Make sure you understand IT fundamentals before accepting analyst assertions about IT Stocks.
I have been around Information Technology all my life and have worked in the industry for about 4 decades, recently retiring from IBM (IBM) as an IBM Distinguished Engineer. I only recently became an active investor in preparation for my retirement. Part of becoming active was subscribing to Seeking Alpha. Because of my background, I began following IT Stocks. I was inspired to write this because so many analyst and commentator discussions of tech stocks make assertions about technology, which are simply not true. These assertions come in the form of four fundamental misunderstandings about IT. I am writing this so that you can examine analyst assertions about IT critically and be better able to assess stock analyses of tech stocks.
Here are the four assertions that reflect fundamental misunderstanding of Information Technology:
- The IT Industry is on the road to commoditized hardware. "Modern Servers" do better than "Legacy Servers" as time goes on.
- A common metric is useful to simply the complexity of dealing with multiple types of servers.
- Servers that don't do well on standard metrics will not survive because along with commoditized hardware come standard performance requirements.
- Since "the cloud" is the wave of the future and the future is about commoditized hardware, cloud implementations will be based on commodity parts.
There are plenty of people in the IT industry that would accept these assertions. In most cases they either have a vested interest in maintaining the "common wisdom" or they are generalizing from experience within a narrow sector of the IT market or of IT solutions. Unfortunately it takes more space than available here to the lay out the technical support for my point of view. Suffice to say that role of servers and the relationship of server types in IT solutions is not as simple as most people would like it to be.
The IT industry is not about commoditized hardware but is actually about integration and customization of hardware infrastructures with software and services for IT solutions. The variety of IT platforms available today is far wider and more diverse than it was 50 years ago when "a computer" was certainly not a commodity. The customization and integration that goes into the creation of server infrastructures today is far from a commodity business. While Google (GOOG) does build infrastructure from a variety of commodity parts, most server infrastructures use servers with purpose built server chips. Many server vendors use Intel (INTC)Server Chips. Among them are HP (HPQ), LENOVO (OTCPK:LNVGY), DELL, and FUJITSU (OTCPK:FJTSY). Even the relatively high volume of Intel server chips is dwarfed by PC chip volumes, which in turn are dwarfed by mobile and embedded processor volumes. Thus modern IT solutions contain critical non commodity elements. The economics of servers still favors purpose built parts, not commodity parts. These parts are more capable than the central processor units of the 20th century, but they still are not standard commodity parts by tech industry standards. All common servers use legacy processor architectures from the second half of the last century. The Intel server is not more "modern" than the mainframe. Its processor architecture appeared only a decade after the mainframe in the mid 20th century. Both architectures support modern and legacy operating systems and application software. Note that the IBM's Intel chip based System X business is on the block. (Seeking Alpha) It is the most commoditized server hardware in IBM's portfolio. Would this be so if the IT industry was moving in the direction of commoditized hardware?
The use of or search for a common metric is a fools errand. The diversity of IT solution designs available today could not be achieved without each of the processor types currently in play. No matter which metric we pick some processor design "styles" will be better than the others and some worse. The problem is that even though the chosen metric quantifies a relationship between servers types, it is unlikely that the relationship is maintained when the servers are actually put to work. It can be shown that it takes at least 2 metrics to describe a match to a workload and that production usage patterns bear little resemblance to any common benchmark. (IBM Enterprise System Journal Article or "ALternative Metrics for Sever RFPs")
It can also be shown that each of the major server types is good in a niche, notch or range and poor in the rest of the spectrum of IT workloads. This applies to Oracle (ORCL) /Fujitsu SPARC, Intel Xeon, IBM Power and IBM Z. No machine is universally superior to or inferior to any other. Thus there is no single metric by which we can create a valid ordered list by performance or capacity. This is especially true when we had cost to the comparison.
Finally, cloud computing is not a new thing. It is more an exploitation of enabling technologies that allow implementation of an old idea called the "Service Bureau", adding "self service" fulfillment, deployment and administration functions. Businesses have long provided specialty services to those who could not otherwise afford computers or IT infrastructures. ADP(ADP) started its payroll service business this way. It would not be surprising to see cloud like self service user interfaces and service charges emerging there. However, the use of commodity processors is not a prerequisite for this type of evolution. "Modern Commodity Processors" are not required by cloud economics.
You should be skeptical or at least careful when analysts support their analyses of tech stocks on an assertion that reflects one of the four misunderstandings about IT technology. For example assertions that a vendor "failed to move away from legacy" fall into this category. These assertions have been applied to IBM, Intel, CISCO and Oracle in various forms all based on the notion that there is a new commodity in town. This argument is a fairly standard way to market a new entry to the market. Consider Paycom's advertising that "The price of technology goes down over time" and the rebuttal responses that can found by googling the quote(example). Keep in mind that each wave of "new technology" has not replaced existing server types, but rather, added to the diversity of technology from which we build server infrastructures for IT solutions. This diversity provides multiple points of optimization and moves the business away from a world of "commoditized hardware".
There is a similar story behind network and storage hardware controllers, which are special purpose servers. The price of per bit of storage doe go down, but the bits required per solution rises. The same is true for bandwidth of network devices. However the evolution of both is toward making them "smart" which brings us back to server technology.
I hope this helps. I do not pretend to offer specific advice on any of the stocks mentioned, only to provide some background to help you examine the assertions made by stock analysts. If you want to understand how the various server types are positioned across the spectrum of workloads, or understand the technical underpinnings to this, contact me at lc-ns.com. You can rest assured that there is solid system science and IT Engineering behind what I am saying and to be sure, you should be willing to challenge, question or be skeptical of assertions based on misunderstandings of IT fundamentals.
Additional disclosure: I am retired from IBM but have no current business relationship with them or any other company mentioned. I still work in the industry and may enter into a relationship with IBM or any of the others in the future. In this paper I am not claiming to represents IBM in any way, nor am I reflecting any input from my former colleagues.