I work for one of the leading semtech companies. It's all about data interoperability and it is inevitably coming. I am not convinced it will come as the result of a company advancing a standard like RDF, rather than as a grass root entity that grows up out of the many existing data formats. The bottom line is that data interoperability is good and it's really not that hard to accomplish. RDF puts forward a scheme to accomplish this with a distributed, directed graph structure that incorporates universally recognizable IDs. I cringe whenever I hear someone talk about how it makes the machine more intelligent or how easy it is to do XYZ using semtech... It's about making extremely precise descriptions of things and gives certain tools (RDF, OWL and inference engines) to make working on abstract models easier, but there is a hell of a lot of work involved in getting something really useful going. As an aside, bing used to be powerset, one of the leading linguistic semantics companies that I saw present in ISWC 07.