For a long time, VMware was IT infrastructure's version of "nobody gets fired for buying IBM." Not always the cheapest, but the one that came with the least explaining. Clients knew it, backup vendors supported it, hardware vendors certified for it, and if something went wrong there was a named vendor to call. That still mattered more than people admitted. The Broadcom acquisition changed the calculus. Perpetual licensing was discontinued in December 2023. Everything is now subscription-based, billed annually, with a 16-core minimum per physical CPU. Miss your renewal anniversary and there is a 20% penalty applied retroactively.
One episode is worth knowing about because it illustrates where things are heading. In early 2025, Broadcom announced a new 72-core minimum purchase requirement, effective April 10. For a 2-node cluster with 16-core processors, that would have meant paying for 40 cores you do not have. The backlash was sharp enough that they walked it back on the same day it was due to take effect, t…. The 16-core minimum remains for now. Whether that reads as a reprieve or as a signal of intent probably depends on how much goodwill you have left for the platform.
There is also a concrete deadline to keep in mind.vSphere 8 is the last version that could be perpetually licensed, with end of general support on Oc… . After that, perpetual licence holders who want continued vendor support have one path: move to a vSphere 9 subscription. vSphere 7 already went end of support in October 2025, so if you are still on 7, you are already running without a safety net. October 2027 is not a distant horizon for a migration that involves production workloads.
None of this means VMware has stopped working. The hypervisor has not degraded. If you are a large enterprise with hundreds of hosts and deep vCenter integration built over a decade, the cost of migration likely exceeds the licensing increase for at least the next few years. But for the SMBs and mid-market organisations we work with across the UAE and GCC, the numbers often no longer make sense, and we have had enough of these conversations to know what the real questions are.
The Two Alternatives Worth Talking About
There are others: XCP-ng, Nutanix AHV, OpenStack if you enjoy suffering. For most organisations we work with, the choice narrows to Proxmox VE or Hyper-V. Both are mature. Both have genuine limitations. Neither is a drop-in replacement for a fully loaded vSphere stack, and anyone telling you the migration is simple has probably not done very many of them.
Proxmox VE
We have been deploying Proxmox for over a decade and run it in our own managed private cloud, so this is not an armchair general's view of battle.
It is open-source, built on Debian, our favourite Linux distribution, using KVM for full VMs and LXC for Linux containers. Clustering, HA, ZFS, SDN, and a web UI are all included. Proxmox Backup Server is a separate free product handling incremental deduplicated backups and offsite replication. Version 9.1 is current. Proxmox Datacenter Manager 1.0 shipped in late 2025 and is beginning to close the vCenter management gap for larger deployments, though it is still early days.
The licensing model is per CPU socket, not per core. A 32-core single-socket server costs the same to license as an 8-core one. Coming from VMware's per-core model, that difference adds up fast. Optional paid subscriptions add the tested enterprise repository and support tickets, running from €115 to €1,400 per socket per year depending on tier. A 3-host cluster with 2 sockets per host at Premium support comes to roughly €8,400 a year. That is less than a third of VMware VVF at list price, and most environments land somewhere in the middle tiers.
Migrating from ESXi is genuinely easier than it was two years ago. The built-in ESXi Import Wizard connects directly to a source host and pulls VMs across. In practice though, there are things to sort before you start. Snapshots need to be consolidated on the VMware side first or the import crawls. vTPM state cannot be carried across, so BitLocker-encrypted VMs need encryption suspended before migration and re-enrolled after. Storage-policy-encrypted disks need decrypting first. vSAN-backed VMs have to be moved off vSAN to a regular datastore before the import will touch them. VLAN assignments do not port over, so network mapping is a manual rebuild. And VMware Tools needs to be replaced with QEMU Guest Agent post-migration for clean shutdown, IP reporting, and consistent snapshots to work properly. Windows VMs will boot on the emulated SCSI and NIC, but performance lags until VirtIO drivers are installed. That is a known step, not a surprise, but across a large VM inventory it adds up.
Backup coverage is no longer a concern. Veeam added native Proxmox VE support in version 12.2 in August 2024 and supports cross-hypervisor restore, meaning you can restore a VMware backup directly to a Proxmox host. Acronis and Nakivo both support it as well. If you are already running Veeam, the migration path is cleaner than it looks. One gap: Veeam's SureBackup Virtual LAB feature, used for automated recovery testing, does not yet support Proxmox as the sandbox environment. You still need a VMware or Hyper-V host for that specific function. For new deployments without an existing Veeam investment, Proxmox Backup Server is more capable than most people expect until they actually use it. Hardware compatibility is generally good. The Linux kernel supports a wider range of hardware than ESXi does, including older servers and consumer NICs that VMware dropped support for in later ESXi versions. If you have hosts that are no longer on VMware's HCL, Proxmox is often the easier path. SR-IOV, NVMe, and GPU passthrough all work in production. The limitations worth being honest about: there is no equivalent to DRS. Live migration exists but automatic workload balancing across hosts is not built in. The networking stack lives directly on the Debian side. There is no visual vSwitch equivalent in the GUI; bridges, bonds, and VLANs are configured through Linux networking, and if you do not know what you are doing there, you will feel it. Cluster management runs on Corosync, which works well on a local LAN but needs proper tuning for geographically dispersed nodes. Latency and packet loss on the cluster network will cause fencing events, and sorting that out after the fact is not a pleasant afternoon. Linux administration skills matter across all of this. If the team is entirely Windows-oriented, the learning curve is real. And for environments where vendor certification is a contractual compliance requirement, Proxmox does not carry the same posture as VMware.
Microsoft Hyper-V
The free standalone Hyper-V Server was deprecated in 2019. Hyper-V today is a role inside Windows Server, which means every host needs a Windows Server licence. Standard covers 2 VMs per host and Datacenter covers unlimited, with Datacenter running $6,000 to $7,000 per host. If you are running more than two or three VMs per host, Datacenter is the only realistic option, and for a 3-host cluster that is $18,000 to $21,000 in host OS licences before you have licensed a single guest workload.
The scenario where Hyper-V makes the most sense as a VMware exit is if you already hold Windows Server Datacenter licences with Software Assurance. In that case, the hypervisor costs nothing additional. That is more common than it sounds in organisations that are already running a Microsoft-heavy estate.
Migration from VMware is well-tooled. Microsoft's VM Converter has been around for years. Veeam handles cross-platform restore natively. Windows guest VMs migrate with less friction than to Proxmox because the driver model is familiar territory. Linux guests on modern distros handle Hyper-V integration services automatically; that kernel support has been upstream for a long time.
Backup coverage is Hyper-V's strongest point. Veeam, Commvault, Acronis, Veritas, Rubrik: every major enterprise backup vendor supports Hyper-V as a first-class platform. If your existing backup infrastructure is built around any of these, migration is straightforward on that dimension.
Support is also the clearest argument for Hyper-V. Microsoft Premier and Unified support are contractually backed, globally available, and well understood in the region. If a named vendor with defined SLAs is a hard requirement, Hyper-V is the only alternative hypervisor that genuinely matches VMware on that point.
Where it gets complicated: the Windows Server licensing cost is real, and without existing Datacenter licences with SA it can close the cost gap with VMware significantly. Hyper-V is Windows-centric and the management tooling, scripting model, and operational patterns all favour Windows administrators. There is no native ZFS. Azure Stack HCI is capable but adds subscription cost and is fundamentally an Azure on-ramp rather than a standalone private cloud option. And clustering beyond basic failover involves Windows Server Failover Clustering, which has its own operational model to learn.
How to Think About the Decision
If you already hold Windows Server Datacenter licences with Software Assurance, look at Hyper-V seriously before anything else. The migration path is mature, backup coverage is comprehensive, and your team probably already knows the operational model.
If the environment is mixed Linux and Windows, or if reducing vendor lock-in is part of the goal, Proxmox is the stronger option. The cost difference is real and the platform is ready for mid-market production workloads. The important thing is to treat it as a proper platform project rather than a hypervisor swap. Storage architecture, HA cluster design, backup strategy, network design, and runbooks all need to be in place before anything important cuts over.
If vendor certification is a contractual compliance requirement, neither option substitutes cleanly for VMware without a compliance assessment first. That conversation belongs before the infrastructure decision.
If your VMware contract is still active, do not rush. Price up the renewal under the new per-core model against your actual core count and use that as the real baseline. The 20% late renewal penalty is worth avoiding, so do not let the contract lapse while still deciding.
What We Do
We have been deploying and managing Proxmox in production environments in the GCC for several years, including our own. We also manage Hyper-V environments and work with clients through mixed-hypervisor estates. We are not pushing a particular platform. What we can do is assess your environment, price up the realistic options, and put together a migration plan that accounts for actual operational risk rather than the version that looks clean on a slide. If you are working through this, reach out at [email protected] or call us on +971-4-2142000.
