The economics are fairly simple.
Your support, validation and sustaining costs don't contribute to the bottom line of your business. If you have a part of the product that takes a unnecessarily large proportion of the bottom line, you look at the value proposition. You do something as simple as removing the client for a platform, you save money.
BUT, if the product is based around open standards, the Linux community has a high probability of making something that will work anyway. For FREE. No support costs for a client, no development and validation costs either. Linux, with it's "Freedom" has an extremely high cost to be an ISV on, you have kernels, X versions, distributions.. All subtley different and all having precious consideration for the cost of operating in that ecosystem.
Google has many examples of killing/not creating a client, but fostering the capability. Google talk is a great example. Google still gets the branding value of the service, but doesn't need to have a client, I have *NEVER* heard anyone talk about "Google's JMPP or Jabber Service". I would expect that this is the same, but for google voice. The people carrying credit will probably be handled.