Right now it would appear that VMware has a better cloud story around desktop virtualization than Citrix. This is thanks to VMware's purchase of Desktone which has given them a "real" desktop virtualization platform that was written from Day One for desktop cloud providers. It has the multitenancy, automation, isolation, security, billing, and serviceability built in. Awesome.
Compare that to Citrix, who has XenDesktop and XenApp. Both of these products were designed and built in the days when everyone ran their own servers in their own datacenters—a time when the only group of users they had to worry about was their own. These products are not "cloud ready" at all. Citrix's temporary solution has been to release the App Orchestration kit, which is essentially a set of scripts that build XenDesktop and XenApp environments for a cloud provider's customer. (With Citrix, cloud providers end up creating full XenDesktop/XenApp stacks that are completely isolated from each other.)
Before I get too far, allow me to preempt the inevitable comments on this. Citrix is doing a great job with this so far. (Over 2600 Citrix Service Providers and 300k customers, mostly using XenApp and XenDesktop. So obviously it's working.) But also let's call a spade a spade. Those 2600+ CSPs are collectively running 300k separate on-premises architectures with millions of service VMs on their hardware. Citrix claims this is multi-tenant and that those who say otherwise are just throwing mud. But that's just a lot of marketing trying to obscure the fact that the latest FMA-based versions of XenApp and XenDesktop have been architected for single tenant, traditional on-premises environments.
So if you were Citrix with your on-premises-oriented products today, what would you do? Do you try to rewrite your stuff from scratch for the cloud? No, instead you do what you've always done. You wait for Microsoft to make changes to their platform, and then you ride them into the sunset. It's worked before, and it will work again.
Have you heard Microsoft talk about the future of Windows Server and how it fits into the "datacenter without boundaries?" It's awesome. The gist is that the next version of Windows Server will be totally integrated with Azure-based cloud services. It won't be so much as "Windows Server" as it is "Azure Light in a box." You'll be able to add roles in Azure (or Microsoft-based clouds from other providers) as simply as you can add them to a server now.
This is Microsoft's answer to the hybrid cloud. They want to make it where it really doesn't whether the hardware is yours, Microsoft's, or another provider's.
Now tie this back to Citrix. If Citrix plays their cards right, they don't have to massively overhaul XenDesktop and XenApp to be multi-tenant or to compete with VMWare Horizon DaaS (Desktone). Instead they can evolve their products just like Microsoft, slowly breaking down the boundaries of the datacenter.
It will take time for Microsoft to get there. We're starting to see it in Server 2012 R2 and will really see it blossom in Server 2015 (or whatever they call it next). In the meantime, Citrix can just focus on building scripts for service providers for their products now, and then they can hit the ground with a cloud-ready (internal or external) product in another year or so that leverages everything Microsoft is doing. The relationship is good and there's no reason Citrix should build the cloud platform themselves.