Two years into Broadcom's ownership of VMware, the conversation has changed. The per-core subscription model, the collapse of the perpetual license, and the consolidation of SKUs into VMware Cloud Foundation have pushed a lot of shops — mine included — to take Proxmox seriously for the first time. This is where I have landed.
The Pricing Reality
VMware Cloud Foundation 5.x is priced per core, with a 16-core minimum per CPU, billed annually. For a modest cluster — three hosts, two 16-core Xeons each — you are now writing checks that were quoted in the tens of thousands for what used to cost a few thousand on perpetual. Real numbers from a recent renewal I saw: roughly 3x the previous annual spend for the same hardware footprint.
Proxmox VE is free. The enterprise repository subscription is ~€110 per CPU per year for community support and is optional. For a three-host cluster that is ~€660/year. Not a typo.
Where Proxmox Matches ESXi
The features a typical mid-market shop actually uses are all present:
- Live migration between nodes in a cluster. Just works.
- HA restart on host failure. Configured via the Datacenter → HA panel.
- Snapshots and clones. Linked clones on ZFS are instantaneous.
- Shared storage via NFS, iSCSI, or Ceph. Ceph is baked in.
- Backups. Proxmox Backup Server is genuinely excellent — deduplicated, incremental-forever, and pairs with PVE natively.
- Hardware passthrough. PCIe, SR-IOV, vGPU with a NVIDIA vGPU license.
- Templates and cloud-init. First-class support.
- Terraform and Ansible providers. The
bpg/proxmoxTerraform provider is production-grade.
The Honest Gaps
Do not pretend these do not exist:
- DRS equivalent. Proxmox has no built-in load balancer that moves VMs based on CPU/memory heat. You can script it with ProxLB or cron + qm, but it is not native.
- NSX-class SDN. Proxmox has an SDN module (VLAN, VXLAN, EVPN) but it is not distributed firewall with intent-based policy. If you live on NSX micro-segmentation, you stay on VMware or move to Cilium on Kubernetes.
- vSAN equivalent beyond Ceph. Ceph is great, but it has a learning curve and wants dedicated networking. ZFS replication is simpler and covers two-node scenarios.
- Vendor ecosystem. Your backup vendor, your EDR console, your monitoring platform — most list VMware first, Proxmox eventually. Veeam support landed in 2024, which helped a lot.
- Phone-a-friend support. Proxmox Enterprise support is email-based, business-hours Vienna time. There is no 24x7 follow-the-sun with a TAM assigned to your account. Third parties like OSIT and Proxmox partners fill this gap if you need it.
Decision Framework
My rough rule for where each wins:
- Home lab and small business (1-3 hosts): Proxmox. Not close.
- Mid-market (10-50 hosts), no hard compliance mandate: Proxmox. Savings fund real headcount.
- Regulated enterprise already running vSphere + NSX + Tanzu: Stay on VMware until your renewal math forces the conversation. Migration cost is not trivial.
- Greenfield enterprise: Honestly, Kubernetes for new workloads, Proxmox for the lift-and-shift VMs. ESXi only if a platform-specific requirement demands it.
A Quick Migration Note
Proxmox 8.2 added a direct ESXi import wizard. Point it at vCenter, pick VMs, import. It handles VMDK-to-qcow2 conversion and fixes VirtIO drivers for Windows guests reasonably well. A 200 GB Windows VM moved in about 25 minutes over 10GbE in testing.
# Import from ESXi via CLI
pvesh create /nodes/pve01/storage/local/content \
--content import \
--filename esxi-import://vcenter.lab/VM-NAME
The answer is no longer "of course ESXi." For a lot of shops in 2026, it is "Proxmox unless you can articulate why not." That is a healthier question to be asking.