QlikTech has announced that it is acquiring expressor. In other words, a BI vendor is acquiring an ETL tool. However, this doesn't appear to be a play equivalent to SAS acquiring DataFlux or Business Objects buying FirstLogic. In other words, this is not about becoming a player in the data integration market. So, at first sight, this seems a rather surprising acquisition. It's not even as if expressor provided federation capability - it's an out-and-out ETL vendor - which might be thought more obviously complementary to a BI tool.
So, what is QlikTech going to do with expressor? It has stated that it will use expressor for loading (big) data into and from QlikView environments but, more specifically, it sees expressor as offering metadata management and data governance capabilities that will enhance QlikView environments. In particular, expressor will provide functionality such as data lineage and where-used analysis. The press release also states that it will enable "the capture and management of the semantics of data while building QlikView applications to efficiently reuse data definitions".
So I think there are two things that QlikTech wants: the performance of expressor - handy, but you wouldn't buy it just for that - and its semantic capabilities. Since it is its semantics that makes expressor different from the horde of other data integration vendors I must conclude that it is this that QlikTech really wants. What will be interesting over the coming months and years will be to see how the company embeds this semantic technology into QlikView. While I don't think expressor's semantics were ever specifically about the semantic web one can see that it would be a good starting point if you want to start running queries against graph databases (triple stores) for example.
One other conclusion must be that expressor will no longer be marketed as a pure data integration tool. Now, QlikTech hasn't said this but, given what it has it said, I think it is obvious that this is the case: the product may continue to be nominally directed at generic data integration environments but I can't see that lasting.