Horizon August 2017 Storage Market Brief

16:32 30 August in Uncategorized

We have seen a slowing within the storage market as we move into Q3 2017. With WD giving conservative guidance, we witnessed Wall Street pull back on stock valuations for both Seagate and WD/HGST.  In the start of 2018, market pundits were forecasting 110 million in HDD sales but recently have walked it back to 105 to 102 million units, due to slowing PC and surveillance sales.  Not only have we seen lower HDD volumes over the past 6 weeks, but also downward pricing pressure.  The SSD market has seen supply catch up with demand within the client and enterprise space from an increase in SSD shipments as TLC based SSD yields improve.

Although HDD fundamentals are weakening, there is guarded optimism as we head into the fall, hoping Intel’s latest processor, Purly, will kick enterprise sales into gear and as they work through their yearend budget. AVGO’s positive outlook for the enterprise market gives credence to OEM’s rosy viewpoint.

The PC market continues to face the headwinds of tight component supply (memory and LCDs), high prices, and slowing demand.  If the enterprise market fails to deliver, to offset a soft PC market, expect to see the market quickly change gears.   With all indications that memory will remain tight throughout 2017 and into Q1 2018, it was a red flag seeing several open computer builders coming to the market with excess enterprise modules.   Reportedly, a hyper converge customer out of the valley de-committed on builds leading to the excess offerings.  Let’s hope this is not harbinger of long-term market weakness.

 

Enterprise HDD

We saw a slight pull-back in the enterprise market with several events contributing to a slow start to the third quarter. Intel’s Purley processor was slow to get out of the gates, not giving the enterprise market the boost to sales it was hoping for.  We heard of delayed builds with a large hyper converged builder reportedly cutting back on a build, resulting in build material (memory and HDD) selling into the open market as excess.  With seven primary participants within the Hyper converged space and a report that one of the Magnificent Seven is catching a cold, the market looks to see if this will be full on flu.

We all recognize storage demand is growing exponentially and the market is migrating to the cloud. Intel is projecting 44 zeta bytes of data to be produced by 2020.   Much of that data is earmarked for the cloud.  Lenovo’s difficulties in the server (on premise) market indicate just how important the hyper converge space is to the storage market, as workloads shift from on-premise to the cloud.   The cloud works on volume, and giving users inexpensive access to data is critical to driving users into the cloud.  Many believe HDDs are still the lowest cost option available for cold storage in the cloud.

There are arguments for maintaining legacy systems, including sunk money in existing infrastructure and code and the high cost of rewriting applications for the cloud. Although the 10k and 15k RPM market is a declining market, Toshiba and Seagate are showing commitment to supporting cloud hold outs with recent product roll outs supporting mission critical applications.     Toshiba introduced the AL14SC HDD line of drives for Q3 volume production. Seagate released its 9th 2.4TB 2.5TB” drive for write intensive applications.

Shifting market conditions put stress on Seagate, who is dealing with quality challenges and a higher cost basis than its competitors.   Seagate missing the 12TB target in the hotly contested HPC market is no joke. Without offsetting desktops sales, Seagate is left in a precarious position with the looming ticking time bomb of SSD.  They cannot afford to fall further behind to chief rival WDC in the hyper converged space. Both Seagate and Toshiba are playing from a defensive position and have been leading with cost to drive qualification as 2nd or 3rd sources for their large cap drives with the big boys within the Hyper Converged Space.  Reportedly, Seagate is quoting $20 to $40 below WDC for 10TB caps to the Amazons and Googles of the world.  Seagate  on their earnings call that they will ship a million 10TB units this quarter.  However, WDC/HGST are several generations deep into their large cap drives, enjoy an 80% share of the large cap market, and have a much better cost basis. We have heard that Seagate is struggling with 12TB qualifications.

Toshiba is struggling to get the air based 8TB drives qualified. Similar to Seagate, they will have to lead with price. With a customer reportedly telling them they must be a sub $160 8TB to get into their Data Center. They are looking to the future by bypassing the 12TB capacity and trying to leap frog to a 14TB Helium SMR drive with 1st half 2018 qualifications.  Based on past history there is question how successful they will be.   More importantly the shift to SMR has implications for the industry with changes in writes/rewrites and the slow IO may result in customer specific SKUs.

The HDD manufacturers will continue to optimize expenses and gather around niche opportunities, such as bulk storage, surveillance, and a shrinking legacy market.

 

Aug SSD

All eyes have been on the acrimonious relationship between Toshiba and WD as they fight for control of their joint venture NAND FAB. Does anyone really remember who had the high bid? This seems to be dragging on and on like a bitter divorce.  However, recent reports indicated that the two were finally sitting at the table to hash out their differences.  Toshiba avoided delisting but cannot afford to allow this to go unresolved for much longer.  They need to get their financial house in order.  While WDC cannot compromise their strategic acquisition of San Disk by losing a stake in the joint venture FAB, currently there is no direct impact on WD’s balance sheet. However, they cannot afford not to have access to technologies for the development of next generation storage.  Unfortunately the latest updated indicates that there is still some distance between them around what WDC’s future ownership stake looks like.

Looking at the competitive SSD landscape, the SSD manufacturers that are vertically integrated (own FAB capacity) fall into the tier 1 bucket and command the lion’s share of enterprise SSD sales, where end users demand performance against the lowest cost option, and the client space, with much loser performance requirements, is driven by cost, allowing for a thriving third party market. Once they work through their issues, WD and Toshiba are positioned to manage the market transition towards SSD, while Seagate is relying on foundry relationships to ensure NAND supply, which could be good or bad.  If SSD capacity surpasses demand, Seagate will be the beneficiaries of cheap NAND without the cost of the capitalization of a FAB. However, realities paint a starker picture.  Not only are they once removed from the supply chain, limiting their control, but they don’t have the same access to technology that WD, Intel, or a Toshiba have for developing new gen products.   When demand outstrips supply, they are relying on their competitors to provide NAND to build SSD.  Not a great position to be in.

The concern over the endurance of TLC based SSD continues to drive open market demand for MLC based SSD. Samsung PM863 and SM863 have been in demand in the open market. Intel’s Intel 3710 and 3610 series have been in high demand due to being end of life, yet its replacement, the S4600 series, will not ramp volume shipments until late Q3.  Although we have seen recent deliveries made for 200GB, 400gb, and 800GB 3710 series parts, they continue to fetch a considerable premium, and we expect this to be temporary as Intel fills last time orders.

The transition to TLC based SSD has been gradual with continued questions about wear performance. We have seen increased volumes of Samsung’s TLC based SSD, PM863a, and SM863a being offered in the open market, and we see prices slide.    We will see greater acceptance of TLC based enterprise SSD, as new generation software is written specifically to take advantage of flashed based SSDs as opposed to current software written to take make SSD work in a HDD environment.  Once this transition takes off, we will see increased adoption of TLC based SSD in the enterprise environment.

 

PC Storage Sales

After an early bump in PC shipments at the beginning of the year, we have seen sales normalize and follow the expected downward trajectory.   As we head into the traditional build season for PCs, the PC manufacturers face numerous headwinds, including a buildup of finished goods inventory in the channel, component shortages (panels and memory), and high prices.  The combination puts pressure on HDD manufacturer sales contending with the cannibalization of HDD sales by SSD as TLC capacity yields improve.  Although high SSD prices have slowed the cannibalization of HDD sales, PC consumers have much lower quality expectations, which makes it a much easier transition to TLC based SSD.  One of the tier 1 SSD manufacturers indicted their backlog has been dramatically reduced over the past 4 weeks. Not good for the HDD manufactures.   A leading indicator of market weakness in the PC space is lowered pricing for the bread and butter 1TB SMR drives.

Bright spots are the hope that gaming PCs and game consoles (Microsoft release of X BOX ONE X) will offset sagging notebook HDD sales. Gaming consoles’ year to date volumes came in at 10.5 million against average yearly sales of 28 million, leaving a backlog of 17.5 million drives to be filled in the back half of 2017 and the hope that the release of Intel’s Coffee Lake-based processors will drive notebooks sales this fall.

The wild card for the PC market is the release of Apple’s iPhone 8 and Samsung’s Note 8 as well as several Chinese handset manufacturers launching new phones. If successful, we may see the launches potential consume NAND capacity and undo recent gains in supply against demand, possibly breathing life into HDD manufacturer’s desktops sales with OEMs substituting HDD for SSD to support the fall launch of Intel’s latest NB CPU.  However, once 3D NAND comes on line in a significant way, expect to see desktop HDD sales drop precipitously.

 

Final Thoughts

Luczo is out as Seagate’ CEO, leaving a company with significant challenges in a shrinking HDD market without a clear answer to SSD. It is simply a matter of supply and cost to drive SSD absorption in the desktop market, while enterprise SSD penetration is less clear.  Questions about TLC wear levels compared to MLC based SSD continue, and SSD is being tethered to SATA and SAS environments.  New technologies, including NVMe and NVMe over fabrics designed specially to take advantage of SSD, will further drive the enterprise transition to SSD.

Again, the wild card is Samsung, Apple’s and Chinese handset manufacturer’s   introduction of their flagship phones, which can tip the demand supply equation into prolonging the SSD shortage. Don’t believe me? Samsung indicated it expects to spend $7B over the next three years to expand NAND production capacity at its Xi’an facility, according to a filing that also indicated it has approved $2.3B in investment.

Stephen Buckler

stephen.buckler@horizontechnology.com