Guest: Lukas Gentele (LinkedIn)
Company: vCluster Labs
Show Name: KubeStruck
Topics: Kubernetes, Cloud Native

Kubernetes autoscaling has long been powerful — but also fragmented. AWS Karpenter brought intelligent bin-packing and cost-aware scaling to EKS users, yet teams running in private clouds or hybrid setups were left behind. vCluster Auto Nodes changes that.

A Universal Autoscaler for Kubernetes

As Lukas Gentele, Co-Founder and CEO of vCluster Labs, explains, Auto Nodes is vCluster’s take on a managed, environment-agnostic version of Karpenter. “We’re baking Karpenter directly into vCluster so it works anywhere — public cloud, private cloud, or bare metal,” he says. By embedding it natively, Auto Nodes becomes “batteries included,” delivering an optimized cluster setup right out of the box.

The system intelligently provisions nodes — whether EC2 instances on AWS, Azure VMs, or OpenStack nodes — based on workload demand. Using Terraform or OpenTofu, vCluster Auto Nodes can define and spin up any node type available to the environment. This flexibility extends beyond cloud providers: “Even in private environments, there are Terraform providers for OpenStack or bare-metal-as-a-service from Canonical,” Gentele explains.

Autoscaling Beyond the Cloud

vCluster’s new approach also integrates with Kubeforce to make autoscaling viable for on-prem infrastructure. For smaller workloads, Auto Nodes can carve out virtual slices of bare-metal machines as VMs using KVM — essentially giving you right-sized capacity on hardware you already own.

For GPU-heavy or AI workloads, Auto Nodes connects directly with NVIDIA’s Base Command Manager (BCM), allowing vCluster to provision and manage DGX systems alongside cloud instances. “You could have DGX nodes for GPUs and add AWS EC2 instances when extra capacity is needed,” Gentele notes.

Hybrid and Multi-Environment Efficiency

What makes Auto Nodes unique is its ability to operate across environments simultaneously. Teams can define node pools that include on-prem GPUs, cloud CPUs, and virtualized compute slices — all tied together by automatically configured VPNs that handle secure communication between nodes.

This hybrid design allows Kubernetes clusters to extend seamlessly from private data centers to public clouds. “We want to build the most optimized Kubernetes cluster possible, even across different environments,” Gentele says.

Takeaway

With Auto Nodes, vCluster Labs is redefining what intelligent autoscaling means in Kubernetes. By combining Terraform’s flexibility, Kubeforce’s virtualization, and NVIDIA’s hardware integration, Auto Nodes offers one of the most complete and adaptive scaling systems in the market — delivering the efficiency of Karpenter to wherever your workloads live.