Runware packs 1MW of AI inference into a 20-foot container
Runware says workloads can be redirected to other pods if an individual unit loses power, cooling or network connectivity.
Credit: Runware
Runware has developed a modular 1MW AI inference pod capable of housing around 1,200 GPUs, offering a distributed approach to AI data-centre infrastructure.
Register for FREE to keep reading
Join 12,000+ scientists, engineers, and IT professionals driving innovation through informatics, HPC, and simulation with:
- Insights into HPC, AI, lab informatics & data
- Curated content for life sciences, engineering & academia
- Access to Breakthroughs: real-world computing success
- Free reports & panels, including the Lab Informatics Guide
- White Papers & software updates for smarter research
Sign up now
Already a member? Log in here
Your data is protected under our privacy policy.
