It really depends on what you're trying to achieve.
This is what the Semantic Mediawiki project has to say on the relationship between Semantic MediaWiki and Wikidata:
The software that powers Wikidata is a set of MediaWiki extensions
collectively known as Wikibase, and though Wikibase has similarities
to Semantic MediaWiki, it is a distinct set of software. However, some
of SMW's backend code has been spun off into a separate library,
called "DataValues", that is used by both SMW and Wikibase as a
framework for storing data.
There is the potential that Wikibase and Semantic MediaWiki will
compete against one another as software, with some wikis choosing to
use Wikibase instead of SMW as their data storage system. This seems
doubtful, however: the Wikibase user interface is geared for a highly
multilingual, highly general knowledge base like Wikipedia. Wikis with
a specific focus and only one or a handful of languages would be
better off with the greater structure and simplicity of Semantic
They also have a dedicated page explaining some of the differences between SMW and Wikidata:
The main use case of Wikidata (a centralised, multi-lingual site that
serves as a data repository) is different from that of SMW (a
data-enhanced MediaWiki), and this leads to a number of differences.
Central to a wikidata statement is that its factual claim is supported
by reference(s) (source of the claim). For example, when SMW makes a
claim about the population of Berlin it would only be annotated with
Berlin has a population of 3,5 Mio where Wikidata would make an
extended statement describing it as Berlin's population being 3,5 Mio
as of 2011 according to the German statistical office.