In what has been a very eventful storage year the most asked question is when will the SSD shortage end? It already has, and it wasn’t an upending event. Instead, it seemed to happen quietly and quite orderly, transitioning the market from a sellers’ market to a buyer’s market. Seagate recognized the changing market conditions and made strategic adjustments and have gotten traction within the growing hyperscaler space with their 10TB drive. While WDC/HGST focus was on the integration of three storage companies (WDC, HGST, and SanDisk) into one allowing Seagate to grab market and mindshare with Seagate reportedly outselling WDC/HGST at a 6 to 1 ratio in the channel.
Q4 HDD numbers are expected to come in right in at 100 million units with Q1 2018 to fall in the 90 million range with some market watchers concerned that sales are vulnerable to seasonality and SSD cannibalization and could call into the 80 million range. Seagate has been aggressive with pricing, setting the stage for additional downward pricing pressure as WDC/HGST and Toshiba respond to Seagate’s resetting of the deck. Improved 3D NAND yields are driving market penetration for SSD, but there is still are questions to be answered around the market demand for MLC based SSD.
Enterprise Storage
Integrating three companies into one and managing a dispute around NAND fab ownership, one can forgive WDC/HGST for reacting to slow to changing market conditions. Seagate was quick to recognize that market dynamics shifted and that it was no longer a seller’s market. When Seagate was aggressively pricing product in Q3 market pundits, including me, focused on the impact pricing would have on Seagate’s P&L, not the strategic corrections they made that proved to be shrewd play calling on their part.
Despite having the upper hand against both Seagate and Toshiba due to lower manufacturing costs and higher manufacturing capacity, WDC/HGST lost ground to competitors. On the hi-cap end of the business, they are selling everything that they can build to hyperscalers. However, because WDC/HGST had not made market adjustments to changing market conditions, their channel sales were impacted negatively. Expect WDC/HGST to respond by shifting focus from profit to top line revenue, and adjust pricing to be more competitive with market conditions. Reportedly they are looking to better serve tier two and three customers by introducing more products targeted to their needs (introduction of 8TB Air) and they are also looking to make it easier to do business with them. They are expected to flatten pricing and reduce the reliance on the complex web of SPAs they have in place.
Throughout 2017 the transition to SSD and the impact that additional NAND capacity will have on the storage market has been one of the most talked about topics. SSD will continue to poach HDD sales, but the HDD manufacturers are not sitting idly watching. The manufacturers are rightsizing their businesses by divesting pieces of business that no longer adequately support today’s storage environment. Instead focusing on where they can leverage the cost advantage of spinning disks by investing in new technologies that playoff off of the decided cost advantage of HHD over SSD. There is a rumor that Seagate is considering exiting the 2.5” enterprise market. SSD offers the better solution in high IOPS environments with lower TCO and better performance. The 15k HDD is all but dead with an insider at WDC/HGST reporting Q3 sales were nonexistent.
Last month, the market debated the merits of new technologies of NAMR and HAMR; both satisfy the need for cheap storage with different approaches. Seagate announced the multi-actuator concept that allows for simultaneous read requests, effectively doubling throughput. Meeting the need for increased IOPS and low cost to stay in front of SSD sales. Expect to see WD and Toshiba follow closely behind with their versions.
10TB drives are the sweet spot accounting for 30% of nearline sales. Seagate has reportedly made inroads with their 10TB (HE or Air) drives with hyperscalers. Although, they are limited in the amount that they can ship with production volumes capped at 450k units per quarter. Toshiba hasn’t experienced the same success with its 10TB, but they are on track for shipping both 12TB and 14TB helium drives that utilize conventional magnetic recording (CMDR) drives, in early 2018. To date, HGST employed shingled magnetic recording (SMR) for its 14TB drives, requiring file systems tailored to work with SMR complexities while Toshiba is using conventional magnetic recording which is much easier to integrate into a storage environment.
With increasing amounts of 3D NAND hitting the markets as 3D FAB conversions mature, it is easy to get caught up in the possibilities new technology can bring. Demand for high-end SSD is strong with HGST reporting to be sold out of both NVMe and SAS SSD. However, HDD is still filling 70% of storage bays with SSD not expected to reach parity until 2021. The SSD market still seems to be finding itself with current SSDs operating in infrastructure designed to support HDD, utilizing protocols that fail to realize the full potential of NAND. Next generation NVMe SSD using 3D NAND are optimized to benefit from NAND’s significant speed increase and adoption in the enterprise market is accelerating, but not everyone is on board. Due to the substantial cell degradation of 3D NAND controller’s algorithms are reset to accommodate the accelerated wear by spreading read writes across more bits. In turn, requiring users to buy additional capacity to achieve the same level of reliability of MLC based SSD. Not all end users are willing to make this change. Those applications that require a minimum amount of storage prefer to stick with older MLC based products that support their installed base and provide better endurance. However, this demand is running into a collision course with the aggressive conversion of FABs to 3D NAND to drive bit growth demand rather than add more wafer starts. This leaves MLC supply short of demand. Also, 3D NAND based SSD will be more plentiful pushing SSD prices down. Here we have two SSD markets operating at deferent ends of the supply and demand equation with 3D based SSD pricing heading lower while MLC based SSD pricing will be increasing.
Although unlikely events that may disrupt SSD price decline are Samsung shifting manufacturing capacity to support DRAM demand, MU Hynix struggling to ramp next-gen fab solutions or an acceleration of high-end phone sales. Overall 2018 appears to be a buyer’s market. As detailed earlier, the 2.5” enterprise market warrants watching with Seagate rumored to contemplating an exit. However, Toshiba is showing continued support releasing their AL 15SE line with its 2.4TB 10.5k mission critical drive. While HGST last introduced a 1.8TB in 2014 indicating they have little commitment to this end of the market. For the large installed based relying on mission-critical drives, things could get quite interesting if Seagate does exit the market. Already we see increased open market demand for 2.5” 10k drives in the open market.
Notebook/Desktop Storage Sales
Reports around the tail end of the quarter suggest that the commercial space showed strength and that retail returned a better than expected result. This is most likely on the back of a strong economy that is driving demand as hiring ramps upwards. We don’t see this as sustainable due to seasonality and doubt that corporate demand will persist at such levels.
A report by Digitime China said that handset manufacturers had cut Q4 orders by 10%. Much of that NAND supply will be diverted to SSD accelerating SSD cannibalization of HDD and driving SSD prices lower.
The lone bright spot in the client space is in gaming. If virtual reality takes hold, expect to see a demand for low-cost storage give life to 2.5” HDD. However, the market is sending mixed signals. Despite the reported strong demand for 2.5” 1TB in the market, Seagate was aggressive lowering 1TB pricing by $1.00 to take advantage of markets conditions to move volume. There is also word of one of the game consul manufacturer canceling a 100k order with Toshiba for 2.5” 1TB drives because they were not able to meet aggressive sales targets.
Final Thoughts
Both HGST and Seagate recognize the shift of storage to the cloud and are looking for ways to increase revenue in the quickly evolving storage market. Both are looking to move up the value chain by offering storage solutions, with both selling storage arrays in various configurations. In the SSD enterprise market, there is not one size that fits all needs so HGST will go back to designing their own controller. This gives them greater flexibility in developing solutions for customers and increased ownership of the solution.
With the market moving so quickly and new technologies such as IoT and autonomous driving competing for storage supply, today’s demand equation is not a foregone conclusion.
I would love to hear your thoughts.
Please comment on LinkedIn page or email me directly at Stephen.buckler@horizontechnology.com
