The Broadcom-VMware renewal is now a board-level line item, not an IT footnote. If your quote arrived several multiples higher than last year, you are not being singled out. This playbook lays out where to migrate, how to move virtual machines off ESXi, and the gotchas that turn a clean cutover into a weekend outage.
Why did VMware costs suddenly explode after the Broadcom acquisition?
After Broadcom closed its VMware acquisition in late 2023, it retired perpetual licenses, collapsed the catalog into a few subscription bundles (VMware Cloud Foundation and vSphere Foundation), and raised per-core minimums. Many organizations report renewal quotes 800 to 1500 percent higher than prior spend, driven by forced bundling more than list price alone.
- ▸Perpetual to subscription. Existing perpetual keys still boot, but you lose security patches and support without a new term subscription.
- ▸Forced bundling. Buying plain vSphere became difficult; VCF bundles vSAN, NSX, and Aria you may never deploy.
- ▸Per-core minimums. Licensing shifted to per-core with minimum floors (widely reported 16-core-per-CPU minimums, later larger per-order minimums), which penalize small and mid-size hosts.
- ▸Partner churn. Many long-time VMware resellers lost program status, pushing renewals direct and removing familiar discounting.
Treat exact percentages as reported by affected customers, not a fixed schedule; your number depends on core counts and current entitlements.
What are my realistic alternatives to VMware in 2026?
The three credible destinations are Proxmox VE (open-source KVM and LXC virtualization on Debian), OpenStack (a full private-cloud platform), and Nutanix AHV (commercial hyperconverged infrastructure). Proxmox suits SMB-to-mid virtualization, OpenStack suits large self-service clouds, and Nutanix suits teams wanting a supported turnkey appliance.
- ▸Proxmox VE 8.x / 9.x runs KVM plus LXC with integrated Ceph and ZFS, clustering, and built-in Proxmox Backup Server. It is free, with optional paid support subscriptions.
- ▸OpenStack (recent releases such as Caracal, Dalmatian, and Epoxy) gives you Nova, Neutron, Cinder, Keystone, and Glance for a multi-tenant, API-driven cloud. Powerful, but operationally heavy.
- ▸Nutanix AHV / Cloud Platform is commercial KVM-based HCI with Prism management and first-party Move migration tooling.
- ▸Also worth a look: XCP-ng and XenServer, Microsoft Hyper-V with Azure Local, and Harvester (SUSE, Kubernetes-native). If you are already reconsidering owned hardware, pair this decision with Cloud Repatriation: When Bare Metal Wins the Math.
When should I choose Proxmox vs OpenStack vs Nutanix?
Choose Proxmox for tens to low-hundreds of hosts where a lean team wants a like-for-like ESXi replacement. Choose OpenStack when you need multi-tenant self-service, quotas, and an AWS-style API at scale. Choose Nutanix when you want vendor-supported HCI and will pay to reduce operational burden and migration risk.
- ▸Pick Proxmox if you run classic VMs, want fast cutover, value a simple web UI with Ceph or ZFS, and have Linux-comfortable admins. Headcount cost stays low.
- ▸Pick OpenStack if you are a service provider or platform team needing tenants, projects, Terraform-driven provisioning, and horizontal scale into hundreds or thousands of nodes. Budget for dedicated operators or a supported distribution.
- ▸Pick Nutanix if you want the smallest migration risk with commercial support, accept per-node licensing, and treat it as a VMware-cost hedge rather than cost elimination.
For EU teams, moving to owned virtualization also advances data-residency goals discussed in Sovereign Cloud and EU Data Residency in 2026: An Engineering Playbook.
How do I actually migrate VMs off ESXi?
Inventory every VM, then convert its VMDK disks to the target format and inject the right guest drivers. Proxmox VE 8.2 and later ship a native ESXi import wizard; virt-v2v and Nutanix Move automate conversions elsewhere. Test-migrate non-critical VMs first, validate boot and networking, then batch the rest during maintenance windows.
- ▸Discover. Export inventory with RVTools or PowerCLI: vCPU, RAM, disk layout, VMware Tools version, guest OS, and app dependencies.
- ▸Convert. Move VMDK to qcow2 or raw using the Proxmox import wizard, virt-v2v, or qemu-img; Nutanix Move and Cloudbase Coriolis handle bulk conversions.
- ▸Drivers. Install VirtIO drivers and the QEMU guest agent (critical for Windows) before or during cutover, and remove VMware Tools.
- ▸Validate. Confirm boot, IP addressing, application health, a working backup job, and monitoring.
- ▸Cut over. Schedule per batch, keep the ESXi source intact until validated, then decommission.
What breaks during a VMware migration (the gotchas)?
The painful surprises are Windows VMs that fail to boot without VirtIO drivers, networking that assumed a distributed vSwitch or NSX overlay, storage features like thin provisioning and snapshots that behave differently, and backup tooling that must be re-pointed. Plan driver injection, network mapping, and backup integration before you touch production.
- ▸Windows boot failures. Inject VirtIO storage drivers first, or temporarily attach the disk as SATA or IDE to boot, then switch.
- ▸Networking. VLANs on a Linux bridge or Open vSwitch replace vSwitch and NSX; MTU, LACP bonding, and VLAN tagging must match exactly.
- ▸Storage semantics. vSAN storage policies map to Ceph replication and CRUSH rules or to ZFS; thin provisioning, TRIM, and discard behave differently.
- ▸Backups. Veeam 12.3 and later support Proxmox VE; otherwise adopt Proxmox Backup Server, Nakivo, or Bacula and test a real restore.
- ▸Guest licensing. Windows Server per-core datacenter licensing still applies on the new hosts.
- ▸Automation. Replace vRealize/Aria and PowerCLI scripts with Terraform providers, Ansible, or the platform API, applying the same Infrastructure as Code: How Terraform Saves Startups Time and Money discipline you use elsewhere.
How do I replace vSAN, DRS, and NSX without VMware?
Every VMware feature has an open equivalent, but not always an automatic one. Ceph or ZFS replaces vSAN, the Proxmox HA manager or OpenStack scheduling replaces HA and DRS, and Open vSwitch, OVN, or Neutron replaces NSX. Expect to design these deliberately rather than inherit VMware defaults.
- ▸vSAN becomes Ceph (RBD) or ZFS replication. Ceph provides distributed HCI storage; size for replica 3 and dedicated storage-network bandwidth.
- ▸HA becomes ha-manager (Proxmox) or Masakari (OpenStack). These restart VMs automatically on host failure; live migration covers planned moves. Rehearse failover the way you would for PostgreSQL High Availability: The Failover You Rehearse.
- ▸DRS load balancing becomes manual or scripted. Proxmox has no automatic DRS; use scheduled rebalancing or third-party tools. OpenStack leans on the Nova scheduler and Watcher.
- ▸NSX becomes OVS/OVN, Neutron, or Cilium. Microsegmentation moves to OVN security groups or eBPF-based network policy.
What does a phased migration timeline look like?
A safe VMware exit runs in four phases over roughly three to nine months for a mid-size estate: build and validate the target cluster, migrate dev and test, migrate production in batches, then decommission ESXi. Keep both platforms live until each batch is validated so rollback stays trivial.
- ▸Phase 1 (weeks 1-4). Stand up the target platform, storage, networking, backup, and monitoring; migrate one throwaway VM end to end.
- ▸Phase 2 (weeks 4-8). Move dev and test, resolve driver and network issues, and write runbooks.
- ▸Phase 3 (months 2-6). Batch-migrate production in maintenance windows, validate each application, and keep ESXi as fallback.
- ▸Phase 4. Reclaim ESXi hosts, cancel the renewal, and capture the savings. If you now run more services in-house, Self-Hosting Internal Services Without a Second Job keeps the operational load sane.
How TuniCyberLabs helps
We run VMware exits end to end: sizing the Proxmox, OpenStack, or Nutanix target, scripting VMDK conversions and driver injection, re-platforming networking and storage, and rehearsing failover and restore before cutover. Our engineering teams across the EU and North Africa design for data residency and hand you clean infrastructure-as-code, not a black box. Get a fixed-scope migration assessment at /services.