Plenty has been written about whether to leave VMware after Broadcom's pricing changes. Far less has been written about the hardware question that follows the decision: the server you specified for vSphere is not automatically the right server for Proxmox VE or Hyper-V. The hardware compatibility lists differ, the storage models are fundamentally different, and passthrough behaviour varies. This is the hardware-selection lens for teams moving off VMware.
The hardware question is separate from the strategy
Choosing your next hypervisor is one decision; choosing the hardware to run it on is another, and the two are easy to conflate. A host that was a certified VMware ReadyNode is not guaranteed to be a clean fit for Proxmox VE or Hyper-V. Each platform has its own compatibility expectations, its own storage architecture, and its own quirks around device passthrough - and those drive different choices in controllers, drives and network cards.
If you are still weighing the move itself, that strategy belongs in our server configuration conversation. This piece assumes the decision is made and answers the next question: what do you actually buy?
Compatibility lists: HCL, certification, and reality
Hyper-V inherits Windows Server hardware certification - if a server is on the Windows Server Catalog and your OEM ships validated drivers, you are on well-trodden ground, and Azure Stack HCI / ReadyNode programmes from Dell, HPE and Lenovo make this turnkey. Proxmox VE, built on Debian and KVM, runs on essentially any modern x86-64 server, but it has no formal vendor HCL - so you lean on mainstream hardware with good Linux driver support rather than a certification badge.
vSphere, by contrast, is strict: only hardware on VMware's HCL is supported, which is exactly the rigidity many teams are leaving behind. The practical lesson is to favour mainstream Dell PowerEdge, HPE ProLiant or Lenovo ThinkSystem platforms with broad OS support, and to verify NIC and storage-controller compatibility specifically for your chosen platform before you buy.
Storage is where the platforms truly diverge
This is the biggest hardware consequence of leaving VMware. vSAN gives way to platform-native storage: Proxmox VE uses ZFS for single-node and small deployments and Ceph for scale-out hyperconverged clusters, while Hyper-V uses Storage Spaces Direct (S2D). These are software-defined storage stacks, and software-defined storage wants direct, unobstructed access to the raw disks - which changes how you spec the controller.
ZFS and Ceph both expect to see the physical drives directly, so you want a plain host bus adapter in IT/passthrough mode, not a hardware RAID controller hiding the disks behind a logical volume. A RAID card in front of ZFS or Ceph fights the filesystem's own redundancy and data-integrity features. Choose the controller deliberately with our host bus adapters guidance - this single decision causes more post-migration grief than any other.
- •Proxmox VE single/small: ZFS on an HBA in IT mode (no hardware RAID in front)
- •Proxmox VE scale-out: Ceph nodes - more drives, fast cluster network, HBA passthrough
- •Hyper-V cluster: Storage Spaces Direct - validated S2D/Azure Stack HCI nodes
- •Across all: plan a dedicated, fast storage replication network, separate from VM traffic
Passthrough, Ceph nodes and networking
If you pass GPUs, NICs or storage controllers through to VMs, validate IOMMU (Intel VT-d / AMD-Vi) and clean PCIe passthrough on the exact platform and firmware you intend to run - both Proxmox VE and Hyper-V support it, but behaviour varies by board and BIOS version and is worth proving in a pilot.
Scale-out designs change the node shape. A Ceph cluster wants several similar nodes with plenty of drives and a fast, dedicated cluster network (25GbE or better) because Ceph constantly replicates across nodes; S2D has the same appetite for east-west bandwidth and RDMA-capable NICs. Build a host or a matched set of cluster nodes for either platform in our configurator.
Buying the right host
In short: keep the mainstream OEM platform, but re-make three decisions for your new hypervisor - verify compatibility for the specific OS, swap hardware RAID for an HBA in passthrough where ZFS, Ceph or S2D will own the disks, and size a dedicated storage network for scale-out clusters. Pick the controller with our HBA/RAID guidance, sanity-check the wider design with our server configuration service, and build the host in our configurator.