VMware Alternatives: What Actually Replaces vSphere

Broadcom closed its VMware acquisition in November 2023. By early 2024 perpetual licences were gone, the catalogue had collapsed from more than 160 products into a handful of bundles, and the pricing metric had moved from per CPU socket to per physical core.

Two years on, most estates have now been through at least one renewal under the new model. Some found it manageable. Others opened a quote several times larger than the one before it and started looking for the exit.

This is a survey of what actually replaces vSphere, including the option of not replacing it. Six paths, what each one is good at, and what it costs you in ways that do not appear on the invoice.

Why teams are looking for a VMware alternative

Four changes stacked on top of each other, and the interaction between them is what produces the headline numbers.

Perpetual licensing ended. Where an organisation once bought a licence outright and paid annual support at roughly 20 to 25 percent of the licence cost, it now pays a subscription every year in perpetuity. For anyone who had already bought their licences, this is a permanent addition to annual IT spend, not a one-time increase.

The unit of measurement changed from CPU socket to physical core, with a minimum of 16 cores charged per CPU. A host with two eight-core processors is billed for 32 cores whether or not those cores exist. Estates built on lower core-count hardware, particularly edge and branch deployments, absorb this badly. Redress Compliance, a licensing advisory, estimates that core minimums add 10 to 30 percent of phantom cores on hosts with low core-count CPUs.

Bundling replaced item-by-item purchasing. VMware Cloud Foundation packages compute, storage, networking and management together, which is good value if you already run vSAN and NSX, and poor value if you bought vSphere alone and never wanted the rest. VCF currently lists at $350 per core per year, down from the $700 Broadcom set immediately after the acquisition. vSphere Foundation, the smaller bundle, has no published list price at all, and advisor estimates of the going rate vary widely enough that you should treat any single figure as indicative rather than a benchmark to negotiate against.

And the version ceiling moved. vSphere Standard and Enterprise Plus are frozen at vSphere 8 Update 3. There is no version 9 of either edition, so anyone wanting vSphere 9 has to move to VVF or VCF regardless of what they need.

Reported increases vary enormously depending on prior discounting and estate shape. Published figures range from 150 percent to over 1,000 percent, and renewal quotes running three to ten times the previous perpetual-plus-support cost are widely reported. That spread is signal rather than noise. It reflects how differently the four changes above land on different hardware.

There is a date attached to all of this. General support for vSphere 8 ends on 11 October 2027. vSphere 7 support already ended in October 2025. If you intend to evaluate alternatives instead of accepting a quote under time pressure, the decision window is 2026 and early 2027.

Proxmox VE

The most common first answer, and a genuine fit for a narrower set of teams than the enthusiasm around it suggests.

Proxmox VE is an open source virtualisation platform built on KVM for virtual machines and LXC for containers, with clustering, live migration, high availability, and its own backup product. The community repository is free. A subscription buys you the enterprise repository and support.

What it does well: storage. Proxmox integrates Ceph and ZFS natively, and for teams willing to design their storage properly the result is capable and genuinely inexpensive. Live migration works. Clustering works.

What you take on is the part that gets underestimated. Proxmox gives you fewer guardrails than vSphere. Cluster quorum and fencing need to be designed rather than assumed, upgrades need planning, and storage rebuilds need someone who understands what they are rebuilding. None of that is difficult for a team that enjoys infrastructure. All of it is a problem for a team that had VMware precisely so they would not have to think about it.

The honest test is whether the person who would set it up is still going to be there in three years. Capability was never the question.

Nutanix

The closest thing to an enterprise-equivalent replacement, and the vendor that has moved most aggressively to capture VMware accounts.

Nutanix is hyper-converged: compute, storage and virtualisation in one platform, with its own AHV hypervisor. The architectural difference from a vSphere estate matters. Rather than separate hosts, a SAN and a virtualisation layer stitched together, storage is distributed across the same nodes that run the workloads. For teams that were already heading toward vSAN, this is familiar territory. For teams running traditional shared storage, it is a change to how capacity is planned, expanded and failed over, and the operational model changes with it.

Migration tooling from vSphere is mature because Nutanix has spent years building it for exactly this moment. Move handles the conversion, and the company has invested heavily in making the path from ESXi to AHV as low-friction as anyone offers. If your primary constraint is completing a migration inside a support deadline with a vendor accountable for it, this is the option with the most institutional weight behind it.

The caution is that switching to Nutanix is a platform decision, not a cost decision. It is enterprise software with enterprise pricing, sold on a subscription model with its own core and node-based metrics, and organisations that assume moving off VMware automatically means paying less are sometimes surprised. There is also a lock-in question worth asking out loud: you are replacing dependence on one integrated stack with dependence on another, and the exit from Nutanix in five years is the same conversation you are having about VMware now. Price it properly, and price the exit too.

Microsoft Hyper-V

For Windows-heavy estates this is often the answer before anyone finishes asking the question.

Hyper-V ships with Windows Server. If you already licence Windows Server Datacenter across your hosts, which many Windows-heavy organisations do because it grants unlimited Windows guest VMs per host, the hypervisor is effectively already paid for. That licensing interaction, more than any technical comparison, is usually what decides it.

The management story has two paths and they are worth separating. System Center Virtual Machine Manager is the traditional answer, familiar to Windows administrators in a way that neither Proxmox nor OpenStack will be, and licensed separately at a cost that erodes some of the saving. Azure Stack HCI, now sold as Azure Local, is where Microsoft is actually investing, and it is billed per core per month rather than bought outright. Teams assuming Hyper-V is free because Windows Server is already licensed sometimes discover the management layer is where the money went.

Where it gets awkward is mixed estates. Linux guests run on Hyper-V and run acceptably, but tooling, documentation and community knowledge all assume Windows. If a meaningful share of your workloads are Linux, you are choosing a platform optimised for the other half of your environment, and the day-to-day management experience reflects that in small frictions that accumulate.

Managed OpenStack

This is where we sit, so read the next three paragraphs with that in mind. Our own pitch is at the end of the OpenStack section rather than in a separate one at the close, which keeps it in the place where you can weigh it against the alternatives.

OpenStack is the open source infrastructure platform that underpins a large number of public and private clouds. It runs KVM, supports live migration, and has no hypervisor licensing attached to it. It also has a genuine operational learning curve, which is why most organisations that run OpenStack successfully either have a dedicated platform team or have someone else run it.

Managed OpenStack removes the second problem. Someone else operates the control plane, the storage layer, the patching and the upgrades, and you consume VMs through an API and a dashboard. The trade is that you give up some control over the platform layer in exchange for not staffing it.

For teams leaving VMware, the relevant part is that the workload shape is similar. You are still running VMs on dedicated hardware with live migration and shared storage. The management model changes, the operational responsibility changes, and the licence cost per core disappears. Windows guests run fine, though the guest tooling changes: VirtIO drivers replace VMware Tools. We have a walkthrough of that in our knowledge base if you want the detail.

Disclosure on what we sell: InMotion Cloud is managed private cloud on OpenStack, on dedicated hardware in US data centres, at a flat monthly rate, with no per-core hypervisor licence and VMware migration included rather than quoted separately. It suits organisations that want their VMs on dedicated hardware without staffing a platform team, and it does not suit teams who want to control the hypervisor layer themselves. Those teams should be looking at Proxmox. Our pricing calculator will model your estate against your current core count.

Moving to public cloud instead

Worth taking seriously, and frequently the right answer for smaller estates.

If your VMware environment is three hosts running a dozen VMs with variable load, the case for rebuilding a virtualisation platform is weak. AWS, Azure or Google Cloud will absorb that workload without you operating anything, and the elasticity is real value if your demand actually varies.

The case weakens as the estate grows and as utilisation steadies. A hundred VMs running consistently at high utilisation will cost more in public cloud than on dedicated hardware, because you are paying an elasticity premium on capacity you never release. We wrote about where that line sits in when private cloud beats public cloud, and when it doesn't.

Lift-and-shift also tends to disappoint. VMs moved unchanged into public cloud carry their original sizing with them, and the sizing was done for hardware you owned, not capacity you rent by the hour.

When staying on VMware is the right call

Sometimes the renewal is the correct answer and the migration is the expensive mistake.

Small estates on modern high core-count hardware often see a manageable increase, because the 16-core minimum costs you nothing when your CPUs have 32 cores anyway. If your quote came in at a number you can live with, the rational move is to sign it and spend the migration effort on something that generates revenue.

Deep tooling investment is the second case. Organisations with years of automation built against vCenter APIs, backup products integrated at the vSphere layer, and staff certified on the platform are carrying a switching cost that never shows up in a licence comparison.

Timing is the third. If your hardware refresh is eighteen months out, migrating twice is worse than renewing once. Renew short, refresh, then decide.

And there is negotiation, which most organisations underuse. Bundle tier is the largest lever available: moving from VCF to vSphere Foundation, where the vSAN and NSX components are not being used, changes more money than most discount conversations. Consolidating onto fewer, denser hosts reduces licensed core count directly, and Redress Compliance reports organisations recovering 15 to 25 percent of an increase this way before any discount is applied. A quote you have not modelled against alternatives is a quote you cannot negotiate.

How to choose an ESXi or vSphere alternative

Five questions, in the order that actually decides it.

How many hosts? Under five, public cloud or managed infrastructure usually beats rebuilding a platform. Above twenty, the economics of running your own start to work if you have the people.

What is the OS mix? Predominantly Windows points at Hyper-V. Predominantly Linux opens up Proxmox and OpenStack. Genuinely mixed favours the platform-neutral options.

Who operates it? The honest version of this question is whether you have someone whose job is infrastructure, or someone whose job is something else and who also looks after infrastructure. Self-managed platforms need the first.

What are the compliance constraints? Single-tenancy requirements, data residency, and audit evidence obligations narrow the field quickly, and they usually rule out shared public cloud tenancy before anything else.

How long is the migration window? Under six months, reduce the scope. A phased migration that moves the easy third first is more likely to finish than a complete replatform on a deadline.

What a migration actually involves

At a high level, four things.

Inventory and sizing, which is where most of the surprises live, because the RVTools export usually shows more VMs than anyone expected and a meaningful number nobody can identify an owner for.

Conversion, which is mechanical. Disk format conversion, guest tooling replacement, and driver changes.

Network reconfiguration, which is where the time actually goes. VLANs, firewall rules, load balancer configuration and anything that was set up by hand five years ago.

Then testing and cutover, workload by workload, starting with the ones that can tolerate being wrong.

Frequently asked questions

Is Proxmox production ready?
Yes, and it is running production workloads at significant scale. Maturity was settled years ago. The open question is operational fit, because Proxmox expects more from the team running it than vSphere does, particularly around storage and cluster design.

What is the cheapest VMware alternative for a small business?
For a small estate, the cheapest option is usually not a hypervisor at all. Managed infrastructure or public cloud removes the platform operations entirely, and at three or four hosts the operational cost of self-managing outweighs the licence saving. If you do want to self-manage, Proxmox VE on the community repository is free.

Can I keep running Windows VMs after leaving VMware?
Yes, on every option listed here. Guest tooling changes, so VMware Tools is replaced by the equivalent for the target platform, and Windows licensing follows Microsoft's terms rather than the hypervisor's. Check the Windows Server licensing position separately, because it is independent of this decision.

Do I lose vMotion if I leave VMware?
Live migration exists on Proxmox, Hyper-V, Nutanix AHV and OpenStack. The feature names and the exact behaviour differ, and shared storage requirements differ with them, but the capability was never VMware-specific.

How long does a VMware migration take?
It depends far more on your network configuration and application testing than on the number of VMs. Inventory and conversion are predictable. Rebuilding firewall rules and validating applications is what sets the timeline.

Can I just keep running my existing perpetual licences?
The licence remains valid. Once your support contract ends you lose patches, updates and support, which for most production environments is not a sustainable position.