WDC MAMR Vs. Seagate HAMR Smackdown and Notes From Japan – Trendfocus

Trendfocus Executive Brief
The MAMR versus HAMR Smackdown and Notes from Japan
 

Trendfocus, Inc. was in Japan this past week presenting at the IDEMA Diskcon Japan 2018 conference while also checking the supply chain and customer sentiment on the latest in HDD and SSD demand.

At Diskcon, both WDC and Seagate presented contrasting, yet compelling views on why the two companies chose different paths for energy-assisted recording technologies to boost primarily nearline HDD capacities beyond the next couple of product generations.

Martin Fink, CTO of WDC, spoke first, reviewing the technology investments and developments that led the company to follow the path of microwave-assisted magnetic recording, or MAMR. He reiterated many of the points presented last year when the company went public with its MAMR plans, including its belief that it could eventually hit an areal density of 4Tb per square inch (comparable to HAMR) and support a 15% areal density CAGR to achieve a 40TB HDD by 2025. The idea that WDC could combine its investments in other technologies and fabricate the types of head structures required for MAMR without significant changes in materials or processing equipment seemed convincing and, if successful, could result in a time-to-market advantage over its rivals. However, whether MAMR can achieve the same ultimate areal density as HAMR is unclear and may be limited by the physics of material properties.

Tim Rausch, technologist at Seagate, also presented a number of achievements the company has made in producing working HAMR drives, including most of the building blocks required to make heat-assisted magnetic recording a manufacturable possibility. Key to HAMR is the near field transducer, or NFT, that converts laser light into electron current oscillating at optical frequencies. In addition, Seagate has developed extremely precise active alignment processes to attach the necessary laser to the optical components fabricated on HAMR heads, among other high-precision processes. Rausch countered many of the criticisms of the state of HAMR technology pointing out that Seagate has already demonstrated an areal density of 2Tb/square inch and has already produced media that can support 3Tb/square inch. HAMR heads have also demonstrated continuous writing hours that should easily exceed current nearline HDD write capabilities of 550TB over a five-year lifetime.

The divergence of technology paths for both HDD companies is certainly interesting and will be a source of many debates over the next few years as to which one will produce the better long-term outcome. The nearline drive capacity roadmap also includes options to further increase HDD component counts, up to 10 disks per drive. Boosting component counts will also lift total HDD capacities and may become a useful lever if advanced recording technologies take longer to realize than currently expected.

On a more near-term note, CQ2 ’18 is rapidly coming to a close and early indications support continued strong demand for nearline HDDs by hyperscale companies. 10TB demand is beginning to give way to the 12TB transition, with the latest 12TB volumes likely to more than double over the prior quarter. In addition, seasonal swings in other categories, such as a strong uptick in CE HDDs for game consoles appear on track to meet prior expectations. HDD build plans for the remainder of the year also appear to support current Trendfocus shipment forecasts of more than 380 million for 2018.

Moving to solid-state storage, some customers sense that SSDs are in an oversupply situation currently, especially on the client side of the market. However, with DRAM supply still relatively tight, those companies that maintain both DRAM and NAND with SSDs continue to push bundle deals that leave DRAM-less companies somewhat at a disadvantage. At the same time, the focus for this year is a transition from SATA-based client SSDs to PCIe models as pricing for both interfaces have fallen sharply in the past two quarters with the gap in pricing between interfaces having nearly disappeared. Emerging client PCIe SSD suppliers will fight hard to gain share against companies bundling DRAM and SSDs, assuming qualifications are achieved. The battle for PCIe SSD share may further accelerate SSD pricing declines later this year, which, in turn, may be an ominous sign for client HDD demand as the year progresses.

Speaking of bundles, Intel continues to push Optane to PC OEMs, promising marketing dollars and other incentives to grow the penetration of the new non-volatile storage technology. While Optane caching devices by themselves have not declined in pricing much, Intel has introduced a new Core + bundle that combines its latest 8th generation Core CPUs with an Optane cache drive. The combination promises to reduce the cost of the Optane cache itself by nearly 30%. The outlook for Optane in client PCs remains modest based on PC OEM feedback though, with most adoption occurring in desktop PCs. That said: Optane usage in PCs will likely be combined with SSDs in addition to the original Optane/HDD configurations that have already reached the market in low volumes.

The impact of technology transitions on the near and long-term future of storage appear as a common theme in both the HDD and SSD industries, creating amplified levels of uncertainty over likely outcomes. Uncertainty also tends to lead to overly simplistic proclamations of the impending death of legacy technologies, but industry watchers and those within the storage device sphere may want to take a rational view of the plethora of long-term opportunities still available to both legacy spinning disk and solid-state storage.