AI Workloads Shake Kubernetes Out of Dormancy, Spurring Security and Portability Push
Kubernetes has shifted from a stable container platform into critical infrastructure for running AI systems, prompting vendors like Broadcom to rethink security perimeters and cross-cloud compatibility.
Once viewed as a maturing technology settling into predictable patterns, Kubernetes is experiencing renewed urgency as enterprises deploy it as the foundational layer for artificial intelligence workloads. This pivot is forcing a reconsideration of security models thought obsolete in the cloud era, while simultaneously raising demands for seamless operation across multiple cloud environments.
Vendors and infrastructure providers are mobilizing to address these shifting requirements. At KubeCon, Broadcom Inc. announced two open-source projects aimed at simplifying deployment and hardening security. Dilpreet Bindra, senior director of engineering in the VCF Division at Broadcom, characterized the moment as a turning point for a platform that had appeared to reach equilibrium.
There were a few that were like, Kubernetes is becoming mature, and so things were stabilizing, and they were less energy, I would say. And this KubeCon just blasts those doors wide open because Kubernetes has become the foundation for AI.
Dilpreet Bindra, Broadcom
Bindra spoke with theCUBE Research's John Furrier and Rebecca Knight at KubeCon + CloudNativeCon NA during an exclusive broadcast. The discussion centered on how Kubernetes now functions as an agent substrate and the steps Broadcom is taking to strengthen security across cloud boundaries while ensuring workload portability.
Securing the agent substrate across clouds
The rise of agents running on Kubernetes is reviving security concepts that cloud computing had largely abandoned. Broadcom's strategy emphasizes layered security, deploying its vDefend firewall at network perimeters and at deeper layers to prevent unauthorized network traffic when rogue agents are detected.
How do you then isolate them in a compute boundary in our vSphere Pods, which is a micro VM, so you can have a very secure runtime. You can employ things like Agent Substrate to be able to run within them.
Dilpreet Bindra, Broadcom
Ensuring portability represents the second pillar of Broadcom's initiative. The company is co-leading a new working group focused on Kubernetes workload conformance, with the goal of eliminating redundant validation cycles that software vendors currently perform for each Kubernetes distribution. The effort encompasses VMware vSphere Kubernetes Service and other variants. Additionally, Broadcom contributed its VM API project to the Cloud Native Computing Foundation, enabling virtual machines to operate as first-class objects within Kubernetes.
https://www.youtube.com/embed/1uybdi7pkFg?feature=oembed
We want Kubernetes to be Kubernetes wherever it runs in any environment. You don't have to do individual validations of VKS, and then do individual validations of other Kubernetes distributions. You can do it as one Kubernetes.
Dilpreet Bindra, Broadcom