Hello OpenNebula community,
As you all know, running modern AI infrastructure comes with real trade-offs:
You want high GPU utilization, but you also need strong tenant isolation.
You want flexibility, but not at the cost of operational complexity.
And you need near-native performance, even in virtualized environments.
OpenNebula Elastic Capacity Management helps bridge that gap 
By pooling GPU and InfiniBand resources and dynamically allocating them across workloads, it enables better hardware utilization, stronger tenant isolation, and a more flexible infrastructure model for modern AI and HPC environments.
Read more and watch the screencast 