First of all, given the choice most computer users buy solutions -- not products. Microsoft is a solution company; they provide desktop computing and back-office infrastructure. They started out as a product company serving up MS-DOS, then Windows 3.1, then Windows 3.11 and so on. Over time, they became a solution company with a far-flung portfolio of integrated products that make up solutions. They also did this with very little competition. They have been successful because they are exceptional at integrating their products. The companies in the Linux space that will last are those that do the same, either by organic development of solutions or integrating technologies potentially through mergers and acquisitions.
Today many people are using Linux, but assembling products -- both commercial and open source -- to solve their problems. These early adopters are not the mainstream; vendors are now moving to address those users who see the benefits of Linux as demonstrated by these early adopters and consequently are doing their best to assemble solutions. The companies that are in it for the long haul will either provide both the desktop and the infrastructure to support it, or make strong alliances to have one vendor offer solutions.
My thoughts are that the companies that will be in the Linux space for the long term will be those Linux companies that become solution companies. I think both Red Hat and Novell fit that bill today. Mandrakesoft (www.mandrake.com) is trying to add products to their portfolio today to become a solution company but they have yet to prove themselves in that respect as they have recently released a number of new products. Xandros (www.xandros.com) is doing the same with their release of their Xandros Desktop Management Server because while they have an excellent desktop product, they aren't a solution company. There are some other newcomers like Mepis (www.mepis.org) and Ubuntu (www.ubuntulinux.org) that have great new products as well. They could be acquisition targets if it ever gets to the point where Linux companies try to acquire subscribers like ISP and telephone companies do today.
Which brings me to Mr.Robertson's company, Linspire (www.linspire.com), which is also a product company that in my estimation has the potential to become a solutions company. They have a very excellent product in their Click'N'Run Warehouse that only works with the Linspire desktop distribution. I have said for sometime that one-click install capability is invaluable. The ability to manage a group of users is also invaluable, and Linspire's ability to execute is admirable. I think an alliance with some other software companies like Mandrake, Xandros, and Ubuntu could rocket Linspire to becoming a solutions company. You see, I think they are close to having a solution, but the solution in my mind is software management (installation, tracking and distribution) and over time, I can see Linux becoming a more unified platform where solutions companies don't focus as much on their distributions (especially in the desktop space) but the services they provide. I think a company like Michael's that could break the dependency of a single distribution and provide that ease of use experience to the entire Linux platform would be a killer solution company. Adding the ability to control software management for any distribution of the Linux platform would be very interesting.
You see, Michael's right: Building an "operating system" is extremely resource-intensive. That's why Linux is gaining popularity, because desktop Linux and Linux in general is just a platform that is being customized. Linux distributors are benefiting from not having to individually develop kernels and file systems individually. They are sharing the burden. What you will see happen is that many of these companies will focus on certain solutions for that platform, I think the Linux Standards Base (www.linuxstandards.org) and the OSDL(www.osdl.org) are becoming good shepherds of the OS that is going to drive this change.
Also, I believe that in any established industry where geography isn't a factor (the Internet has eliminated geography as a barrier in so many cases), what happens is due to the need to realize greater efficiencies through economies of scale there becomes a market leader with the lion's share of the market. Then three or four other players start to challenge that market leader trying to erode that majority. What's happening now is that people are dissatisfied with their options for desktop computing and are starting to search for alternatives to Microsoft. There are opportunities for these second tier players (Apple, Novell, Linspire, and others) to start to erode Microsoft's market share. So rather than predict what companies will form alliances I would say the question is, "What companies' products when combined will form solutions?"