Watch on YouTube
Watch on Vimeo
VMware by Broadcom began its Tech Field Day session at KubeCon North America 2026 by showing how VMware vSphere Kubernetes Service (VKS) fits into VMware Cloud Foundation (VCF) as a core service for an on-premises private cloud, alongside virtual machines and a container service. Timmy Carr, a longtime VMware and Kubernetes product leader, explained that Broadcom renamed the former TKG Service to VKS to end years of confusion across Tanzu product names and make it clear that Kubernetes is included in every VCF and VMware vSphere Foundation (VVF) subscription. Any customer on vSphere 8 or later can enable the Supervisor and start running clusters. Tanzu remains a separate offering focused on Cloud Foundry, Bitnami-based images, and data services. VKS is CNCF-certified Kubernetes, ships each new upstream release within two months, and supports each version for 24 months, giving enterprises a longer runway than the community standard. Platform teams can provision clusters through the UI, CLI, or GitOps, or delegate CPU, networking, and storage quotas to application teams for self-service.
Security hardening was a major focus of the past three releases. VKS supports Canonical Ubuntu (with Ubuntu Pro support included for VCF customers), Broadcom’s own Photon OS, and custom images such as Windows or RHEL. Photon OS is moving step by step toward an immutable node OS: VKS 3.7 locked down the user partition, VKS 3.8 disables SSH access, the OS layer is signed and verified through secure boot, and Advanced Intrusion Detection Environment (AIDE) file integrity monitoring now reports unexpected file changes. For compliance, VKS added OPA Gatekeeper in version 3.6 and policy bundles in version 3.7 that align clusters with baselines such as CIS, leaving the audit itself to the customer and their auditor.
To reduce operational burden, VKS 3.8 introduces chained upgrades, which step a cluster through multiple Kubernetes versions with a single action in the UI, CLI, or GitOps workflow, while checking workload health factors such as pod disruption budgets before proceeding. The add-ons API introduced in VKS 3.6 manages add-on dependencies during upgrades and now supports any Helm chart, and an Argo CD Supervisor service supports GitOps-driven configuration. Carr noted that Broadcom is a top-three long-term contributor to Kubernetes, including Cluster API, etcd, containerd, and CSI, and highlighted two KubeCon announcements: a workload conformance effort with SIG Apps and SIG Architecture to verify that applications are portable across Kubernetes distributions, and the open-sourcing of the VM API that VCF uses to define virtual machines declaratively through Kubernetes. The session closed with demos of a chained upgrade from Kubernetes 1.35 to 1.37 and a one-click installation of the Headlamp dashboard as a cluster add-on.
Personnel: Timmy Carr
Watch on YouTube
Watch on Vimeo
VMware by Broadcom’s Jeremy Wolf, a lead product manager on the VMware vSphere Kubernetes Service (VKS) and Supervisor team, used this Tech Field Day session at KubeCon North America 2026 to show that vSphere is an advantage for running Kubernetes. Enterprises have spent 20 years running mission-critical, highly available applications on vSphere, and VKS lets them apply that same foundation to modern and AI workloads. Wolf walked through the three stages of running a Kubernetes platform. At day-zero, the Supervisor provides a Kubernetes control plane on the infrastructure, and vSphere Namespaces handle tenancy, role-based access control, network assignment, and storage quotas. During the build phase, Cluster API deploys clusters declaratively, and app teams request changes as variable overrides to a cluster class. For day-two operations, VKS clusters directly inherit vSphere HA, DRS, vMotion, and compute oversubscription features such as memory tiering. VI admins keep working in vCenter, owning hosts, capacity, storage, and networks, while platform engineers use familiar tools such as kubectl, Helm, Argo CD, and the VCF CLI.
Wolf described how two schedulers work together: DRS places VKS control plane and worker node VMs to avoid noisy neighbors, and the Kubernetes scheduler then places pods within those nodes, each with its own affinity and anti-affinity controls. When an admin puts an ESX host into maintenance mode, vMotion moves the worker node VM to another host without draining the node, rescheduling pods, or detaching persistent volume claims, so application teams never notice. Workloads pinned to a host through PCI passthrough devices such as GPUs or host-local datastores are the exception, and Broadcom is building a way for those workloads to request a warning so they can drain cleanly first. Unplanned host failures involve three recovery clocks: vSphere HA restarting the VM on another host, the Kubernetes node controller replacing an unreachable node, and pod eviction. In most customer environments, vSphere HA wins that race, and when it doesn’t, it usually comes down to insufficient spare capacity, so Wolf confirmed that customers should keep headroom in their clusters.
On storage, VM storage policies in vCenter are exposed to the Supervisor as Kubernetes storage classes, and the vSphere CSI driver works with Cloud Native Storage in vCenter to place each volume on a compliant datastore, which lets admins decommission a datastore with Storage vMotion while the pod’s storage class stays the same. On networking, the Supervisor supports VDS, NSX, and NSX VPC, and builds L4 load balancing into the platform from the start through the Foundation Load Balancer or NSX. For L7, customers can run Avi with its Kubernetes operator inside VKS clusters for enterprise tuning and visibility, or use an in-cluster controller such as Contour or Istio. Wolf framed Broadcom’s multi-cloud stance around portability: use VKS for on-premises Kubernetes, use EKS, AKS, or GKE in the public cloud, and manage both through GitOps, with Cluster API and the newly open-sourced VM API keeping clusters and VMs portable. He also clarified that Tanzu runs on top of VKS clusters within VCF, and closed by challenging viewers to deploy the Supervisor in a lab, run workloads on a VKS cluster, and put a host into maintenance mode to see the result for themselves.
Personnel: Jeremy Wolf
Watch on YouTube
Watch on Vimeo
VMware by Broadcom’s Elliott Davis, AI architect for VMware Cloud Foundation (VCF), used this Tech Field Day session at KubeCon North America 2026 to explain how enterprises can host AI workloads on VMware vSphere Kubernetes Service (VKS) using open-source components. His central point is that AI agents, rather than humans, will generate most future LLM calls. A single human question can now trigger hundreds of tool calls, retrievals, and model calls as agents plan, retry, and spawn subagents, producing load that Davis compared to a DDoS. He split the problem into two sides: token production on GPUs, which is expensive but measurable through metrics such as time to first token and throughput, and agents, which consume those tokens in bursts and sit idle most of the time while still holding the pod resources they requested. Davis said enterprise agent adoption is still early, and most VCF customers today prefer the platform’s no-code agent builder. He cited a Gartner projection that 90% of enterprise agent workflow steps will run on-premises by 2030 at roughly 20 times lower cost, and stressed that the larger benefit of open-weight models is keeping control of enterprise data.
Davis described a three-layer stack, with VCF contributing at each level. At the bottom, the model runtime pairs an inference engine such as vLLM with Run:ai, which Davis said is now fully open source and handles GPU scheduling, partitioning, and preemption. vSphere adds hardware-level GPU slicing and multi-tenancy, assigning GPU capacity to separate VKS clusters without building separate physical clusters for each team, because on-premises GPUs only pay off when they stay busy. In response to a delegate’s question, Davis placed LLM-D behind the gateway layer for KV cache-aware routing and disaggregated inference, which avoids recomputing the static prompt prefixes common in agent requests. In the middle sits Agent Gateway, the open-source CNCF project from Solo.io, which acts as the policy enforcement point, MCP proxy, and agent-to-agent (A2A) gateway. Istio and Envoy policies force all agent traffic through it, credentials are injected along the way so untrusted agents never hold them, and it decides whether each request goes to an on-premises open-weight model or a frontier model. Broadcom contributes to the project and plans to ship Agent Gateway as a VKS add-on within two weeks. Davis acknowledged that gateway sprawl is an industry problem, with several gateways inside Broadcom alone, and described the pieces as interchangeable building blocks.
The top layer runs the agents themselves through the open-source Agent Substrate project, which treats agents more like network requests than long-running infrastructure. Running one agent per pod would overwhelm a cluster, so Agent Substrate keeps a warm pool of sandboxes and switches agents on and off within roughly 500 milliseconds, multiplexing many agents onto shared resources. Davis has run hundreds of gVisor instances inside a vSphere Pod, with layered isolation: gVisor contains the agent, the VKS node VM or vSphere Pod restricts any escape, and Agent Gateway with deny-by-default network policy blocks outbound access. He closed with a use case drawn from customer requests: an agent that finds orphaned persistent volume claims and idle GPU reservations across VKS clusters and proposes a cleanup plan without acting until a human approves. Listing resources can go to an on-premises model, assessing risk goes through Agent Gateway to a frontier model, the dry run executes in a sandbox, and the agent suspends itself while waiting for approval so it consumes no resources until the user returns.
Personnel: Elliott Davis
Thank you for being part of the Tech Field Day community! Our mailing list is a great way to stay up to date on our events and technical content, and we appreciate your signup.
We promise that we’ll never spam you, send ads, or sell your information. This list will only be used to communicate with our community about our events and content. And we’ll limit it to no more than one message per week.
Although we only need your email address, it would be nice if you provided a little more information to help us get to know you better!