In a turn of irony Microsoft is starting to feel the pain of Moore’s Law that Sun has felt for years. A pain that largely led to the departure of Scott McNealy and to Jonathan Schwartz efforts to open up the company. The Guardian published an outstanding article on this future fate of Windows titled “Why falling Flash prices threaten Microsoft.” In the article they describe “the price differential between the basic Eee PC running GNU/Linux and one running Windows XP is now significant as a proportion of the total cost. One of the main suppliers of the Asus Eee PC, RM, sells the GNU/Linux version with 4Gb of storage and 512Mb of RAM for £199. The cheapest machine running Windows XP costs £259, 30% more, not least because Microsoft’s operating system needs more storage and memory - 8GB and 1GB respectively. It is that difference, far more than any cost of licensing Windows, which means that Linux-based machines can remain consistently cheaper.” While the Guardian points out that Linux is more memory efficient it misses that Linux’s rate of change is also much better because so many people are worried about making it more so (for mobile phones and elsewhere), and anyone can contribute. By contrast, Microsoft has two totally different code bases for Vista and Windows Mobile and so laptops don’t get any efficiencies from the work being done on mobile phones. What is additionally striking is that the OS is now likely the single most expensive thing in a low-end Windows laptop, even though the cost of goods sold (COGS) of the OS is $0. All this adds up to fundamental problems for Microsoft in the world of the affordable PC.
Microsoft can always try to reverse this trend by reducing their prices. Coincidently this is something they just announced. The problem for Microsoft however is not just a matter of lowering costs it is a matter of who profits from a PC. The numbers are clear. Microsoft has profit margins in the 30% range. Intel, the guys who sell the chip are in the 15% range while companies like Dell, Asus, Acer, and others are down in the 5% range. Something is fundamentally wrong. One company is making disproportionately higher margins in this game. IBM understood this so well they exited the PC manufacturing business entirely selling it to Lenovo. Gartner Group even went so far as to predict the “top ten companies in the computer manufacturing industry will exit the market by 2007.” What Gartner did not predict was the role that Linux can play in turning this around. Fundamentally it makes sense for companies like Asus to bring some of that margin home by creating a PC with a custom Linux operating system which is exactly what they did with the Eee PC. Other manufacturers are following including Lenovo, Everex, Shuttle, Dell, HP and others.
Could we be seeing history repeat itself? It wasn’t that long ago that we saw similar fundamentals killing Sun Microsystems in the enterprise. Back then the trend was comparable, Goldman Sachs reported in 2003 “companies may begin to use Linux on higher performance/ lower-cost Intel-based hardware (and, to a lesser extent, on existing mainframe hardware) as a platform for server consolidation.” What was happening then is starting to happen now. Back then the expensive operating system was Solaris and the expensive machines were Sparc boxes. Today the expensive operating system is Windows and the expensive hardware are high priced laptops out of reach of many consumers. Ironically, Intel which was a huge factor in competing against Solaris Sparc, is actually contributing to the pressure on the Windows OS with new low price chips like Atom that support Linux. The brilliance of this move is that Intel is covering their bases in both the low end with Atom and in the high end with their dual core chips.
We’ve seen this story before and we know how it ends. Perhaps Microsoft can learn a lesson from history and Sun Microsystems. The lesson is not to wait. If Sun had open sourced Solaris and enabled its adoption on other architectures sooner it may have prevented the rise of Linux. Unfortunately for Microsoft it may be too little too late and the reality is that monopolies can’t last forever.
Popularity: 34% [?]
5 Comments so far
Good article but I have some observations of my own on this topic, I was just thinking about this.
The hardware to run GNU/Linux “as people want” is much higher than it used to be, although not the Kernel or the core utilities problem it’s the stuff like KDE, GNOME, etc that make it all friendly and glossy that make it require greater hardware (rotating 3D translucent desktops anyone?)
And people want these things, they want them enough that distros like Ubuntu which are supposed to have a very wide market include fairly heavy apps.
Interestingly even in environments where it’s not required (bland offices, schools, call centers) people still don’t buy the most basic hardware. Many office workers could easily do their e-mail, spreadsheets and document processing with a much much slower much much less expensive machine if they ran older software… let’s face it, Microsoft Office hasn’t gained a lot of really important features in a while.
Even though a number of low cost low power notebooks are available, sales show that people don’t want them. People have been saying that this trend would come to an end for a long time, I remember hearing plenty about this in the mid and late 1990s but it’s not true, there has always been cheaper computers than the main stream buys but they aren’t what the main stream buys. Around 2001 I purchased a $1700 machine and a $6100 machine so I could run the things I wanted to, even though my actual work would have been just as easily done using the then 5 year old $2600 machine I had. The people who can’t afford the mid-market new stuff do something stunning… they buy used stuff! Even companies do this, there is a whole sub-industry based around marketing re-furbished hardware to small businesses, NPOs and cheap enterprises. An industry which doesn’t make manufacturers any money (unless they were the ones who refurbished the machines).
The people who can’t afford the new stuff and don’t want used stuff just run the old stuff they have, the people who can afford new stuff buy it, and even though they don’t actually need it, they still find uses because they have it. Rotating OpenGL desktops are a great example of this… they don’t really improve productivity but they do sell hardware.
Industry is stuck on Office, You can say OpenOffice.org is mostly compatible but it’s not, I use OOo and it doesn’t open everything correctly, doesn’t save everything correctly and certainly doesn’t interoperate with all of the tools designed to work WITH Microsoft Office, so I also have a machine with MS Office and use it when I have to (it’s old). If older versions of Office are not available and new versions require the latest OS and the latest OS / Office combo requires moderate hardware the enterprise will pay for moderate hardware which costs more and thusly has the OS and software as a lower percentage of the cost, that means vendors won’t complain as much and the consumer still gets what they want, interoperable predictable software.
So since almost no one buys new (very) cheap desktop and notebook computers and then loads them up with GNU/Linux and free productivity applications and the re-sale of used ones doesn’t make Microsoft any money why should they open source Windows to compete in a market that is miniature and has slim margins?
A really good proof of this is that you can buy these low end computers, even lower than the ones you’re talking about and run GNU/Linux and people DON’T. You can actually get a reasonably productive complete copy of GNU/Linux to run on many Single Board Computers which with a case and power supply (but not an LCD) cost less than $300 and you can run GNUmeric and Abiword et al on there and do a lot of stuff. But people don’t.
So Microsoft doesn’t have to open source windows to ensure they keep selling their other products or to keep customers.
But that’s just my 2 cents, I run Windows on my gaming computer, GNU/Linux on my servers and sometimes on my MacBook Pro. Most of the time I run OS X so I can have fancy effects and run slick applications. I know full well I could do my job on an old 300Mhz Celeron laptop with 256MB of RAM and run Linux, hell I don’t even need a GUI.
*It’s probably best to read this after my previous comment*
Just some more half/baked musings
I don’t think it’s realistic to think Sun open sourced Solaris to save themselves from the torrent of cheap GNU/Linux machines on commodity hardware. You’ll notice they now don’t now sell cheap Solaris boxes with commodity hardware. They still sell expensive SPARC boxes with Solaris and according to the numbers, they move a fair number of them. I think they did it to reduce development costs while still maintaining a legacy platform, same as Novell maintains NetWare because people actually buy it and run it on new high cost high margin hardware.
Sun is now making a migration to a dual Linux Solaris platform ecosystem where their non OS apps and hardware work with both, probably with the plan of eventually phasing out Solaris.
Sun continues to develop their SPARC line, the latest UltraSPARC T2 machines have 8 cores and each runs 8 threads in parallel at 1.2Ghz. That’s a lot of beef! They use power like they have to take it from a sippy-cup. Compare that to an Intel or AMD based cluster of 20 machines (the approximate number to equal a $26k Netra in certain types of performance) and the Sun is on par in price, much easier to program for, easier to manage, takes up less space, uses less power, costs less to cool and makes Sun more money than if they sold you 20 other boxes. *Swish* good job Sun.
So now Sun can sell you a very powerful machine at a good price and not pay themselves to develop the operating system for it and they don’t lose customers by “killing” Solaris, they keep them and gain new GNU/Linux customers. They also gain a lot of friends, especially in the young idealistic hobby crowd, a market that doesn’t buy their hardware, services or software but as they move up through IT food chain, lose their black-jeans and learn to wear golf shirts they’ll be far more likely to buy a shinny new Sun and the associated services than they would be to buy a Hewlett-Lenovo-Dell-Packard (I assume they’ll have merged by then heh) with Windows 2013.
I make the joke about the mergers because in order to keep competing with the companies like Apple (who’s cultivating the coming consumers with cool rather than idealism) and Sun who control the hardware, software and idealism/desire they will have to keep cutting profits and prices which will cause the hardware to lag behind and widen the gap until they die. Kind of like a spider sucking the juices out of a fly Microsoft will suck the juices out of any vendor that continues to focus heavily on them. On this bleak day Microsoft will Open Source Windows with some ridiculous license and let it die. They will release Office and the rest of their highly popular software for the remaining platforms but it will be too late.
So in the long run, without open sourcing windows in the short term and without everyone buying 20 watt computers with 4GB flash drives, Open Source will win, Microsoft will lose and anyone who fails to cultivate positive mind share will be forgotten.
In 2015 I suspect we’ll see a computer market dominated by 3 KINDS of players, commodity hardware with open software, specialized hardware with mixed software and closed hardware with closed hardware. The commodity/open systems will dominate in the developing markets with the specialized/open dominating in the developed markets and the closed/closed being the most expensive systems appealing to image conscious consumers (in the way Apple/OS X does now).
*I must really have been inspired or something.*
Another interesting thing vendors of these low cost, low power, portable systems based on open software could do that would make them very competitive against closed source solutions (even cheap ones) would be to include some basic high current analog I/O controllers into them. Even just two channels of each would be great.
I think that’d be a real boon for the DIYers and industry in the developing world. It would definitely be a great feature in the developing world because it would foster a lot of innovation in ad-hoc products, be it turning existing an drill-presses into simple computer controlled mill or automating valves on an irrigation system, helping people invent is a great way to actually help them AND make money from them.
The open-ness and customizability of the software with real-world useful hardware would sell a lot of units. It doesn’t matter if your $150 comes from a curious kid in Los Angels or a business man in Bangalore, except there are a lot more of the later and he’s going to sell it to other people.
Actually, the next big driver of cost is Intel - who gouge the PC manufacturers. Perhaps they should follow Sun’s lead, too, an open up their silicon to allow 3rd party manufacturers.
Lovely article.
Can we please get a ‘print article’ button / a print css for your blogs? Thank you.