
Share
Cloud-native portability has always been more promise than guarantee. VMware wants a testable spec that proves an app will actually behave the same way across any Kubernetes cluster, and it's bringing its VM API to CNCF too.
Kubernetes was supposed to solve the "write once, run anywhere" problem for modern apps. In practice, anyone who's shipped containerized workloads across multiple clusters knows that's aspirational at best. VMware is now trying to fix that gap with an actual standard, and it's doing so from inside the Kubernetes community rather than bolting something on from the outside.
The company is leading a new Workload Conformance Working Group under the Kubernetes project, aimed at defining a shared baseline for how workloads should behave on any cluster. The group's reasoning, laid out on its announcement page, is pretty blunt about the current state of affairs: "Certified Kubernetes tests the cluster, and Kubernetes AI Conformance tests whether a cluster can run AI and ML workloads. Nothing comparable exists for the workloads themselves, so cluster operators still check each application by hand before routine maintenance or a Kubernetes upgrade."
That manual-checking problem compounds fast. Every operator ends up repeating the same due diligence, and every application maintainer fields the same compatibility questions from every environment they support. It's the kind of redundant toil that a shared spec is built to eliminate.
Prashanth Shenoy, VP of product marketing for Broadcom's VMware Cloud Foundation (VCF) division, told The Register the push stems from a real pain point: apps packaged to run well in public cloud don't always behave the same way once they land in a hybrid environment. That's a problem VMware has serious skin in, along with other hybrid-cloud players like Nutanix and Red Hat, all of whom sell tooling meant to run the same stack both on-prem and in the cloud. If customers can't trust that an app will run predictably across environments, the whole pitch for hybrid infrastructure gets shakier.
The working group is still at the "figure out what we're even testing" stage. Per the announcement, the scope under discussion covers security, resilience during disruptions, resource use, networking, storage, API usage, and observability. None of those areas have settled requirements yet, let alone agreed-upon test methods.
One ground rule is already locked in, though: every requirement in the eventual spec needs an objective, automatable test. As the group puts it, "every requirement needs an objective test that a test binary can run, so nothing subjective goes into the specification." That's a meaningful constraint. It means the working group can't just write a checklist of best practices; it has to produce something a CI pipeline could actually run and get a pass/fail result from.

Shenoy is betting that application vendors will get behind this, for a straightforward reason: they want to package software once and have reasonable confidence it runs the same way everywhere. That's the same logic that made Certified Kubernetes a useful signal for cluster compatibility. A workload-level equivalent would let app teams point to a conformance badge instead of re-litigating the same environment quirks with every customer's ops team.
It's worth being realistic about where this stands. The working group is brand new, the scope isn't locked, and there's no timeline for a 1.0 spec. Standards efforts like this can take years to mature, and plenty never reach broad adoption. But the framing, borrowing directly from the Certified Kubernetes and AI Conformance programs, gives it a plausible path: piggyback on infrastructure that already has community buy-in and extend the same testing philosophy up the stack to workloads.
VMware didn't stop at proposing a standard this week. It also moved its VM API into the CNCF Sandbox, the foundation's on-ramp for early-stage projects that might eventually graduate to incubating or graduated status. The CNCF already hosts KubeVirt, a project that lets you run and manage full VMs inside a Kubernetes cluster. VMware's VM API takes a related but distinct angle: it lets Kubernetes users spin up VMs across different hypervisors, including VMware's own ESXi, using a common API surface.
This isn't VMware's first attempt at making Kubernetes and VMs play nice. vSphere with Kubernetes, part of vSphere 7 back in 2020, was an earlier swing at folding container orchestration into the virtualization stack. That lineage, tangled up with the Pivotal Software acquisition and VMware's years of repackaging and rebranding its container tech, didn't exactly produce a breakout hit. The VM API is a narrower, more composable bet: instead of trying to own the whole stack, it offers a shared interface that other tools and hypervisor vendors can plug into.
Handing the VM API to the CNCF signals this isn't just a VMware vanity project. Developers still overwhelmingly prefer writing for containers, but ops teams frequently end up running those containers inside lightweight VMs anyway, for isolation, security boundaries, or compliance reasons. Containers and VMs aren't going to merge into one abstraction anytime soon. They're going to keep coexisting, especially as organizations run new containerized services alongside legacy VM workloads, all within an infrastructure-as-code world that assumes everything gets deployed and managed automatically.
For practitioners, the near-term impact is limited, both efforts are early-stage. But the direction is worth tracking. A workload conformance spec, if it lands with real test tooling, could cut down the manual validation that currently eats up time during every Kubernetes upgrade cycle. And the VM API, now under neutral CNCF governance, gives teams running mixed VM-and-container environments another standardized option instead of a vendor-specific one. Neither is a finished product yet, but both point at the same underlying problem: Kubernetes won though orchestration, but it still hasn't fully solved for consistency at the workload level, and that's the gap VMware is now racing other hybrid-cloud vendors to close.
Tags
Original Sources
VMware floats new standard to make sure apps don’t break in hybrid K8s environments
↗ https://www.theregister.com/virtualization/2026/11/10/vmware-floats-new-standard-to-make-sure-apps-dont-break-in-hybrid-k8s-environments/5308956
About the author
Kai built ML infrastructure at a Bay Area startup before developing an obsession with transformer architectures and inference optimisation that eventually pulled him out of product work entirely. A stint at a compute research lab sharpened his instinct for what actually matters in a model release versus what is marketing. He writes from the inside — from the perspective of someone who has debugged the systems he is describing at three in the morning. He is allergic to hype and instinctively drawn to the unglamorous plumbing questions that everyone else skips over.
More from The Engineer →This Week's Edition
11 November 2026
14 articles
Related Articles

Tether Evo's New Research Tackles BCI's Calibration Problem with Cross-Subject Models
Models & Research · 5 min

Why Some Engineers Think the Ocean Floor Is the Future of Data Centers
Environment & Climate · 5 min

Europe's Digital ID Push Tests a New Privacy Tool: Zero-Knowledge Proofs
Policy & Regulation · 5 min
Related Articles

Tether Evo's New Research Tackles BCI's Calibration Problem with Cross-Subject Models
Models & Research · 5 min

Why Some Engineers Think the Ocean Floor Is the Future of Data Centers
Environment & Climate · 5 min

Europe's Digital ID Push Tests a New Privacy Tool: Zero-Knowledge Proofs
Policy & Regulation · 5 min
More Stories
© 2026 Cedar & Bloom. All rights reserved.