Much hoopla was recently published about Gordon Moore’s “law” turning 50 years old. An overly simplistic summary basically states that Moore’s Law leads to a doubling of transistors on a microchip every 18 months (lengthened from an original estimate of 12 months). The semiconductor realm has held rather firm to this law over the past 50 years, leading to a plethora of innovations, including the mobile devices we use today and the wearables we may end up fashioning tomorrow.
The better articles about Moore’s Law half-century anniversary point to the fact that Moore’s Law is not a physics-defined law but rather a guiding principle that has driven innovation to improve transistor density at a torrid pace witnessed for the past five decades. However, as physical limits are reached, questions remain as to the longevity of this “law”, especially as lithographic process nodes approach that of a number of a very few atoms in size.
The HDD industry has, up until around 7-8 years ago, followed a similar path, with areal densities increasing at a pace of 50 to 100 percent every 12-18 months. However, since then, perpendicular magnetic recording (PMR) has indeed slowed as the physics of layer structure, material and process technologies have hindered rapid areal density growth.
In the HDD world, several technologies have emerged in an effort to continue the push in areal density, from the performance-limiting shingled magnetic recording (SMR) to the expensive and unproven heat-assisted magnetic record (HAMR). Unlike past technology transitions, the new technologies levy some penalty, either performance or long-term cost premium, which is limiting the adoption to only a subset of the HDD industry’s market segments. The challenge for the industry is to choose the appropriate application(s) that require higher storage density and can afford to pay a premium for the technology. And, will these niche targets grow large enough to offset the investment required to bring these technologies and products to market?
Circling back to semiconductors, the conundrum facing the HDD industry may repeat itself in the solid state world as physical limits potentially restrain Moore’s Law rate of growth. Will new technologies to improve transistor density (or any other semiconductor device density) cost too much for every application to leverage? Or is there another innovation that will increase device density while offsetting the the cost of the transition? The emergence of new applications such as ever smaller and more powerful mobile devices is critically reliant on an engineering commitment to extend Moore’s Law further.
What distinguishes the semiconductor industry from the HDD industry is the difference in the pace of technology development investment. The HDD industry is dealing with a shrinking TAM for client and consumer devices, offsetting much of the growth of solutions for enterprise and hyperscale systems. As a result, development budgets have been cut, workforce reduced and spend re-targeted to fewer, less risky projects. On the other hand, active investment in semiconductor research continues to push new designs and to drive toward ever smaller process nodes. The advantage for the semiconductor industry is that developments can be widely leveraged, from CPUs, to memory, to communications, and to storage and beyond. With applications for new chips expanding, investment in development and manufacturing will continue to grow, and new innovations to extend Moore’s Law well into the future seem more of a sure bet than areal density growth for HDDs…