SlapOS is compatible with most if not all IaaS providers.
SlapOS can be hosted on any bare metal server or on any virtualized server. Success stories of SlapOS show two major use cases: bare metal and VMware. Bare metal is used by 90% SlapOS users since it is the most cost efficient and the only way to compete in price with companies such as Google on collaboration software. VMWare is used by some large corporations which already purchased a virtualization infrasture based on VMWare and on Storage Area Network (SAN). The addition of VMware increases costs by a magnitude of two to four with few if no benefits: the theoretical reliability of virtualized server and storage is often killed by human errors which put down the whole IaaS infrastructure at once for a couple of hours. We have observed until now that many private IaaS infrastructures are less reliable overall than a collection of bare metal dedicated servers running at home.
SlapOS can host various IaaS infrastructures. It is possible for example to allocate qemu-kvm virtual machines directly on SlapOS bare metal servers. This possibility is used by Nexedi developers to create servers and package ERP5 for different GNU/Linux distributions. It is also possible to allocate virtual machines on existing public or private clouds thanks to libcloud library. SlapOS plays in this case the role of a broker between a unified resource allocation API and a collection of IaaS providers, each with its own API.
It would be possible also to use SlapOS as a way to automate the deployment of open source Cloud stacks such as NiftyName, OpenNebula, Eucalyptus or OpenStack. This has been already implemented for some storage backends used by IaaS.