SaaS is not a market. It's just software.

"SaaS market will collapse in two years" reads one headline. "SaaS Market in Asia to Hit $1.16 Billion" reads another. "The State of the SaaS market as reflected by SaaScon" chimes in a blogger.

These are just from the first page of a Google search of "SaaS market" (in quotes, so the results only included items with the terms next to each other) that delivered 109,000 links. But, if you believe  Trisha Gross, CEO of Hubspan Inc. of Seattle, that 109,000 links are to something that doesn't really exist.

"SaaS is not a market," she argues. "There's a market for CRM, ERP, supply chain management, integration software and other markets, but not SaaS."

Hubspan offers software integration in a SaaS model that competes with on-premises software from the likes of Tibco and WebMethods.

Software as a service, Gross contends, is just a way to consume software in one of many true markets. Analysts, journalists, bloggers, and especially Salesforce.com with its 1-800-NO-Software and "End of Software" marketing mantra, have got it all wrong.

For one thing, Gross acknowledges, users can embrace SaaS one day and dump it later. But while they may drop the service, it's unlikely they'll jettison the software.

Gross says, "There may be a competitive advantage to moving the software behind the firewall."

Andrew Berry agrees. He's vice president of sales and marketing for Deposco Inc., a supply chain SaaS vendor in Alpharetta, Ga. According to Berry there are a number of reasons to move from the SaaS model to on-premises.

"There are business reasons you might want to move to on-premises," he says. He points to mergers as one incentive to change, especially if the acquiring company standardizes on an on-premises packaged application.

Like Gross, he says a company may change its model for consuming the software, but not the need for the software itself.

By this definition, then, SaaS is just a channel. If so, it's a unique one. Traditional channels, such as value-added resellers, took existing software packages, tweeked them a bit for a vertical market, and resold them. With SaaS, multi-tenant applications are unique to the vendor and designed specifically for the Web. Not quite the channel as we've known it.

Still, the argument is compelling and refreshing to hear. And it has implications for users, vendors and investors. That is, instead of SaaS vendors being measured against themselves, they should be compared in the same markets as their packaged app competitors, which would greatly diminish their relative size and importance. I'm not sure a lot of SaaS players and their backers would want to go down that road.