Remember when you used to buy a software package and you had to check that annoying little "performance requirements" section to make sure your computer was fast enough? Well, mainstream consumer PC's powered by Intel (INTC) and to a lesser extent AMD (AMD) have reached such high performance levels that we have practically witnessed the end of performance requirements. In my last article, I explained how this performance gap, the gap between the capability of the hardware and the needs of the software, has been one of the key adversaries against Intel's price growth and valuation over the past 5 years. To put it simply, no matter how good Intel's hardware is, consumers won't buy it if they don't have any software and/or services that create a demand for it. I emphasize that this is only an issue with the consumer software market -- the business and academic sectors for example are gobbling up all the hardware performance that they can get. (To be more quantitative, you can see this in Intel's separate segment earnings, down 1.5% in the PC client group in Q4 FY12, but up 6.6% in the Data Center group.) This article isn't meant to focus solely on Intel, but the chip companies are an inseparable element of the discussion.
In this article I take the thesis that the performance gap will be closed -- eventually. The growth in computing performance offers a significant opportunity for new and innovative software applications. However, like I stressed in my last article, the nature of technology development is difficult, and it is not clear how long it will take for this innovation to occur. But the more interesting question from the point of view of an investor is how can I play the situation right now, and moreover, who could I buy? For the rest of the article, I will elaborate on these questions by discussing the potential ways in which the performance gap will be filled, and what companies look like they could benefit from it. My conclusions at the end of the article are that we are still at a somewhat speculative stage in the technology development and therefore it may be too early to make specific buys on this play. However, I do recommend for computer software and services investors, as well as hardware or semiconductors investors, to take note of developments in this space.
First of all, let's discuss the details of how the performance gap has come about, so that we know exactly what we're talking about. I think it's fair to say that this has only become an issue in the last 5 years. In this time period the computer hardware has decisively moved beyond the software, due to a number of technology advances. These have been the process shrinks, going from 45 nanometer (NM) in 2007 (see wiki) to 22 nm in 2012, and 14 nm in 2014 from Intel. There have also been the architectural improvements, including improvements to branch prediction and integration of more and more processing elements on die, such as GPU cores, memory controllers, PCI Express, and the development of new, specialized instruction set extensions. The integration of better and better graphics on-die has been a decisive factor, spearheaded by AMD. With the introduction of next generation Intel Haswell processors sometime in June, advanced graphics processing will become ubiquitous across all PC's, from budget to performance levels.
All these improvements are contrasted against the static 5 year background in mainstream consumer software and services, which haven't moved much beyond the standard OS (e.g. Windows, Linux, or OS X), the standard web browser (e.g. Explorer, Chrome, Firefox), and the basic online services (e.g. Google, Yahoo, Bing, Wikipedia) and offline software (e.g. Microsoft Office) which dominate the computer and internet activity of mainstream users. The linearity in those services is remarkable to me -- in my opinion, they've hardly changed at all in the past 5 years. There are a few notable exceptions like Google's always-connected Chromebook OS, the touch-interactive mobile operating systems of iOS and Android introduced by Apple (AAPL) and Google (GOOG), and a few of the software/services functionalities like autocomplete, which weren't around 5 years ago. In any case, none of these advances have managed to capitalize on the increased performance available to the consumer computer. For example, Windows runs faster on current hardware than it did before, but it's still doing fundamentally the same thing, except maybe now with a touchscreen. Like I said in my last article, I think you have to be pretty critical towards Microsoft (MSFT) about this.
Now let's consider what kind of next generation software and services could stand to make use of the growing hardware resources. Firstly, there is huge room for improvement in operating systems and basic user interface and interaction. IBM (IBM) recently announced how they think the five senses will become increasingly integrated with future computers. Intel has released the Perceptual Computing development kit for applications that have more intuitive user interface interaction. Jack Purcher recently reported how Apple released patents on a 3-D version of OS X, and Gary Marshall discussed the pros and cons of 3-D operating systems in 2011. All of these developments reflect how the current state of the user interface leaves much to be improved. We interact with a 3-D world, yet the computer is still more or less a simple 2-D screen, basically a glorified typewriter. We are now approaching the computer performance capabilities to do things like render 3-D computing environments on the fly, and interact with them through natural gestures measured by motion tracking cameras.
When I first heard about the moves towards gesture interaction, which Intel touted at CES 2013, I thought they were somewhat gimmicky. I didn't really see the point. But now I realize that Intel was just recognizing the same point as this article, that the growing performance capability of the hardware can and should enable innovative new computer interface modalities. The original 2-D GUI introduced by Xerox was never meant to be the final GUI, though it may seem that way now, after 30+ years of continuing implementation. For the most part the Xerox GUI was just an engineering optimization between the need for the user to interact with the software in an easy and efficient manner, against the rudimentary performance capabilities available at the time. As Perlmutter says (and Perlmutter by the way is probably my top pick for the next Intel CEO) we are still only in the infant stages of that optimization process.
The other area where I think the performance gap will be bridged is in search engine technology, which is becoming closely related to basic OS functionality, as represented by Google's Chromebook systems. As said by Christopher Mims, we're already living in our web browsers most of the time. However, search engine technology including user interface technology is still in its infancy, and as of now it doesn't take advantage of the computing resources of the user's system. In the comments of my last article, I wondered why Google hadn't released a Chromebook on Intel hardware, since it could surely figure out some way to use the extra resources. The user NakedOptions reported how Google actually did have such a system in the works. A few days later, the system was formally announced over at Google. They didn't say quite how you would use all those onboard resources, except to say that it would load remarkably fast. But I think the title of their announcement was extremely revealing -- "The Chromebook Pixel, for what's next" (emphasis added).
Clearly, what's next is a search engine that acts as much as high performance software as a traditional web service. Search engines are about much more than just textual interfaces, the way they seem to us now -- they are the fundamental gateway into knowledge and information available on the web. With advances in machine learning driving growth in the "analytics" business, the natural next point is for advanced machine learning technology to enter into web search. This machine learning should be employed both to help understand the available data, but also to visualize it and enable interaction with it through 3-D vision and gesture. Furthermore, it needs to move further into unstructured data, with the kind of open-ended technologies such as Numenta's Grok. These kinds of tasks will undoubtedly be able to benefit from higher performance consumer hardware.
Lastly, let's discuss who exactly we could buy as investors to try to get in on these opportunities early. The landscape is dominated by big companies like GOOG, MSFT, and YHOO in search, and in operating systems, the companies are very similar, basically MSFT, GOOG, and AAPL. IBM has a large influence on software services, but I'm not sure if they'd want to break into the consumer market, after building such a focus in the business services industry. In my last article, I discussed some reservations I hold against MSFT, and so I'm not sure I'd buy into them on this play, although they are clearly on the front lines. GOOG is doing innovative work here and also soon releasing the Glass product, which could be a big part of the interface innovations that boost the demand for higher performance hardware.
There are a few select other companies that could be well positioned for these changes. These are the graphics software companies like Autodesk (ADSK), and the specialized software companies like Wolfram Research and the Mathworks, both private. The reason why these companies could play a part is because both graphics and scientific computing will inevitably be involved in the transition into future interfaces and modalities, either in terms of driving the demands, or driving the software tools for the innovations. Otherwise, I've done a lot of stock screening but haven't found very many smaller software or services companies pushing these kinds of technologies.
In conclusion, I think there is a clearly growing gap in the performance capabilities of consumer computer hardware, and the software and services that are able to use them. I believe that this gap will eventually be filled, and I discuss how advances in user interfaces, operating systems, and search engines look positioned to take advantage of them. Although there are a number of big companies that one could speculatively buy for advances in these areas, with the frontrunner being GOOG, I think we still haven't seen the products to justify a large buy-in on this play. If you've seen some companies I haven't mentioned or heard about, please let me know in the comments.
Thanks very much for reading and I welcome any comments, questions, agreements and/or disagreements.