Convergence or divergence of future data systems?

The rough thought is that there are plenty of different data systems out there, even just in the analytics space (not taking into account transactions):

- Data warehouses (Teradata, ...)

- Big data platforms (Spark, Hadoop, ...)

- AI systems (TensorFlow, MXNet, ...)

- NoSQL systems (HBase, Kudu, ...)

Will they converge? Will they diverge further? Are newer technologies tackling specific use cases going to take over the use cases in the more traditional domains?