I have always been fascinated with the intersection of economics and technology with technology testing economic principles that have been in play for hundreds of years. We have all heard “Creative Destruction” touted as the driver of new ideas and technology and we have talked many times about the transformative impact of SSD on the storage market. I am here to speak up for economic side of the equation. A nice way of saying it comes down to the dollar. A good example of the strong influence cost has on technology is a combination of events that occurred around the tech crash of 2008 that allowed the power of computation to move from the office back closet to the home. After the crash failed broadband companies were selling fiber pipeline for pennies on the dollar making it possible for the masses to affordably access data. Low cost data combined with collaborative software forever changed the world, forming the bedrock of today’s digital ecosphere.
The Hard Drive (HDD) will continue to stake its claim as the low cost workhorse of the digital economy by offering gigabytes for pennies on the dollar compared to SSD. (Huh, awkward. Editor, can you rephrase). As we move into 2018 expect to see enterprise HDD combined with SSD in tiered systems that offer a high performance low cost storage solution.
Waiting in the wings is NVMe, the next iteration of SSD storage. Not only is this important in pushing the performance envelope but crosses a new psychological barrier as the market shifts away from SATA a protocol written for spinning media to one written to support SSD based storage. The NVMe ecosphere is growing as more and more hardware manufacturers introduce NVMe based storage solutions. Once NAND fabs fully transition to 3D expect the market to swing further towards NVMe as NAND prices retreat. Although not a complete death knell for the hard drive the manufacturers need to make sure that they are in the conversation around tiered storage.
Solid State Drives
The divide of the SSD market between client and enterprise is becoming more and more pronounced. Creating two users groups with distinct needs and support structure. Demand for client based SSD is price driven allowing for easy access to customers by third party SSD builders while the enterprise market focuses on performance and quality, preferring to buy SSD directly from the NAND manufacturers themselves. Giving enterprise hardware suppliers access to the latest technology to maintain an advantage in a competitive and evolving storage market.
The sale of Toshiba’s fab will impact the shaping of the SSD market. The direction of the sale continues to be unclear with WD continuing to exert its right to dictate terms of the sale. Upfront the advantageous for SSD manufacturers owning fab capacity are direct access to supply, access to new technology, and a dedicated customer base. The liability is the heavy capital investment and possibility of owning an underperforming asset in a market where supply exceeds demand.
Looking long term one not only needs to think about who ends up with Toshiba’s capacity but also the potential impact of Chinese investment in fab capacity which, depending on what you read, can push the supply demand equation into oversupply. Others say that currently there isn’t enough storage manufacturing capacity to absorb today’s data needs and additional capacity will easily be consumed as data demand grows. The impact on the commoditized client SSD is clear but it is not clear that they will be able to easily crack the enterprise market. Enterprise builders focus on quality and performance, favoring SSD manufacturers that are on the front end of new technology developments and an established history.
Recent reports by both Samsung and Seagate actively pursuing fab capacity indicates that they agree with the view that demand will drive the need for additional capacity. Seagate who does not have an investmant in fab capacity is reportedly teaming up with SK Hynix on Fab joint venture indicating that the advantages of being vertically integrated outweigh the advantages on relying on an out source model to secure supply. Samsung is giving bullish signals with a report that it is in discussion with the Chinese government around the construction of a new fab which would imply continued tightness within the market. Which is very interesting in light of the Chinese having been very vocal about their entering the market and pundits indicating that their entry into the market will push the market into an oversupply situation.
If the Toshiba fab capacity is ultimately picked up by WD we will see a more stable transition and benign supply demand conditions. If sold off to a third party, creating another vertically integrated manufacturers, expect a bit of volatility as the third party entity finds its way in the market.
We at Horizon have seen the client SSD market soften with open market prices and demand falling. Slack demand is more likely attributed to additional NAND availability due to weak Q2 smart phone sales that had been absorbing much of NAND capacity and dram shortages impacting the desktop/notebook market. The Q3 back to school build cycle will give us a good idea of the severity of the SSD shortage.
While the enterprise SSD market is looking past the shortage to the future of storage with the market poised to adopt the speedier NVMe interface. Many hardware ware manufacturers have been launching NVMe based storage solutions growing the NVME ecosphere. NVMe utilizes the PCIe interface which currently commands a 20% to 35% premium over SATA based product. Intel has been aggressively pushing PCIe product based on its 3D NAND and has sent out last time buy notifications for many of its 2D based flashed SSD. SSD manufacturers are pushing to move customers to new gen products that generate higher yields, accelerate return on investment and establish first mover position through early adoption of new products. The missing piece to moving customers to NVMe based solutions is NAND supply. Due to the shortage pricing deltas between SATA and PCIe increasing throughout 2017 the transition to NVMe has slowed.
The move to 3D based NAND is a costly process and the opportunity for missteps are many. Manufacturers are ramping up 3D lines and are beginning shipments of their 3D based NAND storage. Manufacturers are reluctant to provide yields making it difficult to pinpoint when 3D NAND will alleviate the shortage. We expect NAND supply will be tight throughout 2017 and based on Samsung seeking to expand capacity we don’t expect to see a crash in prices but a deceleration and slow retraction in SSD pricing.
Enterprise HDD
Hard Drives manufacturers continues to leverage the SSD shortage to carve out a value play in the storage market. HDD revenues remain stable while shipped capacities increase against 30 to 40% Exabyte growth. Enterprise demand is strong and will carry into the second half of the year and most likely we will run into a tightness within the market driven by drive component shortages, the continued SSD shortage and seasonality. With the SSD shortage extending HDDs runway well into 2018. Expect the 15k HDD to stay in the server mix for at least the next 12 to 18 months. All in all a good situation for the HDD manufacturers with HDD sales expected to eclipse 110 million units per quarter on the backside of 2017.
Last month we talked about demand poised to pick up over Q3 as data centers refresh storage to 10 TB capacities and hyperscale customers switch to the higher capacity drive. WD/HGST are well positioned take advantage of the shift to 10TB capacities as WD id looking to ramp from million 1 million units over the quarter to 3 million units over Q3. The benefits of first mover advantage are considerable. Translating to improved margins, accelerated return on capital invested and the coveted incumbent position with customers. Seagate’s situation is unclear with mixed feedback around their progress with their 10TB drive. Toshibas who has falling behind in the capacity wars is not as focused on the enterprise market but is delivering strong storage results against a healthy market.
Product is showing some signs of tightening with end users willing to secure 4 and 6 TB capacities at premiums to back fill supply gaps. We are also seeing tire kicking around 8TB capacities but no POs placed.
Desktop/Notebook
There are no illusions that personal computer is going to come back but it is reasonable to expect the market to stabilize. All signs are that we have reached that point with IDC reporting that Q1 shipments grew .6% over 2016 Q1 shipments against a forecasted 2.8% drop. HP shows signs of health by recapturing the number one position from Lenovo with an 8% rise in total unit shipments for Q1 2017. A recent report predicts stability within the note book market as detachable tablets, also known as 2 for 1, sales pickup. Which is a plus a SSD storage but not for HDD. On the plus side for HDD are reports of the commercial computing class market stabilizing as corporate refresh cycles kick into gear.
At Horizon we saw price erosion within the note book space with Toshiba making aggressive price cuts for its 2.5” 1TB drive. This fits in with their strategy of focusing on the personal computer market while the other three slug it out in the data center space.
As we approach the traditional back to school build there are a variety of potential headwinds including the combination of a tight SSD market, production limitations due to HDD componentry shortages and capacity being diverted to more profitable lines. We may be in for a tight market over the second half of 2017. However a key variable to pay attention that can quickly change market dynamics is 3D NAND yields. If yields improve supply will loosen easing pricing pressure which will put note book drives under further pressure as manufactures looks to further reduce BOM costs by substituting HDD for SSD.
General Thoughts- in conducting my research for this month updates it was clear that there are differing opinions what future demand for storage is going to look like. Some are of the opinion that we are at the tip of the iceberg for date growth. As we build more and more data intensive innovations these innovations in themselves will spawn increasingly data intensive solutions. A Hegelian three beat rhythm of increasingly complex solutions that will eat up storage as fast as we can produce it. Others see supply overtaking demand resulting in the boom bust cycle we have so often seen in the DRAM market which shares many characteristics with NAND. However one difference between then and now is the migration of the personal computing platform to hand held and tablets. With that the dwindling market for client storage and the growing market for enterprise storage. Unlike client SSD that are readily interchangeable enterprise SSD have specific performance criteria to address specific end user’s needs and therefore commands a premium. Will this mean more price stability for the SSD market or will the Open Compute movement push SSD into the commodity role again?
I would love to hear your thoughts.
Please email me at stephen.buckler@horizontechnology.com
