The answer, as always, is "it depends". Certainly many companies desire to pay for commercial support for many OSS projects, but by the same token, it's obvious that there are some projects that nobody is paying for support for.
In my experience, when companies pay for commercial support, it's ultimately about paying for the comfort factor of knowing that somebody is "on the hook" to fix things if the shit hit the fan. There's a certain level of "CYA" in there as well. If you approve a project using an unsupported OSS project, and it falls over, and the company loses money, then you may suffer some serious repurcusions. OTOH, if their is a vendor behind the product, and they are called in the fix the problem -- and even if the company still loses money -- the vendor absorbs the blame.
Like others have said though, it depends on how complex the software is, what it does, whether or not the firm has deep knowledge of the product already on staff, etc.
Beyond that, a few more thoughts: subscriptions for OSS software, ala the way Red Hat work, are desirable in many ways. For example, with a subscription, you know you always get the latest version as long as your subscription is up to date, so no "big bucks" version upgrades are required. You're also paying with future, inflated, dollars as opposed to todays dollars which are more valuable (assuming the economy doesn't go delfationary). It also aligns incentives between the vendor and the customer, as the vendor has to continue to provide a solid, functional, operational product, for the company to keep renewing their subscription.
Of course all that's true even if the software is proprietary, but the subscription model seems really popular for OSS stuff.