Your VMware renewal is the moment to leave.
Broadcom ended perpetual licences and raised prices around 300%. We audit your VMware estate, migrate it to a sovereign open-source platform — Proxmox or KVM — and run it for you inside the EU. No per-core meter, no new lock-in, and no renewal you did not want to sign.
Virtualization is running virtual machines on a hypervisor that shares one physical server across many isolated workloads. In 2026 the live question for most organisations is not which hypervisor to adopt but how to leave VMware, after Broadcom replaced perpetual licences with subscriptions and raised prices steeply. Argus Root runs the exit end to end: we audit your VMware estate, migrate the machines and storage to a sovereign open-source platform such as Proxmox or KVM, and operate it for you inside the EU — without trading one vendor's lock-in for another's.
In short
- Broadcom ended perpetual VMware licences in January 2024, collapsed 168 products into ~4 subscription bundles, moved to per-core pricing and raised list prices roughly 300%; VMware held about 70% of the on-premises hypervisor base before the deal.
- The exodus is real and measured: Proxmox evaluations up ~340% year on year, XCP-ng ~180%, Proxmox enterprise subscriptions +60% in a year, and Nutanix reporting 40% of bookings as VMware displacement.
- The virtualization software market is heading from ~$110B in 2026 to ~$224B by 2031 (15.3% CAGR).
- The migration wave is slow — most estates are 12–24 months into planning, not executing — so the window to move before your next renewal is open now.
- Proxmox VE (KVM + LXC, Ceph, built-in HA, no licence fee) is the default destination for sovereignty-minded EU teams; Nutanix, Hyper-V and XCP-ng fit specific cases.
Why is everyone leaving VMware?
The reason is economic, and it is blunt. When Broadcom acquired VMware it discontinued perpetual licences, so there is no longer a version you buy once and keep. The product catalogue, once more than a hundred and sixty separate offerings, was consolidated into a small set of subscription bundles, with standalone favourites such as vSAN and NSX folded inside them rather than sold on their own. Pricing shifted from a per-socket to a per-core basis, and list prices climbed by around three hundred percent for many estates. A modest three-node cluster that used to run at roughly eighteen thousand dollars a year became unaffordable for the large majority of smaller organisations overnight.
The scale of the disruption follows from VMware's position before the deal: it held close to seventy percent of the on-premises hypervisor base, so the change touched almost everyone running virtualised infrastructure. For mid-market and smaller firms in particular, the new model removed the predictable, ownable cost structure they had built their operations on and replaced it with a recurring bill they no longer controlled. That is what turned a grumble about pricing into an industry-wide exit, rather than the usual cycle of complaining and renewing anyway.
It is worth being fair about the other side. For large enterprises with deep NSX and vSAN deployments, staying on VMware for the next one to three years is often still the realistic path, because the feature surface is hard to replace at once. Broadcom has also made some concessions on support windows under pressure. But it has not reversed the subscription-only model, and for the broad middle of the market the maths now points firmly toward the door.
The exodus, in numbers — and how the move runs.
The migration is measurable, not anecdotal. Evaluations of Proxmox VE rose by around three hundred and forty percent year on year, the Xen-based XCP-ng by around one hundred and eighty percent, and Proxmox enterprise support subscriptions climbed roughly sixty percent in a single year as mid-market firms moved to low-cost, high-availability clusters. Among the commercial replacements, Nutanix reported that forty percent of a recent year's bookings were direct VMware displacements. Underneath the individual moves, the virtualization software market is forecast to grow from about a hundred and ten billion dollars in 2026 toward two hundred and twenty-four billion by 2031, and sovereign-cloud policy is part of what is driving it.
The pattern that works is disciplined rather than dramatic. Audit the estate fully before touching anything, so every host, machine, storage layout and network dependency is known. Choose the destination from those requirements rather than from a favourite. Pilot non-critical workloads on the target in parallel with VMware before committing. Then move in waves, leaving the storage migration — reliably the slowest part — properly resourced. Run that way, a VMware exit is an ordinary infrastructure project with a clear end, not the cliff-edge the headlines imply.
Proxmox, Nutanix or Hyper-V — which is right?
There is no single answer, which is why a generic recommendation is worth little. Proxmox VE is the open-source default for most teams leaving VMware, and the one that absorbed the majority of the refugees. It runs KVM virtual machines and LXC containers under one web console, with built-in high-availability clustering, live migration, Ceph scale-out storage and integrated backup, all without a licence fee, and its Datacenter Manager now provides vCenter-style central control across clusters and sites. For an organisation that values open standards, data sovereignty and a predictable cost base — small and mid-sized firms, service providers and EU bodies especially — it is usually the strongest fit.
The alternatives earn their place in specific situations. Nutanix AHV is the most enterprise-proven turnkey replacement for a full VMware Cloud Foundation deployment, with a single management plane and a one-click feel, but its cost can approach what you are leaving and it introduces its own lock-in. Hyper-V is the natural choice for an all-Windows estate already licensed for Windows Server Datacenter, since it can reuse existing hardware and integrate with Active Directory. XCP-ng, the open-source successor to Citrix XenServer, is strong where GPU passthrough and VDI matter. OpenStack and OpenShift Virtualization suit organisations that want a full private cloud or a path toward containers. We map your workloads, skills and sovereignty needs to the platform that fits, and say so plainly rather than steering you to a default.
What is the hardest part of the migration?
Storage, by a wide margin. Moving the virtual-machine disk images and rebuilding the storage configuration — datastores, classes, performance tiers — is consistently the most time-consuming part of a hypervisor migration, ahead of moving the machines themselves. After storage come the feature-parity questions: what replaces the network segmentation and distributed firewalling that NSX provided, and how the scale-out storage that vSAN handled is rebuilt, usually on Ceph, which is operationally different even when it is technically equal. Last comes the skills question, because Proxmox and OpenStack expect Linux and infrastructure fluency that the team running a click-driven vSphere may not have in depth.
None of those is a reason to stay; each is a reason to have the migration run by people who do it routinely. The mechanics of moving a single machine are well understood — export the VMware machine, import it onto the new platform, attach its disks to the target storage, set the boot and networking — and the work is in doing it reliably at scale, in the right order, with rollback at every step. The commands below show the shape of a single import onto Proxmox; the service is everything around them.
# export the VM from VMware as OVF, then import onto Proxmox $ qm importovf 142 ./web01.ovf local-zfs ← create VM 142 from the OVF $ qm importdisk 142 web01-data.vmdk ceph-pool ← move the data disk onto Ceph $ qm set 142 --scsihw virtio-scsi-pci --boot order=scsi0 $ ha-manager add vm:142 --group eu-prod ← add to the HA group VM 142 imported · disk on ceph-pool · HA group eu-prod # perpetual licence cost: zero. Broadcom renewal: declined.
Why is the window open right now?
The striking thing about the largest virtualisation migration in a decade is how slowly it is actually moving. Most organisations are twelve to twenty-four months into planning a move but have not begun executing it, because production migrations of large estates are long cycles that few teams want to start mid-renewal. That gap between intent and action is the opportunity. The renewal date is the real deadline, and starting a migration against it — rather than after signing another multi-year term under protest — is what determines whether you leave on your terms or Broadcom's.
There is a second reason the timing favours moving now rather than waiting. The alternatives have matured precisely because of the exodus: Proxmox, Nutanix AHV and OpenShift Virtualization have all shipped substantial feature updates since the acquisition, so the platform you migrate to today is materially better than the one available when the pricing changed. Waiting does not get you a cheaper VMware; it gets you closer to a renewal you did not plan for. The estates that move deliberately, ahead of the cliff, are the ones that avoid both the panic and the premium.
VDI and the Omnissa question.
Virtual desktop infrastructure deserves its own treatment, because the ground there has already shifted. VMware Horizon, the long-standing VDI platform, is no longer a VMware or Broadcom product: Broadcom sold the end-user computing division to the investment firm KKR in May 2024, and it now operates independently as Omnissa. If you run Horizon, in other words, your vendor relationship has changed whether or not you have acted on it, which is reason enough to reassess rather than drift.
We handle VDI as a separate strand from the server-virtualisation exit, because the options and the constraints differ. Graphics-heavy or engineering desktops, where GPU passthrough matters, are often well served by XCP-ng, while other estates may stay on Omnissa or move to a different approach entirely. The important discipline is to sequence the two migrations so that moving the server platform does not destabilise the desktops, and the desktop decision is made on its own merits rather than swept into the hypervisor project by accident.
What we run for you.
We deliver the exit as a managed service from end to end rather than a consulting report you are left to implement. It starts with a full audit of the VMware estate and a destination design chosen against your real requirements, sovereignty needs and renewal timeline. We then migrate in waves — piloting non-critical workloads first, moving machines and storage with rollback at each step — and stand up the target platform with the high availability, Ceph storage and backups the workloads need. Crucially, we keep operating it afterwards: clustering, live migration, patching and rolling upgrades on Proxmox or KVM are ongoing work, and they are the very skills whose scarcity stalled so many migrations.
Running it inside the EU is a design choice, not a detail. A hybrid layout keeps sensitive data on infrastructure you control, under EU jurisdiction, while the platform itself is built on open standards we operate directly with no resale layer in between. The resilience side ties into our backup and disaster recovery practice, the surrounding cloud and server work into managed cloud and server management, and the jurisdictional story into compliance and sovereignty. The exit is one move; the platform that results is something we own with you for the long run.
Sovereign by default, with no new lock-in.
The trap to avoid when leaving VMware is leaving it for another cage. Moving to a proprietary turnkey platform can simply swap VMware's lock-in for a new vendor's, with a fresh meter and a fresh roadmap you do not control. Building on Proxmox or KVM with Ceph and open standards keeps the platform open and portable: no per-core licence, no single vendor owning your future, and the freedom to change operator without re-platforming. Avoiding the next lock-in is part of the reason to move, not a happy accident of it.
That openness is also what makes sovereignty real rather than a label. Infrastructure built on open software and operated inside the EU, with sensitive data held where you choose, is shielded from the jurisdictional and pricing exposure that drove the exit in the first place. It is the same principle behind everything Argus runs: a European operator that builds and runs the platform itself, on open foundations, and tells you the truth about what it costs and where your data sits — rather than reselling a black box and passing on the next price rise.
Questions buyers ask.
Why are so many organisations leaving VMware?
What is a VMware exit?
Which VMware alternative should we choose?
Is Proxmox really enterprise-ready?
What is the hardest part of a hypervisor migration?
How long do we have before our VMware renewal?
Will our applications need to be rewritten?
What about VDI — we run VMware Horizon?
Do we lose high availability and backup by leaving VMware?
Is this just trading one vendor lock-in for another?
How big is the move across the industry?
Can you keep some workloads on-premises for sovereignty?
Why run the new platform with Argus Root rather than ourselves?
Tell us your renewal date. We'll plan the exit around it.
We will audit your VMware estate, map each workload to the destination that fits — Proxmox, KVM, Nutanix or Hyper-V — and give you a migration plan and a cost picture set against your renewal date, not Broadcom's. You get a clear view of what the move takes, what it saves, and what the first wave would carry, before you commit to anything.