Your VMs Are the Load-Bearing Walls in Your Estate
Every year, KubeCon rolls around and the conversation turns, almost inevitably, to the future: a future in which containers have won, Kubernetes is the universal substrate, and virtual machines are relegated to a dwindling “traditional systems” precinct of the application landscape. I understand the appeal of the story. It is clean. It has vision. It is also, of course, untrue.
The majority of enterprise compute still runs on VMs. Not as a legacy embarrassment awaiting migration, but as the deliberate, still irreplaceable foundation of how the enterprise operates. Before I head to Amsterdam for KubeCon + CloudNativeCon Europe 2026, I want to make that case.
Virtualization in Context
The most interesting infrastructure conversation happening right now is not “VMs vs. containers,” but how to manage both well together.
There is a persistent conflation in our industry of “old” (or “legacy,” “traditional,” etc.) and “ready to be replaced.” Virtual machines are certainly old, in the sense that they have been core infrastructure for decades. That’s because VMs deliver resource efficiency, strong security isolation, and service continuity in ways that containers do not replicate.
Containers share an OS kernel. That is a feature; it is precisely what makes them lightweight and fast. But it is also a foundational security issue that organizations in regulated industries and/or anyone running sensitive workloads cannot simply wave away. A VM is an isolated environment, unlike a container.
Containers and VMs are not really comparable choices on a shared spectrum; they are tools with different virtues, suited to different jobs. VMs are the load-bearing walls of the IT estate: You can design around them, add rooms, and modernize the kitchen. But you do not tear them out simply because you believe open floor plans are the future.
Nobody Is Actually Ripping Out Virtualization
I recently sat down with VMware for a briefing on Tanzu, VMware Cloud Foundation (VCF), and their broader cloud-native strategy. Their roadmap, naturally enough, reflects the reality that their customers are living in: a world where virtualization and Kubernetes coexist and are (both) undergoing rapid development on the same plane—and must be managed with that in mind.
The work VMware has done to make VCF a solid substrate beneath Kubernetes clusters is a direct response to this. Enterprises are not choosing between their VM estate and a K8s future. They are running both, and they need platforms designed around that reality.
This is not a VMware-specific observation. There will be numerous projects and technologies on display at KubeCon connecting containers and VMs to resource and service management; signaling that the right question is integration, not elimination—an insight broadly in practice today.
VMware’s presence at KubeCon Europe this year might be worth some attention, even if you aren’t a VMware customer or virtualization maven. They are a Platinum Sponsor and bringing substantive technical content beyond the booth demos. On the platform engineering side, they are presenting a keynote at Platform Engineering Day on why Kubernetes node OS still shapes performance, security, and scale (a practitioner-level conversation). They’re featuring a demonstration of their Kubernetes Service (VKS) at their booth, running VM workloads through Kubernetes APIs via VM Service. Honestly, if you’re the ops side of DevOps at a typical enterprise, that is a pretty cool, helpful view.
VMware Tanzu is worth some understanding as well, best considered separately. Tanzu is a platform-as-a-service offer providing an opinionated, integrated environment for building, running, and managing containerized applications at enterprise scale. While not a security product itself, but it inherits the isolation properties of the VCF layer below it. The security case for VMs is real, and this functionality is a solution for enterprises that need to run containers without avoiding the isolation issue.
A Frame for KubeCon
KubeCon is, and should always be, an exploration and celebration of what the CNCF community has built and continues to build. I am only making the case here for making headspace for the roughly 70% of enterprise workloads that are not containerized. This is a reality to design for. The shops that will most commonly thrive over the next decade will be the ones that build thoughtful, integrated approaches to orchestrating all of their resources: VMs, containers, and whatever comes next.
Virtualization’s core virtues of resource efficiency, governance, and service continuity seem pretty distinct to me from containers’ core virtues of mobility and cloud-native scalability. They do not cancel each other out. They live in the same estate and must be comprehended and run with that in mind. My advice going into KubeCon: bring this holistic context with you. That question will tell you a lot about whether a vendor or project is worth your time.
