I believe that Linux is a significant challenger to Microsoft.
I can point towards the significant (and growing) number of linux installs - 7 million or so. In the world of sotware marketing, it is important to remember that the cost/unit to the maker is approximately 0. This means every unit that isn't sold eats directly into profits. Even if a small fraction (say 1/7?) of the 7 million installs resulted in directly or indirectly not purchasing Microsoft products it is the beginning of something significant. I figure approximately 1 million * $100 = $100 million in lost revenue to Microsoft.
$100 million is not yet significant to Microsoft, so the real question is whether or not Linux will grow. The answer appears to be a pretty resounding 'yes'. First, look at the past:
To me, the interesting question which arises is "How will linux grow?" First of all, Linux is technically superior to Windows NT/95/98 and I don't expect to see this change. Linux can typically do more work in a thinner hardware envelope than Windows XX. Linux also boasts superior (technical support) and is a far cheaper solution running free software on cheaper hardware.
I'll offer some predictions. In the computer world the natural time scale is around 1 1/2 years so I'll confine myself to this time scale.
I've also noted some of the obstacles to linux technology adoption. First, in the minds of corporate managers there is a serious inability to recognize large differences between _value_ and _cost_. This is because traditional economics, which every member corporate management knows (at least intuitively) suggest that value (nearly) equals cost. For software, the natural cost is the price/megabyte of a storage while the natural value has something to do with the technical excellence and work of the programmers building the software. Value and cost can diverge very significantly. I expect to see corporate america struggling with this dichotomy between reality and expectation for several years.
Second, it is far easier under Linux to 'get around' artificial software limitations such as 'expiration dates' or the requirement of a CD-ROM. This ease of 'getting around' is nice for the end user, but could seriously cut down on sales of software products by commercial companies. I anticipate game companies in particular preferring Windows NT/95/98 etc... because Microsoft is willing to conspire with them to artificially make copying harder. Windows XX will become the 'cripple OS' under which new games are released.