Ummm... Revenues have nothing to do with market share. Unix is pervasive, but it isn't generating much in the way of direct revenues these days. It is generating a ton of indirect service revenue.
For example, one of the companies I am currently involved with operates one of the fastest growing fiber optic networks in North America and possibly the world currently, and growing like crazy. In fact, their fabric is becoming an increasingly important part of the North American internet. They have a core mission critical database system that controls their global networks and all realtime billing data and so forth i.e. that system is their business. The application runs on "re-purposed" Windows servers that now run Linux which nobody paid for. Using your metric, this shows that Microsoft is "winning" because they sold a license for these servers, and that Linux/Unix is "losing" because nobody got paid for the deployment.
It is worth noting that these guys (a sharp bunch with a very slick operation) could not have attracted investor money (not that they ever needed it) if they were using Windows to support their operation, since it is a 24/7 business. As in, that was actually a question people interested in investing in the company ask.
The important part here is that while both Windows and Linux are being deployed in high numbers, the mission-critical backend is mostly Unix of one stripe or another. The TCO argument for Windows is weak if Linux is pervasive on the backend, and the less-than-mission-critical positions that Windows servers are put in are frequently (but not always) fungible whereas mission critical services are not. In other words, Linux can replace what Windows is currently doing to a great extent, but Windows cannot replace what Linux is doing. The company I mentioned above will be replacing their current core Linux system with some really Big Iron soon, also running Linux. It is worth noting that this system is more powerful than anything Windows runs on -- as I said, Big Iron. Given that they are committed to running Linux since they require servers that are bigger than anything Windows can do, there is a certain amount of efficiency in homogenizing the other servers to all Linux as well where feasible.
And that is the kind of calculus that MS is facing. The number of things that Linux can do on servers that Windows can't is substantially greater than the number of things Windows can do on servers that Linux can't. Given this selection pressure and the low cost of Linux, Windows will slowly be marginalized, particularly since it is getting it ass kicked most severely in the higher-end markets by a product that is cheaper than Windows. Since they don't own the low-end market either, that is an uncomfortable position and they'll find themselves squeezed as the calculus finds favor in Linux many times.
I'm not a particularly big fan of Linux personally for a number of reasons, not the least of which is that I have a LOT of experience with many different operating systems. Still, I can't reasonably ignore the fact that Linux is cleaning house in the data center. Windows may still control the departmental file server market and such, but I don't see that as a particularly secure position if you are incapable of competing effectively in the data centers.