Even supercomputers are a commodity now!
You may not need a supercomputer with 1728 cores. But below article is interesting because it shows how the commoditisation just keeps on going. Supercomputers used to be custom components from top to bottom and thus also very expensive.
The latest generation of Cray supercomputer, Urika GX, uses standard Intel CPUs (lots of them though in the biggest configuration) and runs OpenStack and support the use of Hadoop, Spark etc.
The general trend is of course to move more and more applications to cloud providers. It will be interesting to see where such supercomputers might find their market. But if you do very high-volume continuous proprietary processing of data this might be a worthy alternative to the economics and performance of lower performance general cloud notes form Microsoft Azure or Amazon AWS. The selection process would probably also have to look other optimised hardware appliances including SAP HANA and Oracle Exalytics - but neither focus explicitly on general big data tools without vendor lock-in.
Read more at: http://techcrunch.com/2016/05/24/crays-latest-supercomputer-runs-openstack-and-open-source-big-data-tools/