VMware After Broadcom: Where Teams Are Migrating in 2026

Broadcom completed the VMware acquisition in November 2023 and immediately did what the industry expected: discontinued perpetual licences, forced a subscription model, consolidated 8000+ partners into a handful, and raised prices on most surviving SKUs by 5–10×. The customer reaction has played out across two years, and by mid-2026 the migration patterns are no longer speculative — they are observable in the order books of every alternative hypervisor vendor.

This is a field report from migrations we have run, plus the published evidence from peer studios and the major alternatives.

What the post-Broadcom contract actually looks like

For a customer running 200 cores of ESXi with vCenter, the pre-Broadcom annual licence was somewhere in the range of CA $40,000–80,000 depending on which SKUs (vSphere Standard / Enterprise Plus / vSAN). Under Broadcom’s VCF (VMware Cloud Foundation) bundle — which is now the default and often the only SKU available — the same fleet renews at CA $200,000–400,000 a year. Roughly 5× the previous run rate.

The contract structure changed too. Perpetual licences with optional support are gone. Customers now buy subscriptions in 1, 3, or 5-year terms, with the per-core minimum raised from 16 to 72 cores per CPU socket (in some SKUs). Many customers who used vSphere Essentials Plus on small clusters discovered their existing licence had no equivalent in the new catalogue at all.

This is not a price increase. It is a different product, repackaged, and Broadcom is comfortable telling that to renewing customers in writing. The reactions have been:

1. Migrate off VMware entirely. This is the dominant pattern for small-to-mid fleets (under ~500 cores).
2. Renew under protest and budget for migration in 18-24 months. Common for mid-to-large fleets where migration is multi-quarter work and the contract bridges the gap.
3. Renegotiate aggressively. Possible for large enterprises with leverage; rare for everyone else.
4. Stay and pay. Surprisingly common among teams who cannot articulate a migration plan their board will approve.

The 2026 migration destinations

By volume of fleets migrated, the four credible options are:

Destination	Best for	Annual cost (per host, rough)
Proxmox VE	Small-to-mid fleets, teams with Linux operations capacity	CA $0 (community) or ~$1,200/host (Enterprise subscription, optional)
XCP-ng + Xen Orchestra	Teams comfortable with Xen, want a graphical management plane	CA $0 community, optional support contracts
Nutanix AHV	Enterprises wanting hyperconverged + commercial vendor relationship	CA $5,000–15,000/host depending on tier
Hyper-V (Azure Stack HCI)	Microsoft-heavy shops with existing Azure footprint	Per-core licensing, integrated with M365 / Azure bills

Public cloud (AWS, Azure, GCP) is a parallel option, but a true cloud migration is a different project. Replatforming workloads from VMware to AWS EC2 is not the same engagement as replatforming from VMware to Proxmox; the latter preserves the operations model, the former changes it. Many teams do both — move dev/test to public cloud, keep production on an open-source hypervisor.

Proxmox VE: the workhorse migration

Proxmox VE has become the default migration destination for fleets under ~500 cores. The reasons:

1. It is genuinely free. The community edition has no licence cost. The Enterprise repository (stability-tested updates, support) is CA $250/host per year at the entry tier — for context, that is less than 10% of the Broadcom renewal for the same hardware.

2. Migration tooling is real. Proxmox 8.2 (released spring 2024) shipped first-class VMware migration support: import an ESXi storage pool directly via the Proxmox UI, with automatic VM conversion. Earlier versions required qemu-img convert and manual VMDK handling; the 2024 release made it a UI workflow.

3. The operations model is familiar. Proxmox is Debian Linux with a web UI on top of QEMU/KVM and LXC. Teams already running Linux do not need new skills. Teams coming from vSphere learn the new UI inside a week.

4. HA, backup, replication are first-class. Proxmox HA (high availability via Corosync), Proxmox Backup Server (deduplicated, incremental, off-site replication), and ZFS / Ceph as supported storage — these are not afterthoughts. Production-grade.

Where Proxmox is the wrong answer:

• You need NSX-equivalent software-defined networking. Proxmox SDN (introduced 2023, matured through 2024-25) covers VLANs, VXLANs, and basic overlay networking but is not NSX. If your VMware deployment depends heavily on NSX-T, Proxmox is a downgrade.

• Your operations team will not own a Linux substrate. Proxmox is Debian. Patching, kernel updates, ZFS pool management — these become your problem. For Windows-heavy shops without Linux capacity, this is real friction.

• You have a strict commercial-vendor procurement requirement. Some enterprises and regulated entities require named, contractable vendor relationships for infrastructure software. Proxmox Server Solutions GmbH offers support contracts, but their commercial relationship surface is smaller than VMware’s was or Nutanix’s is.

XCP-ng: the alternative for Xen shops

XCP-ng is a community fork of XenServer that, with Xen Orchestra (commercial support available), gives you a graphical management plane on top of the Xen hypervisor. The Xen Project is genuinely battle-tested — it powers significant portions of AWS EC2, among other things.

XCP-ng is the right choice when:

• Your team has Xen operations history.
• You want a vendor-supportable option (Vates publishes XCP-ng commercially) without VMware-scale costs.
• Your fleet is mid-range — too big for Proxmox community comfort, too small to justify Nutanix.

In our experience XCP-ng + Xen Orchestra runs ~CA $3,000–6,000/host/year for the commercial tier, considerably less than Nutanix but more than community Proxmox.

Nutanix AHV: the enterprise commercial option

Nutanix has captured most of the large-fleet exits from Broadcom-VMware. Their pitch is straightforward: a real commercial vendor relationship, hyperconverged infrastructure, AHV hypervisor as the included default. Per-host pricing is meaningfully higher than Proxmox, but considerably lower than the post-Broadcom VMware renewal — and the operations model resembles vSphere closely enough that vCenter-trained teams transition without much retraining.

If your board needs a vendor name on the contract and you cannot accept Proxmox’s commercial surface, Nutanix is the next-best answer. Hyper-V (Azure Stack HCI) is the alternative if you are already deeply invested in Microsoft licensing.

What we have actually migrated

Engagements we have run in 2024-2025:

• 3-host vSphere Standard cluster, ~40 VMs, mostly Linux web/app services. Migrated to a 3-node Proxmox VE cluster on the same hardware. Ceph for shared storage. Proxmox Backup Server on a dedicated 4th node, with off-site replication to a Hetzner box. Total migration: 6 weeks, including discovery, parallel-running phase, and cutover. Annual licence cost: CA $1,500 (3 × Enterprise subscriptions) vs the CA $35,000/year Broadcom quote.

• VMware-on-bare-metal lab + dev environment, ~12 hosts, hundreds of ephemeral VMs for CI workloads. Migrated to Proxmox community edition, no support contract. CI pipelines re-pointed at the Proxmox API. Total migration: 10 weeks because of the CI rewrite. Annual cost: CA $0 vs CA $80,000+ proposed VCF renewal.

• Mid-size hybrid migration: ~80 production VMs stayed on a (renewed) VMware contract through the 2025-2026 contract year, while dev / test / lab moved to Proxmox. The plan is to migrate production at next renewal, having proven the operations model on dev first. This is the pattern we recommend when the customer cannot afford a “big bang” cutover.

The hidden costs of migration

Two things consistently surprise teams during migration:

1. Storage replatforming is the real work. If you were on vSAN, you have to choose Proxmox’s storage stack — typically ZFS or Ceph. Both are good; neither is vSAN-equivalent in every dimension. Migrating data while preserving snapshots and reasonable downtime windows is a non-trivial engineering exercise. Budget 50% of the project for storage.

2. Backups need replatforming too. Veeam’s Proxmox support arrived in 2024 (Veeam Backup & Replication v12.1 added Proxmox VE support). For teams already on Veeam, this matters. Teams without a Veeam licence usually move to Proxmox Backup Server, which is excellent but a different operations posture.

We have not had a single migration where the Broadcom renewal was actually cheaper than the migration cost over a 3-year horizon, including engineering time. That is a strong statement, and the math gets stronger as the fleet grows.

What to do this quarter

If your VMware renewal is in the next 12 months:

1. Get the renewal quote in writing now. Many teams discover the quote only at renewal, with no time to evaluate alternatives.

2. Pilot Proxmox on lab or dev infrastructure. Two or three weeks of Proxmox operation, on hardware you do not depend on, gives you a real basis for the production migration plan.

3. Cost-model 3 years. Year-1 migration is more expensive than renewal. Year-2 and 3 are dramatically cheaper. Most boards approve migrations when they see the 3-year total.

4. Decide on storage early. ZFS, Ceph, NFS-from-existing-NAS — the choice shapes everything downstream. We default to ZFS on local NVMe + Proxmox Backup Server for backup, with Ceph for clusters that need shared block storage.

If your situation is large enough that none of this is a one-engineer problem — and especially if you have an NSX-heavy deployment or strict compliance certifications by name — get an architectural opinion in writing before you talk to a vendor.

Where to read more

We wrote about Proxmox vs VMware specifically for Atlantic Canada SMBs — the same decision framework, scoped to local-business operations realities.

For an architectural opinion on your specific fleet, write us a brief at setkernel.com/contact. Two paragraphs is enough: your current fleet size, your renewal date, your team’s Linux capacity, the workloads you cannot move. We respond in writing within one business day.