It’s an exciting time to be in the memory business. Every decade or so, a set of technologies mature at the same time, leading to a sum of technologies much greater than the sum of the parts. Last decade, the emergence of mobile connectivity, low-power processors, MEMS technology, and NAND Flash memory led to the creation of the smartphone and tablet revolution. Now the continued cost reduction of Flash memory along with the emergence of mega-datacenters needed to service all these connected mobile devices is leading to another technology paradigm shift. Companies such as Google, Amazon, and Facebook are pushing the industry to rethink datacenter designs to effectively deal with the unprecedented amount of data that needs to be stored and analyzed.
I recently attended the Flash Memory Summit and noticed there are numerous companies riding this wave and creating technologies that will enable the next generation of datacenters. Notably, Flash memory prices have dropped to the point where it is now cost effective to replace large amounts of DRAM and HDD (hard disk drives) and many companies are developing Flash Memory appliances to handle the ever-growing demand. However, not all is sun and roses in the memory world. Both DRAM and Flash memory cell technology are reaching scaling limits that could slow the relentless price decline that has driven innovation over the last few decades.
The decline of certain Flash technologies (i.e. Floating Gate NAND or FG-NAND) has been predicted for years only to have clever engineers continually find ways of extending the technology. However, FG-NAND may finally be dead but only because an alternative road has been found, such as the shift to 3D NAND technology. Rather than trying to continue scaling to smaller and smaller dimensions, the new 3D NAND technology reverse scales and moves back a few generations in lithography. Density and cost reduction is achieved by building multiple layers of memory in the back-end wiring levels or BEOL. Given the importance of this shift, we expect most Flash memory manufacturers to follow suit over the next few years. And there are certainly developments under way looking at technologies that could supplant DRAM or create a new storage layer in between DRAM or Flash. Such technologies are sure to allow datacenters continue growing their memory capacity even as DRAM scaling slows.
What’s driving this shift continues to be the pressure that the dramatic growth of digital data has put on existing computer architectures, creating opportunities for innovation in storage systems and memory hierarchies. Here at Rambus, we’ve been following these trends and opportunities with considerable interest for years. Following our acquisition of Unity Semiconductor, we have continued to invest both resources and manpower into new solutions with the goal of expanding our reach and playing an increasingly important role in the evolution of memory hierarchies and storage systems. We’re looking forward to what the future holds, and we’re excited to help our customers navigate the waters of this new world order for memory.
Leave a Reply