Without advances in how the world physically stores and retrieves data, today’s most useful devices and algorithms would not exist. The dominant memory chip technologies such as NAND Flash and DRAM rode the wave of innovations in materials, process and device technologies to scale down the path of Moore’s law. Although physical scaling is becoming increasingly difficult, the forces and market pull driving cost, power and density scaling are growing relentlessly. The amount of memory in systems for example is increasing geometrically while the power budget continues to decrease. In addition, the applications and resultant memory requirements continue to diversify and expand from traditional handheld devices and larger data centers.
A commitment to innovation and creativity at a system level design is required to meet demands of the data age. These innovations will help fuel the next generation of technologies such as self-driving cars, space exploration, artificial intelligence and machine learning, which sounded like science fiction not so long ago. Several technologies have been proposed over the years with no clear winner. Some of the critical factors which need to be considered for a successful implementation of a new technology include; why and when alternate memory technologies may be needed, what are the performance criteria and related requirements, and what needs to happen in the ecosystem to support a successful new technology. The result of this reality is that bottoms up development for a new memory technology may not feasible due to technical risks and cost and we must target application-specific solutions for new future markets.