Linux server growth threatens MS revenue supply
One slice of the market pays lots, another doesn't - how long can that last?
Analysis This week's IDC report on the server market has some tantalising numbers for Linux fanciers. Linux server shipments will climb faster than the rest of the market, resulting in 2004 in Linux shipments of around 4.7 million, finally within spitting distance of Microsoft, at 5.7 million. But although this holds the possibility of Linux taking the lead in 2005 or 2006, that's not the real story - the changing underlying economics of the market are what Microsoft really has to be worried about.
As our old friend IDC pundit Dan Kusnetzky colourfully puts it in the release: "From a revenue perspective, Linux is nothing more than a speck of sand on an ocean beach." Total Linux sales last year, he says, amounted to about the same as Microsoft's operating systems business generated by the third working day of January 1999. So that's point one - Linux server shipments will start to catch up on Microsoft, but there will remain a vast chasm between relative revenue levels, so maybe Microsoft shouldn't worry, right?
Well maybe. The market as a whole will grow 17 per cent a year, with Linux chalking up 28 per cent within this. But revenues will be flat, with growth of only one per cent a year. This can partially be ascribed to the higher growth rate of Linux pulling the average down, but IDC expects a surge of shipments of upgrades to Microsoft server products, which will bring the number of shipments up but will only generate upgrade revenue.
That suggests Microsoft will be pedalling faster to stand still, in a market where prices are falling and where it's increasingly difficult for it to win new customers. Smaller amounts of upgrade revenue will tend to be coming from existing customers.
Although it'll have shipments accounting for 38 per cent of the market by 2004, Linux sales revenues will only be $85 million, IDC tells us. Typically, the outfit reckons that Microsoft sales tend to go to businesses wanting prepackaged solutions, while Linux ones go to companies who want to build their own.
But how sustainable is this over the next five years? It's first of all difficult to see how the market can balance nearly 40 per cent of shipments costing practically nothing (we should note here that those high valuation Linux outfits had better hope their services revenues climb like crazy), while a rather larger section of the market, but not that much larger, pays rather a lot. It would be reasonable to expect that model to start to fall apart on the way to 2004.
Then we've got the likelihood of packaged Linux solutions beginning to chip away at Microsoft's core server market. And of course, the .NET/rental strategy. Microsoft.NET currently seems dependent on two things; first of all, switching customers over to a services-based model, and from the look of the price indicators so far, winning more revenue from existing customers. The second part, the one that has so far tended not to work terribly well for Microsoft's server business, is to make the products so compelling that market share expands rapidly.
But again, you could see the dynamics of this changing - .NET is essentially the future of the Internet translated into Microsoftspeak. If Microsoft hadn't invented it (which it didn't) it would still happen anyway. You can expect rival services based on Linux and Unix servers to compete here, and given that the services will be Web-based, where Linux and Unix are the servers of choice, you can see Microsoft having a tough time of it, and again coming under price pressure.
That pressure, rather than the shipment numbers or the continuing disparity between revenue levels, is what's really important. Microsoft will continue to rake in vastly more revenue than the Linux vendors, but Microsoft is a monster that needs to eat a lot more money than Linux does, and needs to keep growing to keep the whole deal spinning. If that ceases to be the case, the monster is in big trouble. ®
Sponsored: Transform Your IT Infrastructure