The plain-English definition
A hypervisor, also called a Virtual Machine Monitor (VMM), is software that lets one physical server safely pretend to be many. It carves a host's real CPU cores, memory, storage, and network into isolated virtual machines (VMs), each running its own operating system as if it had hardware to itself. The hypervisor sits underneath, scheduling who gets the real silicon when.
Why it matters: before virtualization, servers ran one application each at 10–15% utilization. Hypervisors pushed that to 60–80%, which is why a 12-host cluster can run 300 workloads. Roughly every enterprise data center, and every public cloud, is built on this layer. When the dominant vendor of that layer changes its pricing model overnight, as Broadcom did with VMware, the entire industry feels it (that story here).
Type 1 vs. Type 2 hypervisors
| Type 1, bare metal | Type 2, hosted | |
|---|---|---|
| Runs on | Directly on server hardware; no host OS | As an app on top of Windows/macOS/Linux |
| Performance | Near-native; built for production | Noticeable overhead; fine for desktops |
| Used for | Data centers, clouds, anything 24/7 | Developer laptops, testing, demos |
| Examples | VMware ESXi, Microsoft Hyper-V, KVM (Proxmox, AHV), Xen | VMware Workstation/Fusion, VirtualBox, Parallels Desktop |
Everything in a VMware migration conversation is about Type 1. When you evaluate ESXi alternatives, you're choosing among other Type 1 platforms, all of which are mature, widely deployed technology, not science projects.
The market landscape after Broadcom
For 20 years ESXi was the default and everything else was niche. Broadcom's 2023 acquisition and the 2–5× price increases that followed (licensing details) redrew the map. Here's the field in mid-2026:
| Platform | Hypervisor core | Licensing model | Best fit |
|---|---|---|---|
| VMware vSphere/ESXi | ESXi (proprietary) | Per-core subscription (VCF/VVF) | Deep VMware tooling dependence; or via managed providers at scale pricing |
| Nutanix AHV | KVM-based | Commercial subscription (capacity-based) | Enterprise HCI shops wanting full vendor support, compare |
| Proxmox VE | KVM + LXC | Open source; paid support subscriptions | Cost-focused teams with Linux skills, compare |
| Microsoft Hyper-V | Hyper-V | Included with Windows Server licensing | Microsoft-centric shops, compare |
| OpenShift Virtualization | KVM via KubeVirt | Red Hat subscription | Teams converging VMs and containers, compare |
| Hyperscaler VMware (AVS/GCVE/VMC) | ESXi, provider-operated | Cloud consumption | Fast data center exits, compare |
| XCP-ng / Xen | Xen | Open source; paid support | Open-source-first environments |
Two things worth noticing. First, KVM won the open-source war, it underpins Proxmox, AHV, OpenShift Virtualization, and most of the public cloud, so betting on a KVM platform is not a fringe bet. Second, the hypervisor itself is rarely the deciding factor anymore: live migration, HA, and clustering are table stakes everywhere. The real differences are management tooling, ecosystem integrations (especially backup), licensing economics, and your team's skills. That's why platform selection is a fit question, not a feature-checklist question, and why a free assessment starts with your environment, not a product.
Capabilities to evaluate in any hypervisor platform
- Live migration, moving running VMs between hosts with zero downtime (vMotion in VMware-speak)
- High availability (HA), automatic VM restart on a surviving host when hardware fails
- Clustering & resource scheduling, pooling hosts and balancing load (DRS in VMware-speak)
- Snapshots and backup APIs, what your backup vendor hooks into; verify support before choosing
- Storage model, hyperconverged (vSAN-like), shared SAN/NAS, or both
- Network virtualization, VLANs, SDN, microsegmentation if you use NSX today
- Management plane & API, the thing your team lives in daily; demo it with your own admins
Glossary: terms you'll hear in every migration conversation
| Term | What it means |
|---|---|
| VM (virtual machine) | A software-defined computer with its own OS, running on a hypervisor |
| Host / node | A physical server running the hypervisor |
| Cluster | A group of hosts managed as one pool of capacity |
| vCPU | A virtual CPU presented to a VM; multiple vCPUs share physical cores |
| vCenter | VMware's central management server for hosts and VMs |
| vMotion / live migration | Moving a running VM between hosts without downtime |
| HA / DRS | High availability (auto-restart on failure) and dynamic load balancing |
| vSAN / HCI | Hyperconverged storage: pooling each host's local disks into shared storage |
| NSX | VMware's network virtualization and microsegmentation layer |
| HCX | VMware's bulk workload-mobility tool, used to migrate to clouds/providers |
| KVM | The open-source hypervisor in the Linux kernel; basis of Proxmox, AHV, most clouds |
| VCF / VVF | Broadcom's VMware subscription bundles (full stack / compute-focused) |
| Per-core licensing | Pricing metered on physical CPU cores, the post-Broadcom standard |
| RTO / RPO | Recovery time / recovery point objectives, how fast you must recover, and how much data you can lose |
| DRaaS | Disaster-recovery-as-a-service: replicating VMs to a provider's cloud for failover |
| P2V / V2V | Converting physical-to-virtual or virtual-to-virtual (e.g., ESXi → KVM) |
Where to go next
If you're here because of a VMware renewal: start with what happened to VMware, see what a move costs in the cost guide or the calculator, and compare every platform side by side in the comparison matrix.