What are the alternatives to VMware after the Broadcom acquisition?
The alternatives to VMware fall into four groups: another hypervisor such as Proxmox VE, Microsoft Hyper-V, Nutanix AHV, XCP-ng or OpenStack; virtual machines on Kubernetes with KubeVirt; containerizing applications and running them on Kubernetes; and moving workloads to a public cloud. Most organizations combine two of them, because the right destination for a legacy Windows VM is rarely the right one for an application a team still develops. The search for alternatives started after Broadcom bought VMware in November 2023 and moved it to subscription-only, per-core licensing.
What changed at VMware under Broadcom
| Date | Change | Source |
|---|---|---|
| November 22, 2023 | Broadcom completes its acquisition of VMware. | Broadcom press release |
| December 11, 2023 | Sales of perpetual licenses and of new support contracts for them end. The portfolio is consolidated into two main subscriptions, VMware Cloud Foundation (VCF) and VMware vSphere Foundation (VVF). | The Register |
| Current terms (November 2026) | Subscriptions are licensed per physical core, with a minimum of 16 cores per CPU even on smaller CPUs. | Broadcom KB 313548 |
| February 2024 | The free ESXi hypervisor is discontinued. It returns with ESXi 8.0 Update 3e on April 10, 2025, for non-production use and without support. | The Register |
| April 2025 | A minimum order of 72 cores is announced for April 10, then withdrawn within days. The 16-cores-per-CPU minimum remains. | heise |
Perpetual licenses that customers already own keep working, but they can no longer get new support contracts, so patches and support end with the last contract. From then on, the same software is a recurring per-core subscription sold in bundles, which can include products a customer does not use. The date of that renewal sets most migration timelines.
The four types of alternative
| Path | Examples | What stays the same | What changes |
|---|---|---|---|
| Another hypervisor | Proxmox VE, Hyper-V, Nutanix AHV, XCP-ng, OpenStack | VMs, guest operating systems, the operating model | Management tools, storage, backup integrations |
| VMs on Kubernetes | KubeVirt, OpenShift Virtualization, Harvester | VMs and guest operating systems | Everything around them: APIs, networking, storage, skills |
| Containerize | Kubernetes on your hardware or in a cloud | Application code | Packaging, deployment, operations |
| Public cloud | AWS, Azure, Google Cloud and European providers | Applications, if rehosted as VMs | Cost model, networking, data location |
Another hypervisor
A hypervisor swap is the smallest change for teams that run hundreds of unrelated VMs. Proxmox VE is an open-source platform that combines the KVM hypervisor with Linux containers (LXC) and integrates Ceph and ZFS storage. Since version 8.2 it includes an import wizard for VMware ESXi that connects to an ESXi host or vCenter and copies VMs over. Hyper-V is the hypervisor built into Windows Server and is a common choice for Windows-heavy estates. Nutanix AHV is the hypervisor that ships with Nutanix’s hyperconverged platform. XCP-ng is an open-source Xen-based platform managed with Xen Orchestra and supported commercially by Vates. OpenStack fits organizations that want to run a private cloud with self-service APIs at large scale, and it needs the largest team of the five.
For how Proxmox relates to Kubernetes, see Proxmox vs Kubernetes.
Virtual machines on Kubernetes
KubeVirt adds virtual machines to Kubernetes as custom resources, so VMs and containers run on one control plane. It became a CNCF graduated project on November 10, 2026. OpenShift Virtualization from Red Hat and the open-source Harvester platform are built on it. This path fits organizations that already run Kubernetes and want to retire a separate VM platform. Running VMs on Kubernetes with KubeVirt covers how it works and how it compares with Proxmox.
Containerize applications
Applications your teams still develop gain the most from moving to containers: faster deployments, autoscaling and portability between data centers and clouds. Containerizing takes engineering time per application, so it rarely covers a whole VM estate. Packaged software from vendors that only support a VM install stays on a hypervisor. The difference between containers and virtual machines explains what changes for an application.
Public cloud
Rehosting VMs to a public cloud removes hardware from the equation. It suits workloads with variable load or a planned data center exit. It moves costs from licenses to hourly compute, storage and egress, and it puts data in a provider’s region, which matters for regulated data. Whether to run on your own hardware or in the cloud at all is covered in Kubernetes on-premises vs cloud.
How to choose
| Criterion | Question to ask |
|---|---|
| Number of VMs | Can the team migrate them before the next renewal date, or does a short renewal buy time? |
| Windows workloads | How are Windows Server and application licenses counted on the new platform? |
| Storage | What replaces vSAN or the current SAN: Ceph, a hyperconverged platform, or the existing array through a new driver? |
| Backup | Does your backup software support the target hypervisor or KubeVirt? |
| Networking | Which NSX features, such as micro-segmentation, need a replacement? |
| Skills | Does the team know Linux and KVM, Windows Server, or Kubernetes best? |
| Migration tooling | Is there an import path: the Proxmox ESXi importer, Forklift for KubeVirt, or a cloud provider’s migration service? |
A common sequence is to sort the VM estate into three lists: VMs that move as they are to another hypervisor, applications worth containerizing, and workloads that leave for a cloud or for retirement. Each list then gets its own target, and the VMware renewal covers only what is left during the move.
How Cloudfleet fits
Cloudfleet Kubernetes Engine (CFKE) is a managed Kubernetes service whose clusters accept any Linux server as a node, including bare metal, VMware vSphere and Proxmox VE virtual machines, next to nodes in public clouds. Teams that containerize applications during a VMware exit can run them on their existing hardware without operating a control plane. See self-managed nodes for the requirements.