Proxmox VE vs VMware Cloud Foundation

How VCF licensing actually works

• VCF is licensed per physical core, per year, with order minimums that often inflate what you actually pay: 16 cores per socket and 72 cores per order. A 4-host cluster with 8-core CPUs is still billed as if it had 72 cores.
• Proxmox VE subscriptions are per CPU socket, per year, with no core counting, no order minimums, and no per-feature uplifts. The platform itself is fully functional without any subscription — subscriptions buy you the enterprise repository and Proxmox commercial support.
• Discounts on VCF depend on who you buy from. Broadcom's tiered partner programme means Strategic customers see roughly 52% off list, Pinnacle partners around 45%, Premier partners around 38% — and everyone else pays list. Resellers and distributors eat into their own margin to retain customers, so what you actually pay varies. The calculator above lets you enter your actual negotiated discount.

What the bundle actually costs you (beyond the line item)

The headline per-core price is only part of the story. VCF is a bundled suite — vSphere, vSAN, NSX, Aria, and VKS (vSphere Kubernetes Service, Broadcom's rename of Tanzu) — and you pay for the whole bundle whether you use every component. For most Australian organisations we work with, three things drive the real total cost:

1. Order minimums

The 16-core/socket and 72-core/order minimums hit smaller environments hardest. A two-host cluster running 8-core CPUs is billed as if every host had 16 cores. A new four-host edge cluster with 6-core CPUs is billed as 72 cores even though the physical hardware has 48. These minimums don't apply to Proxmox.

2. vSAN raw capacity entitlement

VCF includes a per-core vSAN entitlement (1 TB raw per billable core). Exceed it and you're charged overage at US$120/TB/yr. Proxmox uses Ceph (or ZFS, NFS, iSCSI, FC — whatever fits) and there's no storage-capacity-based licensing component at all.

3. Suite components you don't use

NSX, Aria, and VKS are in the bundle whether or not your team uses them. For the majority of mid-market organisations, NSX microsegmentation isn't load-bearing, Aria automation is unused, and VKS is not the K8s platform in production. You're paying for capability you don't run.

What about the features VCF is genuinely strong at?

A fair comparison acknowledges that VCF still wins on a narrow set of capabilities: deep NSX microsegmentation for organisations with mature east-west security policies, the Aria suite for shops that have standardised on it for operations and automation, and VKS for teams running it as their K8s platform. If any of these are load-bearing for you, the comparison is more nuanced than the cost numbers alone suggest. We talk through these trade-offs honestly in a discovery call.

What about CRS — doesn't Proxmox now have its own DRS?

Yes. Proxmox VE 9.2 introduced the Cluster Resource Scheduler, closing the long-standing DRS gap. Live load balancing across cluster nodes is now a native Proxmox feature, not a workaround.

What's the actual migration risk?

For most environments, a VCF-to-Proxmox migration is a methodical exercise rather than a leap of faith. We work it in phases: assessment first (what's actually running, what's truly load-bearing, what depends on VCF-specific features), then a parallel Proxmox build, then workload migration in scheduled batches with full rollback at every step. The pattern is well-trodden — see a recent case study.