Description

It’s finally time to short hard drives!!  I am short Seagate Technology (STX).

Before I begin let me note that Im a generalist value investor, and not a tech innovation groupie who has been crowing the merits of shorting the HDD companies since 2009.  On the contrary, I was long HDDs for a few years starting in early 2011 (post-merger announcements and pre-Thai floods) on an industry consolidation and pricing power thesis.  In 2011 I believed that HDDs would be around in a meaningful way for longer than people expected.

Four years later what appeals to me about the short opportunity is that the smart money and everyone else has stopped paying attention – investors have been lulled into a state of buyback and dividend induced coma precisely at the same time the risks to the oligopoly structure and fundamental demand are higher than they have ever been.  Both the March and June 2015 quarters will be the worst quarters that either STX or WDC have printed since the Thai floods in late 2011 (but you wouldn’t know that looking at their stock price charts).  

The industry does admit that their traditional business is under pressure but they have convinced people there is enough growth in high capacity enterprise business to support growing consolidated revenue.  They make this assertion despite the fact that organic revenue growth has been clearly negative for the last few years.

Both WDC and STX trade at a similar multiple of just north of 10x forward P/E.  I prefer shorting STX, as WDC has executed reasonably well in SSDs, has material potential synergies ($500mm-$1bn?) if MOFCOM approves a WDC/Hitachi subsidiary consolidation, and is net cash [note that the one negative is that it is has an inferior tax structure resulting in a bigger offshore cash problem].

First, some basics

It is worth having these fact sheets from STX and WDC handy:

http://www.seagate.com/www-content/investors/en-us/docs/fiscal-q3-2015-supplemental-information.pdf

http://investor.wdc.com/common/download/download.cfm?companyid=WDC&fileid=824315&filekey=9BA16E20-0A00-40E1-9B98-686C94ABDA0A&filename=FY15Q3_QFS_Version_1_draft_5_.pdf

-    We have a declining business trading at a 10x-11x forward P/E (based on consensus)

     o    Declining segments: PC (notebook and desktop), high performance enterprise (10k and 15k RPM HDDs)

     o    Flattish segments (aka future decline candidates): branded (the “backup” drives you see at Costco and Best Buy), consumer electronics (primarily gaming consoles and DVR)

     o    Stable/Growing segments: high capacity enterprise (aka nearline)

     o    ~10% of industry units fall into this growing high capacity enterprise category (note that the HDD industry conveniently lumps all enterprise drives into one bucket so investors cant see the decline in high performance drives or the growth in high capacity drives)

     o    While the high capacity drive revenues are growing the revenue growth is likely underwhelming.  The HDD companies refuse to disclose the revenues or the revenue growth rates in this segment.  I don’t have access to Gartner or IDC data but I suspect it would show enterprise nearline revenue growth rate of 10%-20%, or enough to offset the decline in high performance enterprise drives and show a tiny bit of net growth

     o    HDD companies also make the case that margin contribution of these enterprise drives is higher than other types of drives.  This is true but the HDD volume that will disappear with time has also been absorbing costs

-    Revenue and earnings have been eroding post flood.  This is despite multiple tuck in acquisitions that have contributed material revenues and in some cases earnings

     o    Seagate acquisitions: Xyratex ($374mm purchase price, added $500mm-$600mm of revenue), LSI flash solutions ($450mm p.p.), Lacie (branded business, ~$175mm p.p., ~$300mm of revenue))

     o    WDC acquisitions: Virident ($685mm p.p.), STEC ($340mm p.p.), Skyera (Note: WDC’s SSD business is at ~$1bn revenues per year)

Why short now?

-    PC demand is in secular decline and the decline could be accelerating

o    The last quarter was just a taste of what is to come.  Yes, people will still buy PCs.  But large form factor phones and Chromebooks are capable of a lot – who would have expected to make bank check deposits from your phone a few years ago?  Said differently, while phones and Chromebooks aren’t considered PCs, they are mobile computing devices where the hardware and apps are becoming more useful and sophisticated with time.   What will happen to the usage of PCs when you can accurately dictate to your phone?  For US readers, keep in mind that the rest of the world pays up front for a mobile phone and who would argue that a new PC matters more to the average person than a new phone?  Qualcomm’s India country manager was recently quoted saying “The right kind of word to describe what has happened in the smartphone segment is probably revolution. The way it's happening, perhaps at some point in time, one won't need the laptop. Lots of the most recent compelling applications and services are being designed for smartphones first. The number of smartphones being sold is far more than laptops, and not only that, the amount of time the younger generation spend on smartphones is far more.”

Incidentally, mobile phones becoming more tablet-like led to a tablet market that is now in decline.  Lower PC volumes = lower HDD units

-    SSD penetration rate in notebooks is at ~20% and is now hitting the sweet spot of the adoption curve; this adoption will be accelerated by step function improvements in 3D-NAND  cost and performance vs. planar NAND

o    3D NAND is going to drive a cost decline curve that is steeper than the current planar NAND cost decline curve and is also at least 20 percentage point higher than HDDs on an annual basis for each of the next 3 years.  At that point HDDs may use HAMR technology and the cost curves might become similar… but a cumulative >60% relative gain for NAND over 3 years would be significant

o    In addition to the step function lower costs with 3D NAND, you also get better performance than planar NAND: better read/write endurance, higher bandwidth, better retention (at least some providers like INTC/MU are claiming this).  Memory companies are promising 3D TLC solutions with performance characteristics able to address market segments that planar TLC could never address.  3D-NAND will continue to have the other favorable characteristics that consumers value in SSDs today – namely smaller form factor, lower energy consumption and durability.

o     Yes, HDD sales for PCs have thus far held up better than a lot of people expected but there is an important wrinkle.  Many PCs with hard drives have some amount of flash today (either incorporated into the HDD (what Seagate call a hybrid drive) or with a flash module (in addition to the HDD) that is home to the key boot or execution files.  This HDD + NAND approach is a temporary solution – people want the performance of NAND but historically havent been able to afford all-flash solutions.  The number of PCs that have an HDD but also have <$10 worth of enbedded flash is tens of millions per year.  

o    an OEM with $50 gets ~128 GB of client SSD capacity today and in 3 years should be able to buy ~350 GB of SSD capacity for that same $50.  An OEM might be able to spend $20 in 3 years and get a 100-128GB SSD while a PC HDD should cost $30 at a minimum (while the HDD would have more storage than the SSD the minimum cost of an HDD does not decline given the mimimum BOM of having one platter, one head, etc)

o    It is worth noting that flash is cannibalizing both high end and low end HDD sales.  Chromebooks and most of Apple’s PCs use NAND storage.  It’s rare when a technology addresses both end markets

o    The combination of a faster reduction in NAND costs and higher performance = lower sales of the highest margin HDDs into both enterprise (high performance drives) and PCs (notebooks with thin hard drives)

o    3D NAND products will hit the market in 2016 (with some Samsung products in 2H 2015)

o    Intel and Micron have also been working on new storage class memories which some tech folks think will be disruptive but that remains to be seen

-    As Fred Hickey (High-Tech strategist) readers already know, Fred is a highly respected tech investor who thinks there has been a huge overbuilding of cloud/datacenter capacity over the last few years and there are a some signs that companies are waking up to reality

o    While I don’t follow the datacenter market closely, I suspect he has the broad strokes right… it was vogue to have a cloud business and free money and no earnings expectations can be a powerful investment driver

o    HDDs for data centers and hyperscale clients have been the only source of growth for STX and WDC and so if and when that business cracks it will get ugly

o    YTD the HDD companies have been painting a picture of cloud/hyper scale demand that is H2 2015 weighted (the ole "in the future")

o    Dave Mosley, STX’s operations guy recently said of the leading cloud players “just one little optimization tweak, they might be able to put off their buys for 3 months or something like that”.  The leading cloud and hyperscale storage providers take an approach to storage that individual consumers rarely do – they use compression and they utilize close to 100% of their capacity before investing in more.

-    The oligopoly isn’t as strong as you think it is and declining demand will test industry behavior

o    STX and WDC will take a disproportionate hit as Toshiba has no reason to reduce units.  The STX CFO recently said “they [Toshiba] have capacity of 22mm drives a quarter, and they want to move 22 million drives per quarter.  So you let them move 22mm drives a quarter”

o    MOFCOM (Chinese antitrust) and China Inc. are asserting themselves in technology as we can see by their actions with Qualcomm, other semis, and technology more broadly.  

o    More than 3 years after the HDD mergers, MOFCOM is still requiring WD and Hitachi to operate as independent subsidiaries!  AND requiring STX and Samsung to operate as separate units.  MOFCOM doesn’t like 3 supplier markets

o    MOFCOM is smart enough to know the future of HDDs are high capacity nearline drives (currently at 6 and 8 terabytes per drive) – given that Toshiba isn’t a significant player in this market it begs the question whether MOFCOM would force a break-up of the current duopoly in the only market for HDDs that is growing or force some significant price declines.  It is a threat, not an opportunity, that Toshiba is subscale and doesn’t have the appropriate resources to invest in new products like 8TB helium drives.

o    Some industry participants (mainly STX) have started to express concerns over the HDD pricing environment

A few other things to keep in mind

-    The trouble with the bull case

o    one key thing that people forget to talk about is that HDD data transfer rates don’t improve with time.  This is because you can only spin an HDD so fast before heat, power, reliability and other physical/mechanical limitations kick in.  The speedier HDDs transfer data at a rate of low 100s MB / S (megabytes per second) and the cheaper PC HDDs may be at half of that speed depending on conditions.  These transfer speeds don’t change with time even though the storage density and HDD capacity growth with time.  10 years ago an 80 megabytes per second transfer rate was fine, but as the rest of the tech world grows geometrically and we move into the world of GBs, terabytes, petabytes, exabytes, it’s just a matter of time that HDDs become less useful (the analogy of drinking a pool through a straw has been used)…. Meanwhile, SSDs have forced the evolution of interface standards (SAS, PCIe, etc) so that data can move at a rate closer to the limits of the underlying NAND technology (a Macbook Pro might have a data transfer rate closer to 1GB / s).  3D NAND coupled with evolving interfaces will support even faster transfer rates.  This HDD mechanical limitation issue is also why it takes the OEMs “forever” to test an HDD.  Seagate recently commented that between wafer and test time you have >20 weeks in the supply chain. HDD companies also have to spend more money on testing equipment and that is why capex isn’t declining as much as you would think despite a declining TAM.  

o    Within the next 5 years it seems very likely that HDDs will become the equivalent of what tape is today – a cheap technology with severe performance limitations that is useful for mass storage and backup of cold data.  

o    Cloud storage is cheap and easy and most importantly, efficient.  Amazon offers unlimited pictures + 5GB other data for $13 per year; Google offer 15GB free, Sugar Sync ran a promotion this week for referrals where 250GB is $25 for one year.  In the future, there will be few reasons for individuals to have HDDs; people will have NAND based solutions in their phone and other computing and gaming devices and the cloud will have a mix of NAND (for speed) and HDDs (for capacity).  The cloud storage players don’t need performance – they buy for cost and optimize compression so that 20 or 100 people might share an HDD.  Data may be growing but moving data from individual users to the cloud is not good for HDDs all else equal.

-    Insider sales

o    Steve Luczo is a great trader of STX stock , having bought 1 million shares at less than $5 in the panic of 2009 [and also having led the original LBO which made Silver Lake several billion dollars and likely made him several hundred million dollars]

o    In mid-2012 he told shareholders he was would not sell more stock (other than an existing 10b5-1 running through 2012).  He kept that informal commitment for less than a year.  He has been consistently monetizing $30mm+ of stock for each of the last few years.  He has sold 230,000 shares YTD.  I very much doubt he needs to sell given his other sources of wealth and liquidity.  Steve is a commercial CEO and at minimum his selling should be a signal that he doesn’t think the stock is undervalued.

-     Ultimately the HDD companies seem very unlikely to win in NAND

o    There is probably close to a 0% chance that Samsung, Hynix and Micron feel that it is acceptable to cede the enterprise SSD business to other market participants over the medium to longer term

o    WDC seems to be saying that while they can do without a memory fab in the short to medium run in the long run it makes sense to have access to a memory fab

o    WDC and STX are both at risk of overpaying for NAND or flash solution assets as they are forced to remake their business at an accelerated pace; It’s likely that STX and WDC weren’t the only bidders for LSI flash, Virident, STEC, Skyera, etc…

o     STX has already showed signs of desperation; Micron recently commented at an investor event that they got the better end of the Micron/STX SAS SSD joint development deal

-    STX has backed themselves into a corner with a large dividend that consumes most of their FCF

o    Last quarter they reported $1.08 of non-GAAP earnings and paid a 54 cent dividend.

o    The company also says they want to raise the dividend from the current $2.16 annual rate by 10% each year for the next couple years (and do this while maintain an investment grade rating).  This guidance gets us to an expected dividend of $2.87 in 3 years.  As I discuss below I think earnings are very unlikely to support this level of dividend.

o    Seagate can continue paying the dividend for a long time from cash on hand and credit facilities but I doubt they would be so foolish if the business were clearly in trouble [I havent checked but in todays credit markets what are the odds that an IG company has material covenants?].  While a dividend cut would be bad, STX isn’t a stranger to cutting their dividend – STX sported a ~2.5% dividend yield in 2008 before the dividend was abandoned in the recession.

My View and Valuation

STX trades at 12x last quarter annualized and 10x forward earnings estimates.  Seagate has reasonable quality of earnings and D&A/capex is fairly well matched over time, resulting in a free cash flow yields that is similar to the earnings yield.  

The important question is: what are industry profits over time?  Earnings are likely to go significantly lower due to the reasons discussed above.  I suspect COGS are mostly variable over time and in a lumpy way (the supply chain employees several hundred thousand people and lots of capital equipment given the large number of parts and precision, electro-mechanical nature of the system).  I consider opex to be a fixed cost because that spending is needed to try to evolve the business model, both in SSDs and in driving HDD areal density and solutions.  Assuming STX loses 10% of its sales over the next year at a 35% decremental margin (i.e., just a tad higher than current GMs) and that all of that falls to the bottom line given their low and mostly fixed-dollar tax structure, that results in an EPS hit of approximately $1.40.  That is large percentage of the $4.32 LQA earnings base.  The reality is that over 2 or 3 years sales are likely to drop more than 10% and the company will have to make tough choices between investing in the business and paying a large dividend.  

This is the type of business where in a bad cycle earnings go to 0 or negative and the stock trades down to a single digit multiple of future normalized EPS using conservative inputs.  I think we are headed back to a world where $2 of normalized earnings is what STX shareholders expect.  The HDD industry macro sensitivity is also exacerbated by the fact that macro weakness likely leads to lower NAND prices and greater SSD substitution.  

Last week I was busy researching STX and I noticed on that day trading volume was 2 million shares.  In 2011 and 2012 50 million and 100 million share days happened with surprising frequency.  The perception of a stable oligopoly and investors seeking a dividend yield have brought stability to the stock.  

Seagate is guiding to 5% to 7% revenue growth starting in fiscal 2017.  If an HDD company guiding to a specific revenue growth rate years into the future (despite the current negative organic growth rate and despite missing their own commentary and guidance for TAM, revenues, and earnings with surprising frequency) isn’t a sign of the top I don’t know what is.  You have to be an optimist to run an HDD business and if you follow management commentary over many years it becomes clear that they have no ability to forecast their business.  It is comical that Seagate has been talking about a shortage of media and heads since 2011.  Where is the shortage?  Eventually the numbers will be too real to ignore and slap people across the face.  The stability in the stock is not sustainable.

Over the long term I suspect the industry will produce a lot fewer drives (maybe 100mm high capacity drives annually) at a nice ASP (pick a number - $150?) and will be able to make a living doing that at a much reduced cost structure.  Between here and there it’s likely to be a rough ride as STX and WDC figure out what they want to be when they grow old.

A few parting pictures.  

Note that the June 2015 quarter should look worse than the March 2015 quarter based on STX and WDC guidance and the continued very weak trends in PCs and tech more broadly.

fyi, the Thai floods happened in Oct. 2011.  By CY Q3 2012 conditions were normalized

Catalyst

Earnings.  Eventual dividend cut.