I have actively worked on Intel HAXM as part of emulator project [1]. I think there's a fragmentation problem in hypervisor scene: KVM is Linux-only, Hyper-V is Windows-only, HVF is macOS-only. Some toy hypervisors support a small subset of modern features, while lacking in other areas. Some support either Intel VMX or AMD SVM, but not both. To make things worse, Windows/macOS are making incredibly difficult for developers to deploy their own hypervisors. I believe Intel HAXM took a step in the right direction at aiming to be cross-platform, the community has unofficially even ported it to AMD CPUs as well, but more awareness is needed
I don't think https://libvirt.org/ is a solution to the fragmentation problem. Whenever a new virtualization feature appears (and this happens quite often!), having to update N different hypervisors, and expecting their interfaces to consistent enough and consumed by libvirt is wishful thinking: In practice it just does not happen. This hurts the development of cross-platform software that require cutting-edge hardware virtualization features.
I don't think https://libvirt.org/ is a solution to the fragmentation problem. Whenever a new virtualization feature appears (and this happens quite often!), having to update N different hypervisors, and expecting their interfaces to consistent enough and consumed by libvirt is wishful thinking: In practice it just does not happen. This hurts the development of cross-platform software that require cutting-edge hardware virtualization features.
[1] https://github.com/AlexAltea/orbital