Linux Platform Ecosystem to Grow to $36 Billion by 2008

by Timothy Prickett Morgan

So just how big is the Linux ecosystem today, how fast is it growing, and how big is it going to get? These are the three questions that IT managers, independent software vendors, and IT hardware suppliers are all trying to answer as they map out their plans for the next several years. The answer to those questions, according to a study by IDC is big, fast, and a lot bigger.

IDC worked in conjunction with the Open Source Development Labs to gin up an economic model of the Linux market with respect to PCs, servers, and packaged application software. This aggregate, worldwide Linux platform market--and I have some problems with this characterization of these three elements as a fair representation of the Linux market--is expected to grow at a compound annual growth rate (CAGR) of 25.9 percent to $35.7 billion by the end of 2008, up from just over $11 billion in 2003. To put that in plain English, the Linux market (as defined by the IDC report) is going to triple in between 2003 and 2008.

The submarkets of the Linux ecosystem are growing at different rates, of course. In the server space, where Linux got its first beachhead in the commercial IT community, the value of servers running Linux as either their primary or secondary operating system is expected to continue to grow, even after the phenomenal growth rates of the past few years. But revenue growth is cooling, at about 21 percent CAGR between 2003 and 2008. Nonetheless, Linux servers are expected to account for 3.3 million units (representing a 24.5 percent CAGR for server shipments) and $11 billion in sales in 2008. The installed base of Linux servers (including the population of repurposed machines that originally shipped with another operating system) will balloon to 9.7 million units by 2008, according to IDC. In 2003, the Linux server market represented just over 1 million unit shipments and just under $5 billion in sales.

In 2004, IDC reckoned through surveys of IT shops that for every five new Linux servers shipped, another two Linux servers were either created from a used machine or had Linux as a secondary operating system using virtual partitioning technology. (Specifically, IDC says that the installed base of Linux servers in 2004 was 37 percent greater than the net new systems shipped when repurposed and virtualized machines are taken into account. Our way of saying it makes it more clear what the impact of repurposing is.) IDC says that on a worldwide basis, Linux represents about 16.5 percent of server shipments and redeployments in 2003 (with a higher rate in the Americas and lower rates in Europe and Asia), and that this penetration rate will grow to about 28 percent by 2008. In the Americas in 2008, IDC says that about a third of all new, redeployed, or virtualized servers will ship with Linux, with about a quarter of servers in Europe and Asia having Linux in 2008.

On the PC front, IDC expects that Linux will gain more traction in Europe and Asia than in the Americas. The company is forecasting that shipments of Linux PCs will grow from about 4.5 million units shipped and just under $4 billion in sales in 2003 to just over 17 million units shipped and $10 billion in sales by 2008. The company is predicting that the total worldwide Linux PC installed base will be 42.6 million machines by 2008, including repurposed machines. This is a big enough market to foster Linux-specific application development.

As for Linux-based packaged application software, IDC is figuring the market will utterly explode, from about $2.5 billion in sales in 2003 to over $14 billion in 2008. That's a CAGR of 44.3 percent, and represents one of the fastest-growing market segments in the entire IT sector. Considering that many application software suppliers in the Linux market do not charge a lot of money for their software and support, that revenue growth is stunning. However, the bulk of that revenue is going to come from mainstream ERP software suppliers who increasingly will support and peddle Linux-based solutions to a Linux-hungry IT market.

While the IDC study did try to more accurately characterize the Linux ecosystem by including recycled PCs and servers and machines with dual-boot capability, a more accurate representation of the Linux market would take a stab at trying to quantify the amount of money that companies don't spend by using Linux. For instance, a lot of companies have many more instances of Linux running on their gear than they paid for, which is not illegal since Linux is open source. The count of machines on paid-up tech support, which is where the Linux money comes from, is probably even lower. Many companies that use Linux to support stable workloads find that it is reliable and tend not to pay for support for machines that do not see a lot of changes in their software. In essence, these machines become appliances. There is not an easy way to measure the economic value of a four-year-old server running an open source implementation of Linux or a commercial distribution that users download and compile themselves (and do not pay for), except to look at the cost of the server and operating system that they didn't buy as a result of the availability of this used (and economically valueless) iron and free Linux. When you add all of this up, there has to be billions of dollars of value (but no actual money) more in the Linux ecosystem.

At IT Jungle, for example, we have several redundant Web servers running Linux and FreeBSD and we do not pay for support for these boxes since it is prohibitively expensive to do so. If a Web server dies (which has yet to happen in two years), we let it die and another one takes over; we would then fix the broken server and add it back into the cluster. Our core email and groupware server is, however, on paid tech support because it is the heart of our business. By the way IDC measured the Linux server market, we have only one Linux server, when in fact we have six.

The other thing that the IDC analysis did not include was the economic value of the open source development tools that are part and parcel of the Linux platform. On other platforms, programmers have to pay for compilers and change management systems. There are millions of programmers who work on Linux machines who have not paid for expensive Unix or Windows development tools. It is hard to say what the economic value of this part of the Linux ecosystem might be, but it probably represents a few thousand dollars per seat, on average, and that is again several billions of dollars of money not spent.