The growing explosion of digital data motivates renewed emphasis on new architectures for future data-centric workloads. At the same time, the inability of power to scale with increasing transistor counts has led to a recent focus on “dark silicon” designs where transistors are used to design specialized accelerators to improve energy efficiency and performance. Combining these trends, in this paper, we examine the applicability of accelerators in future data-centric system architectures. Specifically, we focus on indexing, a fundamental and time consuming component of databases, and propose a new “indexing widget” to improve energy efficiency and performance. Through preliminary characterization of a representative scale-out database, VoltDB, in conjunction with performance, power and area models, we show that our proposed approach can achieve 1.24X improvement in energy efficiency relative to non-accelerated designs.