July 10, 2012 - 8:55am EST by
2012 2013
Price: 31.53 EPS $10.15 $7.50
Shares Out. (in M): 269 P/E 3.1x 4.2x
Market Cap (in $M): 8,482 P/FCF 2.8x 4.2x
Net Debt (in $M): -864 EBIT 3,008 2,290
TEV ($): 7,618 TEV/EBIT 2.5x 3.3x

Sign up for free guest access to view investment idea with a 45 days delay.

  • Hardware
  • Industry Consolidation
  • Misunderstood Industry
  • Low multiple
  • FCF yield


 “The reports of my death are greatly exaggerated.”

Mark Twain and the Hard Disk Drive industry


Western Digital

WDC is such a glaring valuation anomaly that it’s practically forcing a look from investors, and the question should not be why to own this stock, but rather why not to own this? The investment thesis on WDC is straightforward - the industry has consolidated down to two major players controlling 80% of the market, which should result in structurally higher profitability. Additionally, an act of God that knocked out a large swathe of industry capacity caused a temporary price spike windfall profits that are being returned to shareholders. The artificially high pricing and margins are not sustainable; both will decline meaningfully from 2012’s elevated level, but we expect them to wind up well above the historical average. In light of that, we chose to value WDC based on a normal operating environment in 2013. The market clearly doesn’t believe in this thesis, otherwise the stock would not be trading at 3.5x 2013’s earnings (inclusive of share buyback) with a 24% FCF yield on today’s market cap (26% on the EV). Expectations are so low that we don’t need many things to go right to make money. If our thesis is correct, the stock has a 100%+ upside opportunity to $72 per share (8x $9 in earnings in 2013, including share buyback). If management got really serious about the low valuation ascribed to their stock, they could buy back an enormous amount of stock (potentially 35-60% of the shares outstanding) and do a public market LBO right in front of us.


March 7th, 2011 is an important date. That’s the day WDC kicked off the final round of industry consolidation. The floods in Thailand are seven months away. WDC is a year away from announcing the intent to repurchase ~20% of their stock. And the new ‘normal’ operating environment of higher profitability is still 24 months out. It’s safe to say that WDC has a much improved position relative to where they stood on March 7th, 2011, yet their stock price is lower than at that point. That makes no sense.


Key drivers of the investment case

1)      HDDs are not going away

2)      Pricing and margins will be structurally higher going forward due to industry consolidation

3)      Management of both major players are being disciplined, and could return even more capital to shareholders

4)      Valuation



Industry Background

Hard Disk Drives (HDDs) are vital to the storage of information. Over the past ~25-30 years HDDs have supplanted tape drives to become the dominant media for storing digital information. In 2010, notebooks were the largest end market for HDD units, accounting for 42% of the total, followed by desktops at 39%, consumer electronics (think DVRs, gaming consoles, external HDDs) at 14%, and enterprise at 5%. However, the emergence of Flash memory (throughout the report we’ll interchangeably use Flash to also refer to Solid State Drives - SSDs) simultaneous with growth in mobile computing (handsets, tablets, ultrabooks, etc) is causing traditional PC growth to slow and substitution to Flash in notebooks. Additionally, enterprise is starting to be penetrated at the high end. Flash’s competitive advantages are its faster read/write speeds, smaller real estate footprint (total size and thickness), lack of moving parts, and lower power consumption. The major drawback to Flash is its cost, approximately 10x more expensive vs. a HDD on a per gigabyte basis.


Information Creation: the rate of digital information creation is astounding, up over 10x in the past five years, and the incremental growth in 2011 (~600 exabytes) was nearly equivalent to all of the information created in 2009 (650 exabytes), as shown in the table below.


in exabytes (EB)          ’05       ’06       ’07       ’08       ’09       ’10       ‘11

Data created                120      161      280      410      650      1,200   1,800

% change                                 34%     74%     46%     59%     85%     50%


*1 exabyte = 1 billion gigabytes. 1,800 exabytes is equivalent to 450 billion movies at 4 gigabytes each, or equivalent to enough memory to support 28.1billion 64GB iPads.


Data from EMC and IDC.


The explosive growth is being driven by digitized media - streaming movies and TV from Netflix and Hulu, YouTube clips, downloading MP3s from iTunes, recording TV shows onto DVR, taking mobile photos and videos, growth in internet traffic, sending emails, growth in social media (Facebook, Twitter, etc), and cloud storage (iCloud, Amazon Cloud Drive, Google Cloud Storage, Dropbox, Sugarsync, etc. etc). Content creation and consumption per user is growing as are total users.


Demand for Data Storage: Not all of the data that is created gets stored - think of digital TV programs you watch but do not save. Referencing the table below, the storage rate of newly created information in 2006 was 25%, dropping to 13% five years later in 2011, and will continue its downward slope. But the law of large numbers starts to impact demand: consider a hypothetical example where data creation grows 50% in 2012 while the storage rate remains flat at 12.5%. That would require the storage industry (both HDDs and Flash) to supply 338 exabytes of new supply (2,700 EBs x 12.5%), a 50% increase Y/Y. The equation is not perfect, but the key point is that as long as information gets created, some fraction of it will be saved on something.


in exabytes (EB)                      ’05       ’06       ’07       ’08       ’09       ’10       ’11       ’12e


Information created                120      160      280      410      650      1,200   1,800   2,700

Total storage capacity             110      150      260      390      410      625      850      1,188

% change                                             36%     73%     50%     5%       52%     36%     40%


Incremental capacity addition                        40        110      130      20        215      225      338

Storage rate*                                       25%     39%     31%     3%       18%     13%     13%

*Storage rate attempts to measure how much of newly created information actually gets stored in a particular year. Equals Incremental capacity addition divided by information created that year.


Data from EMC and IDC.



Storage Production: What the following table attempts to capture is the memory storage production per year. Numbers are approximate, but the trend is accurate.


in exabytes (EB)                                  ’06       ’07       ’08       ’09       ’10       ’11       ‘12e     ‘13e


HDD units, millions                            437      500      540      557      653      623      672      720

Capacity per HDD, in GBs                 125      156      195      244      305      381      477      596

Total storage production, in EBs        55        78        105      136      199      238      321      429

% change                                                         43%     35%     29%     47%     19%     35%     34%


Flash production                                 1          2          5          7          10        17        29        49

% change                                                         100%   150%   40%     43%     70%     71%     59%    


Total storage production                     56        80        110      143      209      255      350      475

% change                                                         44%     38%     29%     46%     22%     37%     36%


Flash data from iSupply, HDD data from IDC and various sell-side research.



Key drivers of the investment case

1)      HDDs are not in secular decline.

This is the area of greatest controversy and misperception. Consensus opinion is terrified by the thought that HDDs could be completely obsoleted in a few short years by Flash memory, which sounds scary until you think analytically about it. It’s an impossibility for Flash to totally displace HDDs. We’d agree that Flash memory will continue to take share, but that does not mean HDDs cannot grow modestly. This is a very contrarian viewpoint, but one we can support with logic:

a)      Information creation continues to grow and the need for storage grows along with it

b)      HDDs are the lowest priced storage medium. Flash costs 10x as much per gigabyte of storage. HDDs enjoy a significant cost advantage, mainly because of the fixed cost nature of a basic hard drive. Each unit requires a controller, head, suspension, arm, platter, etc, regardless of capacity size. As the size of the drive increases, the incremental cost to build a larger HDD declines meaningfully. The incremental cost to build a 1TB drive vs. a 500GB may only be a ~25% difference, an additional 1-2 discs plus 1-2 heads, yet has 2x the capacity. Flash operates closer to a linear cost curve, with the cost to produce 256 GB ~2x the cost to produce 128 GB.

c)      Cannibalization of mobile is partially offset by growth in Cloud storage. In olden days one’s digital media – photos, music, etc, would get stored on their desktop HDD. The enormous growth of mobile computing with much lower storage capacities (500GB desktop hard drive vs. 32GB tablet, for instance) is requiring a secondary source of storage. Presumably data that does not (or cannot) get stored on an 8GB iPhone (photos, video, MP3s) is getting stored somewhere in the Cloud. Proof is the proliferation of Cloud-based storage services – iCloud,Amazon Cloud Drive, Google Cloud storage, Dropbox, Sugarsync, etc, and that’s good for HDDs because they buy HDDs, the cheapest solution. Demand from cloud storage will likely be a very significant driver of demand for HDDs.

d)     Hybrid drives could be a realistic competitor to Flash in certain mobile applications. A hybrid drive, comprised of both an HDD and Flash memory, offers the best of both worlds – much higher storage capacity and more comparable performance in terms of speed and power consumption, relative to pure Flash, while costing considerably less.

e)      Desktop PCs, are not dead yet, exhibiting very modest growth. Clearly in western countries desktops are struggling to show any growth, but that’s a very Western-centric view that overlooks growth in emerging countries where desktops, because of price, are the cheapest computing option. As the cost of notebooks decline their adoption will increase, but that will take time to play out.

f)       There is simply not enough Flash capacity in the world to cannibalize 100% of HDD supply. If the incremental demand for data storage in 2012 is 338 exabytes (50% data growth at a 12.5% storage rate), that’s 12x the estimated Flash capacity in 2012. It’s physically impossible for Flash to meet that demand and eliminate the role of HDDs.  

g)      The cost of adding new greenfield Flash capacity is prohibitively expensive. Samsung recently completed building the world’s largest fab, the 300mm Fab 16, capable of producing 5 exabytes of storage per year, or ~20% of the industry capacity pre its birth. The cost of Fab 16 was $10 billion, or $2 per gigabyte of capacity ($2 billion per exabyte). To match HDD capacity, Flash OEMs would need to build 279 exabytes of capacity, or 10x their current production, at a cost $558 billion (279 exabytes x $2 billion per exabyte). The industry doesn’t have that kind of money.

h)      The threat of virtualization has already occurred. Servers and storage equipment have been getting ‘virtualized’ for 10+ years, this is not new, and has mostly played out, yet the HDD industry was able to grow through this. If today was day 1 of virtualization, I would have a different opinion.



2)      Pricing and margins are structurally higher going forward due to industry consolidation

Two major players now control 80% of the market which should lead to rational decisions regarding pricing, capex spend, and capital allocation. The industry has seen over 100 competitors come and go over the past 30 years, and at its peak had 25 OEMs. The past six months have seen the final step of consolidation down to two major players – Seagate took out Samsung in December 2011 and Western Digital took outHitachi in April 2012. Today, WDC has ~40% market share, STX has ~40%, and the only other competitor, Toshiba, has ~20%.


Historically, it’s hard to find an industry that experienced meaningful consolidation and did not see a structural change to profitability. In a duopoly market the two dominant players presumably will find no incentive to compete intensely on pricing, which is especially true if no player has sufficient excess capacity with which to use price as a lever. The U.S. steel industry is the only one that jumps to mind where consolidation helped for a brief time period (2004-2007) but was subsequently overwhelmed by the macro (U.S. housing and construction bust, recession, and China).  The point is that going forward, this industry’s profitability should be under a more favorable industry structure than that looking backward.

  1. aerospace and defense
  2. airlines
  3. beverage packaging (Ball Corp, e.g.)
  4. railroads
  5. steel
  6. HDDs ?



An additional factor that temporarily pushed up pricing is due to the supply shock from the Thailand floods. Thailand was home to 40-50% of global HDD production that got knocked out for nearly 6 weeks beginning in October 2011 due to severe flooding, causing pricing to spike 50%+. Expectations are for industry production to return to pre-flood levels by the September 2012 quarter. Consensus is confused as to what normalized pricing is, but we believe it will remain above pre-flood levels for the following reasons:

a)      Industry consolidation, as discussed previously

b)      Greater scale – with the acquisition ofHitachi, WDC’s capacity goes up by 50%. Although they are not allowed to fully integrate the two businesses (China Ministry of commerce rule), there’s still synergies that can be gleaned in terms of purchasing power, increased internal sourcing, etc.

c)      Supply chain inventories are still incredibly lean, which will help add to demand for a few quarters

d)     Production costs could remain permanently higher post-flooding, exerting upward pressure on pricing. HDD OEMs helped get their vendor base back to healthy levels by paying up for components, subsidizing them directly with upfront payments, paying faster, etc. And HDD OEMs have begun to diversify away fromThailandto countries that are modestly higher cost (Thailandis a cheap place to manufacture, that’s why so much manufacturing wound up in there in the first place).

e)      Could customers be willing to pay a small premium to ensure supply? Not having access to HDDs meant OEMs could not sell equipment (PCs, enterprise boxes) and could not recognize revenues or cash flow. Possibly ensuring supply is worth a slight premium.


We additionally think gross margins normalize at a level 400-500 basis points higher than their historical average. WDC’s 10 year average gross margin is 18%, the 5 year average is 20%, and they printed 20% during the September 2011 quarter, the last ‘normal’ pre-flood. Then the Thai floods occurred, boosting pricing which lead to a 1,200 basis points jump in the gross margin to 32% in the March 2012 quarter. That level of profitability is not sustainable, however the new normal should be mid 20% range. It’s hard to believe that the impact consolidation will have on pricing that profitability and margins could worsen to a point where the industry is actually less profitable. An interesting question is whether or not WDC was under earning in March – they faced significant manufacturing absorption costs with volumes 24% lower than September 2011 quarter, and they were subsidizing their supply chain. Could margins have been higher?  



3)      Management of both major players are being disciplined

We have solid evidence of discipline within the industry:

a)      Capex plans are not aggressive and the money is mostly being spent on getting production back to pre-flood levels, not on new capacity, despite having record operating cash flow

b)      Windfall profits earned this year are getting returned to shareholders in the form of buybacks and dividends. In January of this year STX committed to a buyback equivalent to 25% of their outstanding stock, $4 billion, within twelve months while increasing their dividend 39%, and in May WDC announced a $1.5 billion buyback of nearly 20% of their stock (5 year authorization).


It’s hard to understate the enormity of the signal WDC sent with their buyback announcement. Between 2002 – 2011, WDC never was a net buyer of its own stock, issuing a cumulative $140mm and paying zero dividends. They hoarded cash and built up a massive war chest that one point was equal to half of their market cap. Then suddenly management announces a huge buyback after having already repurchased $305 million in the June quarter. That signals a sea change to management’s thinking, and that they “get it” regarding capital allocation – low growth businesses generating humongous cash flow should be buying back stock when it trades this cheaply. No doubt WDC management was under pressure by investors to enact a similar capital allocation strategy to STX. The question to ask is whether or not more capital will be returned to shareholders. And if management “gets it” on capital allocation and moderate capex spend, shouldn’t they also “get it” in terms of remaining disciplined on pricing?



4)      Valuation

We value WDC based on a more normal environment in 2013, which is more indicative of true normal earnings power, and heavily discount 2012’s artificially high profitability (although we’ll gladly take the cash flow!).


WDC’s stock is ridiculously cheap on our 2013 base case, trading at 4x earnings of $7.50 per share, 2.3x EBITDA, and generating an impressive 24% of its market cap in free cash flow. To reiterate, our 2013 forecast incorporates a price and margin decline. An additional source of upside is the company’s $1.5 billion share buyback authorization, that if somehow could be completed in one fell swoop at today’s price, would takeout 48mm shares, or 18% of the share count, pushing 2013’s EPS estimate higher by $1.55 to $9.00 per share, equating to a 3.5x P/E multiple. See the base case fundamentals table and valuation table below.


Base Case Fundamentals, no share buyback

In $ billions unless otherwise noted

                                CY10              CY11              CY12*                CY13

Units, millions             204                  190                  264                  328

ASP**                        $49                  $49                  $63                  $57

Sales                          9.9                   9.3                   16.7                 18.7

Gross margin              21.3%              22.1%              30.4%              24.0%

OpEx                          0.9                   1.1                   2.1                   2.2

EBIT                           1.2                   1.0                   3.0                   2.3

D&A                            0.5                   0.6                   0.8                   1.0                              

EBITDA                       1.8                   1.6                   4.0                   3.3

Interest exp.                -                       -                       -                    -0.2                 

Pretax profit                1.2                   1.0                   3.0                   2.2

Taxes                           0.1                   0.1                   0.3                   0.2

Tax rate                       8%                   8%                   9%                   9%

Net profit                    1.1                   0.9                   2.7                   2.0

Avg shares O/S, mm   235                  237                  263                  269

EPS                             $4.75               $3.94               $10.15             $7.50              

OCF                            1.9                   1.5                   4.2                   3.1

Capex                         -0.8                  -0.6                  -1.2                  -1.1     

FCF                             1.0                   0.9                   3.0                   2.0


*Acquisition of Hitachi HDD business closed on March 8, 2012

**2012 ASP begins the year at $70 but exits at $55, -21% from peak.



Valuation                   CY11              CY12              CY13

P/E                              8.0x                 3.1x                 4.2x

EV/EBITDA                  4.7x                 1.9x                 2.3x

EV/Sales                     0.82x               0.46x               0.41x

FCF/EV                      12%                 40%                 26%

FCF/Mkt Cap              11%                 36%                 24%



If management got serious and really wanted to take advantage of the fact their equity is dirt cheap (or cheaper?), what level of buyback could they do, assuming they can buyback all they want at today’s price?

a)      Spending 100% of FCF this year ($3.0 billion) would enable a 95 million share repurchase, or 35% of the total outstanding share count. Earnings power could be $11.50! ($2.0 billion in net profit divided by 174 million shares)

b)      Leverage the balance sheet up to 1x net debt/EBITDA would give management $4.2bn of firepower ($900 million net cash currently on the balance sheet plus $3.3 billion of additional borrowings) could takeout 133 million shares, or 50% of the outstanding shares. Earnings power could be $14.75! ($2.0 billion in net profit divided by 136 million shares). The free cash flow generated from 2012-2013 would more than payoff the debt leverage!

c)      Spending 100% of FCF both this year and next ($5 billion) would enable a 158 million share repo, or 59% of the total outstanding shares. Earnings power could be $18! ($2.0 billion in net profit divided by 111 million shares)


The amount of cash flow this company can generate makes it totally unbelievable as to why a takeout has not already occurred or a stealth-public market LBO is not underway.


We are not suggesting that we think the company is going to buy back 59% of its outstanding shares.  Either we are very wrong on earnings power, management is going to colossally fail on capital allocation, or the stock price is going to go up significantly making the above analysis moot.


The disaster scenario is as follows: unit shipment show no growth off of 4Q12 run-rate of 328mm drives (82mm per quarter), pricing is the worst in at least 6 years at $45 per drive, revenues fall to $14.8 billion, 20% gross margin (about in line historical average), $3.0bn in gross profit, $2.0 billion of OpEx, $1.0bn in EBIT, $1 billion in D&A, $2.0 billion in EBITDA, $230 million in interest expense, $770 million in pretax profit, 9% tax rate, $700 million in net profit, $2.60 in EPS (ex any buyback), $1.7 billion in OCF, $1.0 billion of capex, and $700 million of FCF. That works out to a valuation of: 12.1 x earnings, 3.8x EBITDA, and a 8% FCF yield on market cap. The numbers look ugly, but the valuation still looks reasonable.



[The views expressed are those of the author and do not necessarily represent the views of any other person. The information herein is obtained from public sources believed to be accurate, reliable and current as of the date of writing.  The author will not undertake to supplement, update or revise such information at a later date.  The author may hold a position in the securities discussed.]




    show   sort by    
      Back to top