB200 totaling 400 GPUs plus a small Grace CPU partition, expected to open next year; in the meantime Mimer has been serving customers for eight months on existing resources. The factory's primary mission is helping startups and SMEs shorten time to market while building trustworthy AI, alongside academic and public sector users. Beyond compute access, Mimer offers AI expert support from idea sanity checks through development, fine-tuning, testing, and validation to production migration, though it does not host inference. It works on AI Act risk classification, participates in national and European standardization, and runs training, workshops, and hackathons. Focus areas are gaming, life sciences, autonomous systems, and materials science, though the largest customer group falls outside these areas. The customer journey moves from eligibility screening through scoped projects to wrap-up and hand-off, often to European cloud providers positioned as alternatives to hyperscalers. Mimer reports 285 customers in the pipeline, 65 completed, and staff growth from one to 70 people in eight months, with many customers building LLM-based services and agentic AI products.
Topics: eu ai factories · sme support · trustworthy ai · gpu infrastructure · ai act compliance · innovation ecosystems
AI-generated summary of an auto-generated transcript (~1,260 words in full). Details may be imprecise — verify against the session recording.
Auto-generated captions from ISC 2026 session recordings · transcription errors likely, verify quotes against the video · timestamps are offsets into each recording · independent tool, not affiliated with ISC · a Radixia Labs experiment
