How Long Will The HDD Dominate Data Storage? More Than Long Enough

Includes: HTCH, STX, WDC
by: Thomas Lewis

In my previous contribution to Seeking Alpha (Hutchinson Technology: Extraordinarily Undervalued, About to Become a Lot Less So) readers were asked to accept at face value a set of assertions recently made by the two CEOs whose companies control about 87% of the hard disk drive (HDD) industry. This article is the more or less promised examination of those assertions. What might be termed the Luczo-Milligan Paradigm posits that over the balance of the decade, demand for data storage will grow at a staggering rate (33%+ CAGR) and that most of that data (85%) will be stored on HDDs. It was further noted that keeping up with this demand growth by improving areal density (i.e., how densely data can be packed onto storage media) has gotten more difficult, which means that HDDs will be using more heads & disks and so the suspension assemblies which "manage" every head over every disk. The valuation of Hutchinson (NASDAQ:HTCH) ($3 when I started writing) relative to the posited intrinsic value of company was such that a profitable investment might be realized even if the future did not quite live up to this expectation. Getting at just what the full investment potential of HTCH [and Seagate (NASDAQ:STX) and Western Digital (NYSE:WDC) for that matter] might be entails further study of this worldview. After all, there is no law of nature dictating that CEOs are somehow more immune than the rest of us to the Kool-Aid of popular delusion (Exhibit A: Nasdaq @ 5000, c. 1999). (See here, here and here for more on these three stocks.)

Bottom Line: it's even better than I was thinking. This article will delineate a half dozen or so assumptions about the next five or so years that the reader is invited to consider. If these assumptions seem likely to turn out mostly true, then the three or so year investment outlook for HTCH, STX & WDC will be compelling. If some meaningful combination of them appears improbable or downright wrong, then you should consider prospecting elsewhere, or placing a bet against these stocks. But before I can get to my case against the imminent demise of hard disk storage, some clarifications are in order. The first would be to remind the reader that we are making assumptions and not predictions about a thoroughly inscrutable future. An investment in any stock is a bet against all the trouble in the world. Thirty years at this craft and an avid interest in history have informed me that there is no limit to the number of things that could come along and chase money out of equities. This article only touches on that handful of issues that are most germane to "the non-imminent demise" of these suppliers.

The assumptions I am explicating are specific to the aforementioned five or so year time frame, and I hold no brief as to what either the immediate near or indefinite distant future might hold for these stocks. Neither is worth too much mental effort. In the case of the former, we can rest assured that sometime in the next year or so, there will be yet another adverse mood change. When that seeming "change in the cue cards" happens, or whether it's benign like in 2010 or 2012, or nasty like Q3 11, or stark raving, hopefully-once-in-a-generation ugly like in 2008, is guesswork. But it will happen, and when it does, we will all but certainly be reminded that stocks that have "worked" like these ones have lately tend to be in the hands of "investors" who decide to sell with even less thought than went into the decision to buy, which is to say hardly any at all.

The indefinite distant future is synonymous with a word we have seen used a lot in the case against HDD technology: Eventually. This is a dangerous word for astute investors. Eventually, we are all dead, the Constitution and so our blessed experiment in ordered liberty is no more, and the sun runs out of hydrogen and begins its progression to red giant and then cold cinder. Eventually, for one reason or another, spinning disks will no longer be the primary means of data storage, but in the eight years I have been paying attention, eventually has shown no signs of getting any closer to making its presence felt. (I would appreciate hearing from anyone who can help me out here. If you know of a device that will supplant HDD for mass storage, please let me know. Just be sure to include a capacity point, a unit price and an availability date.) The case for owning stocks like HTCH and, presumably, holding them through a Market mood swing or two, does not rest on a slew of eventualities. It rests on an assessment of the likelihood that a couple of trends will persist and that a couple of other things will not change much over the next five or so years.

So what do we have to believe to like the risk/reward of the HDD makers and their suppliers, besides management continuing to execute in the manner which I have observed over the past several years, and the Four Horsemen and their innumerable minions remaining at bay? Or put the other way, where is the industry or technology specific weak link in the HDD story, and so perhaps the basis for betting against it?

Data generation, and so the need for places to store it, will continue to grow at crazy rates. 30%? 40%? 50%? They all seem like guesses to me, but the growth rate for data generation and the zettabytes that will be needed is clearly a great big number, and there is no good reason to expect it to slow to a more GDP-like growth rate in the next five years. How do we know? It is a function of more people (the second billion getting connected a lot faster than it took to get the first billion) with more data capture devices (many with more than one, some with several) presuming on bandwidth improvements and so sending more data-rich content (as in HD). It can't really be precisely measured, like a temperature, but we know it is huge and absent some awful extraneous shock nowhere near slowing down. It's those 350MM photos being uploaded to Facebook (NASDAQ:FB), or that 86K hours of video added to You Tube, every day. For me, the defining anecdote came several years ago as I passed by a tiny church in a tiny town one Sunday in rural Texas. A First Communion service had adjourned to the parking lot, and the children were surrounded by dozens of friends and family, who seemed for the most part to be immigrants from points not very far south. And it seemed like every one of them had their cameras and phones out, snapping away. I wondered how often how many of those pictures got shared. (I can wonder today how many of those pictures are still taking up space at Facebook or some photo sharing site.) We can be sure that 100s of millions of even less well situated people have since that time joined the "connectedness on terms of one's own choosing" that is the World Wide Web, and that there are still a billion or more who would like to.

Areal Density will continue to improve (but not enough to matter). The simply amazing improvements in areal density that have been wrought by both HDD and NAND makers will continue, but the way forward has gotten a lot tougher for both of them. HDD makers have gotten about all the AD they are going to get from PMR (perpendicular media recording) technology). A couple of new technologies (Heat assist and shingled media) will like give AD a boost in the 2014-15 time frame, but not quite enough to catch up with data growth. In terms of physics, NAND technology seems to be up against a more daunting barrier. Both technologies must deal with the challenge that as the memory cells get smaller and closer together, it gets harder to make them hold a charge when you want them charged and release it when you don't. This property, known as coercivity, is specific to particular elements. Not surprisingly, the ferritic material used in HDD is a lot more coercive that silicon. As I understand it, this has something to do with the write-cycle limitations that seemed to arise as MLC (multilayer, as opposed to more expensive single layer) NAND gets down around 20nm.

So expect to keep hearing about higher-capacity-for-not-much-more-money HDDs over the next five years, but no breakthrough that catches up with 30-50% data growth. Expect to see more NAND in devices. Expect to see NAND grow its share of enterprise or cloud storage, becoming a much more perceptible slice of a rapidly growing pie. But don't expect to see NAND or any other technology grow in a way that accommodates the expected need for zettabytes the way only HDD will be able to. Seagate has done as good a job as anyone explaining the economics of this.

One last thought about substitution risk, the likelihood of some other technology besides NAND supplanting HDD as the primary medium for mass data storage: Eventually, it just might happen, but it won't be in the time period we care about. This is because even if something that really could store twice the T-bytes at half the cost showed up in a lab today, it would probably be five years at least before it could even begin to matter at retail. Think of the amount of testing that would have to be done before customers would consider installing it in hundreds of millions of devices, stamped with their brand and backed by their warranty. Consider time and process technology involved in getting from lab volumes to flawless mass volumes, and the very building of plant and equipment to do that production. This is not a risk that sneaks up on us, the sort of secret that could be kept until it is ready to go to market.

All Those HDDs will keep Spinning (but only for so long). One element of unit demand for HDDs that I don't find anything written on revolves around there being a very large installed base of devices in service, a certain number of which reach the end of their service life every day. The implications of this will grow more apparent not only as that installed base (and hence the "replacement", as opposed to "growth in the need", element of demand) continues to burgeon but as the cloud becomes the predominant repository of everybody's data. This replacement aspect has been an evolving muddle, probably too much so to be worth trying to understand, for as long as the client device was the dominant storage nexus. It will likely be a lot less so at the end of the decade.

Historically, HDDs got replaced when they failed, when the device owner could upgrade to something more capacious (and was tired of de-fragging every week) or when it was time to trade up to a new device (which in the heyday of the WinTel era was about every two years). Since those early days, few things have evolved as dramatically as HDD technology in terms of value rendered so much so that it is well nigh impossible to know in a statistically meaningful way what impels owners to retire an HDD. I suspect that at the device level, it is more than ever the retirement of the device (PC, etc.) itself. This has become problematic for HDD makers as PCs have for most households evolved into appliances, which we all know only get replaced when they quit working or when the household has discretionary income and doesn't have a more compelling use for that income. (Point of View: This explains a big part of weak PC sales lately. Either households lack discretionary income, or there is a more compelling use for what they do have. Households that don't yet have a tablet or smart phone will likely buy that before replacing that long-in-the-tooth but still functioning appliance that is their four year old lap top. It also matters, IMO, that by altering the answer to the "But will we get value back for what we spend?" part of the home improvement question, the recovery in residential real estate has made a hugely pent up demand for home improvement a whole lot more compelling, at the expense of less compelling uses of not exactly abundant discretionary income.)

The replacement element of HDD demand will probably become a lot more understandable as the preponderance of data storage shifts to the cloud. This is because replacement in the enterprise setting is mainly a function of the economics of equipment providers' service contracts. As was explained to me by someone who has worked in enterprise storage technology about as long as I have worked at investing, the disk arrays that make up enterprise storage come with a service contract of a specific, several year duration, covering the prompt replacement of any HDDs that fail. Apparently, the terms for extending such a service contract are such that it makes sense to just replace the disk array. The array provider might recycle the equipment to a less demanding venue (a data center for an African airline still operating B-727s?) but only so much. What this suggests is that as the cloud grows its share of a rapidly growing "data storage in service", a higher portion of HDD demand will come from relatively predictable equipment replacement cycles.

Of course, this shift to the cloud is where "more heads & disks" comes into play and does damage to our habit of thinking in terms of an HDD TAM (total available market). The "average heads per disk" metric also appears headed to share the fate of the man who drowned in a river he heard was on average only six inches deep. This is because the market has begun a bifurcation towards a lot more client devices with no or only one or two heads and cloud storage drives with as many as fourteen heads. As this trend continues (and even accelerates if thin, one-disk solutions for client devices proliferate) these metrics will become less useful. For instance (these numbers are illustrative, not factual) consider a 500MM unit TAM not unlike the recent past, wherein 80% of devices (client) averaged 3 heads per drive and 20% (enterprise) use an average of six. That would require 1.8B heads (1200MM + 600MM). But if the market evolved to just 70% of the drives needing an average of only 2 heads and 30% needing an average of 10, and the TAM remained unchanged at 500MM (again, just an illustration), the number of heads (as so suspension assemblies) would grow by 22%, to 2.2B (700MM + 1500MM). No wonder we are hearing more about "more heads & disks", at least for the years immediately ahead of us.

Wireless Internet Will Inexorably Grow More Robust (but the problem of latency isn't going away soon.) One scenario HDD makers would have a hard time adjusting to would be if infrastructure that supports mobile computing got so "fat" and ubiquitous that everyone could rely on what amount to portable dumb terminals, with all the computing and data storage happening somewhere else. This might work out for George Jetson, but for investors considering the next three to five years, the build-out will only foster rather than hinder demand for portable storage. Eventually, there might be one great big hot spot. In the time frame we are investing for though, except for those individuals who never leave their apartments (or, perhaps, those peculiar folks who brag of never leaving the Island Manhattan), there will still be the annoyance of dropped signals and pokey hot spots (even in Manhattan, no doubt!), not to mention the numerous and quite often numinous places we like to go that are still not covered by 4G LTE. If one wants to enjoy the superior performance of a SSD, this problem can be ameliorated by careful management of what and how much data is on the drive. A better solution would be to own a wireless portable HDD. It likely costs about the same as a year's worth of cloud storage but lasts for 4-5 years, and a really mobile user will never have to wonder if it will "be there" for them. An even better solution is probably going to be having a SSHD (solid state hybrid drive) in the device. I am inclined to believe that these are going to be very successful because they will offer ample "away from home" storage at a much lower cost than a barely adequate SSD, giving up only an imperceptible amount of performance (like 12 seconds to boot instead of 10? or a few less minutes on an all-day battery?) So while it is entirely plausible that in five years most users will think of the cloud as their primary storage nexus (if they think of it at all), there will also be a vast expansion of the notion of caching data, and if it is by the Gigabyte or Terabyte, it will be done on spinning disks.

Human Nature: Barely Mutable, if at all (Part 1) Television has gotten a lot of mileage out of hoarders, those people who can't seem to throw anything away. But who among us is anything but a hoarder when it comes to digital content? But there is a very good reason that even those who are fanatically clutter-free in the physical realm hardly ever get around to de-cluttering their hard drives. Storage is just too cheap, and the stuff most of us have on our HDDs is an admixture of some of our most precious possessions (too precious to risk mass deleting) with more worthless trash than we want to even think about wading through. Unless one can zero on really big chunks, like HD movies, who among us can be bothered to take the time and effort to delete those files that are only taking up space. Just to free up a hundred or so Gb (that cost us how much)? And the same problem will play out with cloud providers. Perhaps they will come up with ways to make it real easy for customers to let go of at least some of their ten year old kitty pictures, etc., but it will have to be real quick and real easy. (Who knows, somebody could write an app that measures "precious" and tags files that don't register as such.) A diminution of pack rat like behavior is not the kind of thing that is going to put a meaningful dent in the zetta bytes of storage likely to be needed for the foreseeable future.

Human Nature: Barely Mutable, if at all (Part 2) A final factor that makes me think that the world of five years from now will see Hard Drives Everywhere derives from our tendency to believe that our most precious possessions will be safer if we can hold them in our hands and hide them in a place that only we know about. As opposed to handing them off to somebody we have never met to look after them wherever they choose to put them. I get that objectively speaking, data on a cloud providers drives (they do use disk space much more efficiently than any of us possibly could, but they replicate it all over the place) is safer (from loss) and probably more secure (privacy wise) than on a drive in my house, but only in my head. My heart is still not so sure. Some people in some places might already be trusting enough, or simply haven't accumulated much in the way of digital content, to completely walk away from a physical connection with it, but I suspect they are few and far between. Their ranks might very well grow over the balance of the decade, but not likely by much, and especially not in those parts of the world where trust, in both infrastructure (to not fail) and authority (to not abuse) is not as abundant as it perhaps is in the USA.

Putting all this together, the case for a whole lot of spinning disks five years from now is solid. The cloud will continue to be built out, and all kinds of data capture/data share devices (the PC/tablet/phone distinction continuing to blur, and more things we aren't even thinking about showing up), some with HDDs, many without, will proliferate. With respect to HDD use there will be outlier groups: some whose only interaction with an HDD will be at their cloud storage service through their smart phone. There will be others who will eschew the cloud and get by with just one HDD in their device, although their on-line activity will entail storage demand on many other HDDs. But I suspect that the vast preponderance of users, at least among the most prosperous billion of us, will have a personal cloud connected with a cloud service. They will pay the equivalent of a not particularly lavish dinner for two, or an hour of an auto mechanic's time, for a year's worth of cloud service. They will also pay a similar amount for the HDD at the heart of their family's little cloud. And if in the SSHD we get "the best of both worlds (the high capacity and low cost of HDD, the speed of SSD) there will likely be a whole lot more HDDs per household than can be estimated from data in hand today. A couple of billion consumers, and the businesses and governments that serve them, keeping all of their digital content redundantly safe, secure and instantly available, at a cost that gets easily lost in the shuffle within our household budgets.

If this thesis holds up reasonably well, then the years just ahead should see substantially higher valuations for the HDD makers and suppliers like HTCH. Of course, they will also all but certainly see macro and geopolitical "bumps in the road" that overwhelm the patience and fortitude of all but the most stoic (or somnolent) investors. Despite their strong recent performance, it is probably not too late to establish a position in HTCH, STX and WDC, based on their three to five year appreciation potential. But such a purchase should not be so large that one could not live through a 25% (with the visceral prospect of more) detour along the way, or better yet summon the nerve to buy more once that happened rather than selling out. In any case, until such a moment arrives, a bet against HDD technology remaining the primary answer to a huge and growing problem for years to come is going to be an awfully tough bet to live with.

Disclosure: I am long HTCH, STX, WDC. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.