Actually all 3 of those are true.

I don't know exactly what this company's product is, but if they offer an API for creating maps on a website it would be a rival product.

Next, the barrier to entry for online mapping is huge. The raw data to generate and create maps is not easy to collect. Especially for satellite images and topo data. And then there's the infrastructure required to actually process and serve the data.

Switching mapping APIs is not always trivial. Along with the different interfaces, most of them have slightly different features. Once a site is using a specific mapping API it's anywhere from a hassle to a major PITA to switch to something else.

I think you misunderstand the economic meaning of "rival good": http://en.wikipedia.org/wiki/Rivalry_(economics)

True, but API's use hardware that has actual costs so it is a "rival good". For example, GPL code is a non "rival good" in that someone running it on their own hardware costs nothing, a download service that provides GPL code is a "rival good" in that a finite number of people can use it at one time.

Now if Google hosted their map's as a torrent and said, use this for whatever you want. That would be close enough that calling it a non rival good would be a reasonable defense. But, they do charge some customers for using their API which right or wrong opens them up to predatory pricing laws.

Yeah, rival-ness is not a binary value though. In a case where marginal costs of providing a service are very low, it's reasonable to describe it as being mostly non-rival.

> if Google hosted their map's as a torrent . . .

Then it would be non-excludable, which is a different concept from non-rival. It would be rival to the same degree as before, in that it would require marginal computing resources to use. Just the resources required would be coming from the user rather than the provider of the service.