KVM would be less profitable, in the sense that less people buy it. Not enough is sold for the price compared to OpenVZ. From the numbers I keep hearing, KVM is a minor amount of sales and it's not THAT much more expensive.
It would seem this is likely because the people newer to virtualization have no idea how to install their OS or do not want to bother with it. They want a point and click 30 second install like OpenVZ provides. I think it is not until people become more technical that they start to work with KVM. That being said, just based on the amount of offers that are around, I would say OpenVZ is the quickest platform to make a buck from as it takes less overhead as far as time to install and manage as almost anyone can get it setup out of the box using SolusVM. KVM can still be setup similarly but is a completely different beast when it comes to watching how resources are used and maintaining a server compared to the jailed environment that OpenVZ provides.
Also, you can oversell OpenVZ a good bit and it still be stable as it provides a platform for stretching and overselling your resources. While you can oversell with KVM/XEN you can't do it to the same extent, so there is less money in it from that prospective. So the bottom line is that you can likely make more profit with OpenVZ.
I stopped providing KVM VPS completely over time and only retained existing customer. Main reason, dealing with CPU/IO abuse and customers who refuse to admit they are using sufficient resources to warrant a dedicated server. I used to charge more for KVM, and still profit wise it's nowhere near OVZ nodes but yet I have to deal with more KVM customer issues in a single node vs OVZ which has at least twice the number of VPS.
OVZ CPU abuse I can just save a copy of the CPU graph for the VPS itself or point the customer to it, and pretty much no argument, graphs and numbers don't lie. KVM I literally have to go back-forward with the customer multiple times because they'll never admit to CPU abuse or simply don't know how to fix it and I have no easy graph to slam on the table and go "look here". IO usage on KVM is worse too, customers don't know/want to use virtio and result in massive IO abuse especially for Windows, I'm guessing swap usage.
Might be just unique to me since majority of my customers are from the Asia region, my US/EU KVM customers don't have as much abuse issues or are at least able to identify and rectify quickly.