VMware alternatives in 2026: what the companies actually leaving are doing

The most useful number in the VMware exodus is not the one in the headlines. In a January 2026 survey of 302 North American IT decision-makers at enterprises of a thousand people or more, commissioned by CloudBolt, 86% said they were reducing their use of VMware. In the same survey, only 2% had moved more than three quarters of their environment, and 44% had moved a quarter or more.

So the honest description of the market is not an exodus. It is a long, partial, workload-by-workload withdrawal that most organisations are two years into and nowhere near finishing. CloudBolt's own phrase for it is a calculated retreat. If you are somewhere in the middle of that and feeling behind, the data says you are average.

This post is about what the people actually doing it are choosing, what the renewal arithmetic looks like, and — the part most vendor content skips — which workloads are worth moving first and which are worth paying to leave alone.

What changed, factually

Broadcom announced the end of availability of VMware perpetual licences on 11 December 2023. Everything since is a consequence of that single decision: the portfolio consolidated into subscription bundles built around VMware Cloud Foundation, with vSphere Foundation as the smaller tier, and licensing moved to a per-core model with a minimum core count charged per CPU socket.

What it costs is harder to state cleanly, because Broadcom does not publish a price increase and the figures in circulation come from consultancies with a commercial interest in the renewal. Treat these as indicative rather than official: Redress Compliance, a negotiation consultancy, reports VCF list pricing in the region of $350 to $400 per core per year, multi-year discounts of 18 to 38 per cent, and first quotes arriving at two to four times a customer's prior vSphere run rate — that last figure being their own observation across 35 to 50 renewals, not a vendor statement.

The buyer-side numbers are better sourced. In the CloudBolt survey, 59% had seen increases of more than 25%, and 14% had seen their prices double. 88% were worried about future increases. Those are reported experiences from a stated sample, which is more than can be said for most of what is written about this.

If you are mid-renewal, the single most valuable thing on this page is that spread. A first quote at two to four times your old run rate is normal, and settlements land well below list. Do not treat the opening number as the price.

Where the workloads are actually going

From the same survey, with respondents able to pick more than one destination:

• Public cloud IaaS — 72%. By a wide margin the most common answer, and the one that changes the most about how the estate is run.
• Hyper-V or Azure Stack — 43%. The like-for-like move, and the path of least retraining for a Windows-centric shop.
• SaaS replacements — 34%. Not a migration at all: retiring the workload rather than rehoming it, which is the cheapest exit available where it is possible.

Proxmox and Nutanix dominate the comparison searches and the practitioner conversation, but note what the survey says: the most popular destination is not another hypervisor at all. The plurality answer is to stop running a hypervisor yourself. That is a bigger operational change than a platform swap and it is the one most likely to be under-costed.

The three exits, and what each actually costs you

Like-for-like hypervisor swap

Proxmox, Hyper-V, Nutanix, OpenShift Virtualization. Lowest conceptual risk: the virtual machines stay virtual machines. The costs that bite are the ones around the hypervisor rather than in it — backup tooling that must be re-certified, monitoring and automation rebuilt, storage and networking redesigned where vSAN or NSX were doing work you had stopped noticing, and a team that knows one platform deeply now knowing a new one shallowly.

Right when you have a large estate of long-lived VMs, a hard renewal date, and no appetite to re-architect applications at the same time.

Rehost into public cloud

The most-chosen destination and the one with the widest range of outcomes. A straight lift-and-shift of VMs into EC2 or Azure VMs is achievable on a renewal timetable and will usually cost more to run than the VMware estate did unless somebody does the rightsizing work afterwards. The saving is not in the move, it is in what the move makes possible: autoscaling, managed services, turning things off at night.

Right when you were already heading to cloud and the renewal has brought the date forward. Wrong as a pure cost play, on its own, with nobody owning the bill afterwards.

Retire the workload

A third of respondents are replacing something with SaaS rather than moving it. This is the cheapest exit there is and the easiest to overlook, because it does not look like an infrastructure project. Worth doing the inventory before the migration plan: every VM you decommission is one you do not have to license, move, test or support.

What to move first, and what to leave

The sequencing advice that holds up, and it is not what a migration vendor usually says: move the easy and the expensive first, and be prepared to pay to leave the hard ones where they are for another term.

1. 01Non-production first. Dev, test, staging and build infrastructure carry the licence cost without the risk. Moving them proves the target platform and reduces the core count on the renewal at the same time.
2. 02Then anything already stateless or containerised. If it is behind a load balancer and does not hold data, it is a weekend rather than a programme.
3. 03Then the cost outliers. The handful of oversized VMs driving the core count. The core-minimum model means a few fat hosts can be a disproportionate share of the bill.
4. 04Leave the regulated and the deeply integrated for last, and plan on at least one more renewal term covering them. Greyhound Research's Sanchit Vir Gogia makes the point precisely: VMware sits underneath ERP, clearing, healthcare and manufacturing control systems, which makes exiting it a multi-year programme rather than a procurement switch.

The corollary is uncomfortable and worth saying out loud: a partial migration that leaves you paying for both platforms during the transition is the normal case, not a failure. Budget for the overlap rather than discovering it.

Why migrations stall

The survey asked what would stop a migration. The answers were migration complexity and risk (25%), costs higher than expected (23%), and technical limitations (21%) — though 62% still called the obstacles challenging but doable.

Two softer findings explain more than those percentages do. 63% of organisations had changed their VMware strategy two or more times in two years, and 41% reported increased executive pressure to act — most often from the chief executive (19%), the board (16%) and the chief financial officer (15%).

That combination is the real failure mode. A technically sound multi-year plan meets a board that wants the line item gone this year, the strategy changes, the team restarts, and two years later a quarter of the estate has moved. If you recognise your own organisation in that, the useful intervention is not a better target platform. It is a written sequencing plan with the overlap cost in it, agreed with the people applying the pressure.

There is no such thing as a frictionless migration. — Matt Kimball, Moor Insights & Strategy, quoted in CIO

Where we fit, stated plainly

We are not a VMware migration product and we will not tell you we have a tool that makes this painless. What this work actually needs is competent infrastructure engineers for a defined period — people to do the inventory, build the target platform, rebuild the automation and backup around it, move workloads in waves and keep the lights on while two platforms run in parallel.

That is a staffing problem with a deadline attached, which is what we do. The honest version of the pitch: this is a capacity problem you cannot solve by hiring, because by the time you have hired the renewal has passed. Published rates are on the pricing page so you can model it before talking to anyone.

And the case against using us: if your estate is small enough that one experienced contractor could do it, hire the contractor. If the hard part is your regulated workloads rather than the volume, you need an architect and a compliance conversation more than you need hands. We would rather say that now than three months in.

Sahil Dubey

Group Managing Director

Runs engineering and compliance across the group. Twenty years in infrastructure, most of it on the operations side of systems that were already in production when he inherited them.

• CISA
• AWS Solutions Architect – Professional
• Microsoft Certified: Azure Administrator Associate