PCs: Raging at the Dying of the Light
July 28, 2010
PCs: Raging at the Dying of the Light
- Now in its 30th year, the PC market is enjoying an upgrade-cycle fueled rebound from a recession induced dip. However, an alternative approach, by which mobile non-PC devices, like smartphones and tablets, access apps running on large centralized “cloud” servers, threatens the x86/Windows architecture in the same way that the PC threatened the minis and mainframes that came before it. We believe that the demand for desktop PCs will wane, as virtualization and cloud applications allow many users and organizations to skip future upgrades and downscale configurations. Mobile PCs, the primary driver of industry growth, will feel pressure from the cloud-enhanced capability of non-PC tablets and smartphones. PC-based servers will suffer from ASP degradation with an on-going shift to lower cost rack-mounted blade servers. As such, we expect a comprehensive and irreversible shift away from PC architecture will be evident within 5 years
- 2009 was a tough year for PCs, with global recession punctuating a relatively tepid three year reception for Microsoft’s Vista PC operating system upgrade. While the 4Q09 intro of Windows 7 prompted anxiety, 1H10 results along the PC value chain suggest that some of the aging installed base of XP desktops is being replaced, yielding a flat rather than declining demand curve. This is augmented by rapid growth in notebook and netbook PCs, particularly in the consumer market. At the same time, the rapid growth of dense, rack-mounted blade servers for private virtualized systems and cloud providers is offsetting a decline in traditional enclosed servers
- The simultaneous rise of portable computing platforms and Internet resident cloud applications benefit the PC value chain in the short run, but will sap demand for powerful and interoperable PC-based user platforms as alternatives gain momentum. Smartphones and non-PC tablet computers have great appeal from their superior cost, portability, ease of use and communications capability, while cloud-based applications on powerful shared servers allow users to work effectively with the more limited processing and storage capability afforded by mobile platforms. Enterprise desktop PCs will see pressure from inexpensive “net-top” computers, which can run Windows applications off of virtual and cloud servers whether or not the “net tops” have PC architecture. This paradigm change would hurt the entire PC value chain, creating opportunities for non-PC architectures
- Rack-mounted blade servers are less than 15% of the total server market, but are growing rapidly, particularly relative to traditional enclosed tower servers, which can be installed on racks, but with far less density and at much higher costs than blades. We expect equilibrium to tip decisively toward large, dense cloud hosting installations and virtualized enterprise data centers, favoring an even sharper shift toward the lower cost blades and an increase in the bargaining power of customers. This shift would affect systems vendors far more than semiconductor or software suppliers
- We do not expect the decline of the PC to be immediate or abrupt. Strong momentum for notebook and netbook platforms has yet to be tempered by either tablets, demand for which was only recently kickstarted with the iPad, or Smartphones, which have relative input and display shortcomings vs. PCs that are yet to be addressed by vendors. Moreover, technology demand has significant inertia – the IBM 360 mainframe is approaching its 50th anniversary and still has annual sales of more than $10B. However, we do believe that the market and technology trends will result in a life-cycle inflection point for total PC-architecture demand within 5 years and that the coming erosion of the category will become apparent to investors even sooner
- A shift away from PC architecture at the device level would have its greatest impact on PC systems vendors, the largest of which are HP, Dell, Lenovo and Acer, PC dependent semiconductor vendors, such as Intel, TI, AMD and NVIDIA, and PC disk drive suppliers, such as Seagate and Western Digital. The future for companies in these categories will depend on their ability to offset future PC market deterioration with success in alternative devices and servers. PC software publishers, such as Microsoft, Symantec, Adobe, and others, have largely already begun to see a shift toward server based products, often running on virtual machines or clouds, away from higher revenue, but lower margin PC applications. Here we expect the opportunity to be strong, albeit with a wide opening for competitors to enter previously concentrated market
A Star is Born
The IBM PC was introduced to the world on August 12, 1981. The initial model, the 5150, ran the Microsoft developed PC-DOS operating system on an Intel 8088 4.77MHz microprocessor with two floppy disk drives and a maximum of 256KB of memory, selling for $1,565. It had an open architecture, inviting any and all manufacturers to produce compatible products without need of license. While IBM used a variety of proprietary elements in its hardware and software to try to hold off the makers of “clones”, ingenuity would not be denied and companies like Dell, Compaq (now part of HP), and Acer were built on the back of IBM’s initiative, finally gaining prominence in the ‘90’s.
Of course the PC also nourished the growth of numerous suppliers to the emerging platform. Microsoft and Intel are obvious, but companies like AMD, Micron, Seagate, Western Digital, Lotus (now part of IBM), Symantec and Adobe all owed their lives to the PC. To press it further, without the standardized PC platform, there would not have been a market for the consumer internet – email, on-line shopping, search, social networking, etc.. Indeed, Time Magazine’s widely ridiculed choice of “the computer” as its person of the year in 1982 was, in retrospect, prescient.
The Comeback Special
2009 was a tough year for nearly everyone. PCs were no exception. Global unit production dropped 5% YoY (Exhibit 1), with revenues down further on ASP declines (Exhibit 2). The biggest part of the decline was the recession, but the economic downturn also coincided with what had been a weak market response to the 2006 introduction of Microsoft’s Vista release of its Windows OS. Ordinarily, a major Windows revision had stimulated major accelerations in PC demand as users and enterprises upgraded their systems to meet the requirements for the new software. Each new version needed faster processors, more memory, bigger storage disks, etc., pulling along new demand along the entire value chain. Vista had been late to market, marked by early user complaints and foot dragging by would-be customers, and capped by the budget sapping recession.
Microsoft’s October 2009 release of its next upgrade, Windows7, was straight into the teeth of the economic storm. While the taste of Vista and the tough environment caused investors and observers to temper their expectations, the first half results from 2010 have been very encouraging. Microsoft and Intel reported very strong growth in their recent 2Q earnings, with the rest of the PC value chain echoing the bullish sentiment. Enterprise uptake of Windows 7 appears to be stimulating demand for more capable machines to replace the large base of aging machines still running the 8 year old XP version of windows, along with a return to growth for portable notebooks bound for both enterprise and consumer markets. Correspondingly, PC related stocks have largely enjoyed a strong rally over the past two weeks.
The New Kids on the Block
Over the near 30 year history of the PC, its head-to-head competitors have made little traction in blunting the momentum of the Windows/x86 architecture. Despite its disproportionate share of media attention, Apple’s non-PC Mac product line accounted for just 3.2% of global unit shipments in Q1 2010 (Exhibit 3). The Unix variant Linux has a rabid IT geek following, but appears on fewer desktops and notebooks than Apple’s Mac X OS.
Gartner projects global PC unit production to roughly double over the next four years, with mobile PC devices growing nearly 25% per year over that time. By the end of that period in 2014, Gartner assumes that home mobile PCs will account for about half of all PCs sold at nearly 240 million annual units. With mobile PCs higher priced and more profitable than their desktop cousins, that would appear to be good news for the PC value chain.
Not so fast. The iPad is leading a wave of non-PC portable device challengers sporting non-Windows operating systems – e.g. iOS, Android, WebOS, etc. – and non-x86 processors – e.g. Snapdragon, OMAP, XScale, etc. – along with their built-in wireless internet connectivity and pixel-dense touch screens. These tablet devices are designed for portability, reliability and ease of use, that the more powerful, but also more complex Windows/x86 architecture has difficulty matching.
On a stand-alone basis, these devices might have difficulty meeting user needs for processing power or storage, but tablets do not stand alone. The rise of “cloud” applications, by which devices access software resident on powerful servers via the Internet and leverage the seemingly limitless processing and storage available there, obviates the need for added firepower on the device. For the average consumer user, internet access, including email, search, e-commerce, streaming media, social networking, basic document processing, games and other apps, is the primary reason for buying a device. These cloud applications may even make increasingly capable smartphones, such as the Apple iPhone and the many Android models, reasonable alternatives to a mobile PC for many users.
Desktop to Nettop
For 30 years the mechanics of the PC market have been driven by Microsoft OS upgrades. Each new version of Windows has been bigger than the last, requiring faster processors, more memory, increased storage, ad nauseum. To exploit the new capabilities offered by each new release, users and enterprises would upgrade their desktops to bigger, faster PC models. Windows NT begat Windows 95, which begat Windows XP, which begat Windows Vista, which brings us to today’s Windows 7 (Exhibit 4).
Enter virtualization. Virtualization is a technology by which software can partition the resources of a large system such that each partition, called “virtual machine”, can operate as though it were a separate computer with dedicated processing, memory, and storage, running its own operating system and applications. By this, enterprises can take advantages of substantial economies of scale in running applications on large, data center servers, rather than on individual PCs in departments and on desktops. Because the applications use the resources of the centralized server, the need for processing power, memory or storage at the desktop are greatly reduced. Thus, as enterprises virtualize, they can upgrade their OS and applications on the server without requiring a matching upgrade to the device. The same is true for applications that move outside the enterprise to “virtual machines” operated by 3rd party “cloud” hosts on the internet. Users do not have to have upgraded devices to exploit new operating system functionality and applications. Moreover, it would allow non-PC devices to run PC applications in a virtual PC on a server.
As noted above, these non-PC devices may well be mobile platforms like Smartphones and tablets. They could also be cheap desktop devices optimized for accessing applications running on virtual machines on big data center servers. Just as mobile carriers have marketed inexpensive mobile internet optimized note-book PC like devices as “net books”, IT marketers have coined the term “Net-tops” for these inexpensive PC replacements. In the long run, we believe this is a major threat to the future of the desktop PC, and while recent sales data confirms that a Windows 7 upgrade is happening at the desktop, we fear that it may be happening for the last time.
Are You Being Served?
The market for servers, overwhelmingly based on x86 architecture, is also seeing a substantial recovery from a sharp drop in 2009. Here, the trend toward virtualization and clouds is a favorable long term factor. However, there are a few caveats to consider.
First, Gartner divides the server market into three major segments. Tower servers, self-contained stand-alone computers, have historically been the largest category, but have now entered long term decline (Exhibit 5). Rack optimized servers are still self-contained, but have been designed to be mounted in standard data-center racks. This category is growing at a modest pace. Blade servers are the electronics of a server installed on a compact board operable only in the context of a dense rack installation with heat dissipation, grounding and other protective functions integrated into the rack. This category is the smallest but the fastest growing part of the server market. With the strong trend toward virtualization and cloud computing, we anticipate that the shift toward compact, efficient and lower cost blade servers will accelerate, pulling down system ASPs in the process.
Second, with the rise of very large cloud hosts, such as Google, Amazon and Microsoft, the bargaining power of those buyers will likely put more pressure on server vendors than the customers that they replace. The same is probably true of virtualization, which could help IT departments gain better control of computer use that has been outside of their control.
Finally, the rise of blade servers appears to be attracting new entrants to the market (Exhibit 6). Networking leader Cisco has already entered the blade server market with its “Unified Computing System”, ready to parlay its deep data center expertise and partnership with virtualization leader VMWare into a viable competitive position.
The Chain Reaction
Our base scenario projects non-PC architectures to take a meaningful share of the mobile and desktop computing market by 2014 and as a result, for the 30 year growth story of the Windows/x86 PC to reach an inflection point shortly thereafter. This is not to portend an abrupt and precipitous end, but rather a slow, and permanent decline in the device platform. The rise of virtualization and cloud computing, which is greatly contributing to the deteriorating position of the PC architecture at the device level, will serve to keep it alive in the data center in the form of increasingly large and dense deployments of blade servers. What does this mean for participants in the PC value chain?
At the systems level, the PC market is only modestly concentrated, with HP, Dell, Lenovo and Acer collectively holding just under 44% share. While PC manufacturers have opportunity to participate in the rising market for non-PC mobile platforms, they will face formidable competitors also moving into the space from the Smartphone market without the scale advantages – volume purchasing, manufacturing efficiency, distribution – that they enjoy in their PC businesses. Of the top PC manufacturers, Acer, Lenovo, Asus and Dell all derive more than half of their revenue from desktop and notebook PCs (Exhibit 7). Systems companies with greater exposure to the server market, such as IBM and, to a lesser extent, HP, would be less sensitive to a negative outcome for PC architecture in the device market.
Semiconductor vendors also have strong stakes in the PC market (Exhibit 8). Intel, AMD, and NVIDIA each generate at least 70% of their sales from PC processors, with an additional 15% or more derived from the more promising server market. Chip makers specializing in other aspects of the PC – e.g. memory, power, I/O, etc. – are generally far less dependent on the PC market, as these functions are broadly applicable to other large electronics end markets.
Software is somewhat less dependent on PC architecture, as applications can be run in clouds or virtualized servers and can be ported to alternative devices. Microsoft, assumed to be nearly synonymous with PC architecture, generates only 25% of its revenue from PC device software (Exhibit 9). Security software companies, such as Symantec and McAfee, have traditionally had high exposure to the PC market, but would appear to be well positioned to address the security concerns that have been raised around the cloud computing model.
Hard disk drive vendors, such as Seagate and Western Digital, have exposure to PCs approaching 80% of their revenues. Moreover, new mobile device types like tablets and smartphones have excluded disk drives in favor of solid state DRAM storage. At the server level, mass storage requires very different products than are used in desk top and notebook PCs, offering little opportunity for PC drive specialists to offset the possible deterioration of their core market. Similarly, storage vendors to the server market, such as EMC, HP, Oracle and Hitachi, have little exposure to the desktop or notebook computer markets.
Contract manufacturers have had a difficult time of it over the past 18 months, although PCs have been only a part of the problem (Exhibit 10). While many manufacturers participate in the PC market, their exposure is skewed toward the more attractive mobile platforms. Moreover, most of these manufacturers also have strong exposure to the non-PC platforms which threaten the PC, offsetting their risks.