Intel (INTC 2.13%) has managed to build a near-monopoly in the server chip market during the past few years. Competitor AMD has largely become irrelevant, claiming a low-single digit share at best, and Intel's x86 architecture has become the de facto standard in the server world.

International Business Machines (IBM 0.13%), which recently sold its x86 server business to Lenovo, sells servers and mainframes built around its POWER architecture. While many industries are dependent on IBM's mainframes, Intel's x86 architecture has become the king of the data center.

Intel's dominance has allowed the company to extract outrageous profits from its data center segment, and companies like Google, which buys a tremendous number of server chips each year, have little choice but to pay whatever Intel charges for its chips. IBM took the first step toward shifting the balance of power last year with the formation of the OpenPOWER Foundation, a consortium of companies working to make the POWER architecture a viable alternative to Intel's products. With companies like Google, NVIDIA, and Samsung backing the effort, OpenPOWER represents a real threat to Intel's lucrative server business.

The plan to take down Intel
IBM is licensing its new POWER8 architecture, as well as all previous versions, to members of the OpenPOWER Foundation. Companies will be able to design their own POWER8 chips, and have them manufactured at the foundry of their choice, as well as design other server components that will tightly integrate with the POWER8 processor.

IBM recently announced the first servers under the OpenPOWER initiative, which include a GPU accelerator card from NVIDIA, and run the Ubuntu Linux operating system. Canonical, the company behind Ubuntu, is also a member of the OpenPOWER Foundation.

IBM's POWER8 chips are extremely powerful, and the company claims that the ability of the chips to tightly integrate with coprocessors from NVIDIA and other companies give POWER8 servers the ability to analyze certain Big Data workloads up to 1,000 times faster than comparable x86 systems. The advantage is much smaller when considering just the processors themselves, but IBM's POWER8 chips are still more powerful than Intel's offerings on a per-core basis.

It will take more than just a powerful processor to challenge Intel; but with support from big companies like Google, which would likely benefit if it could replace Intel chips in its data centers, POWER8 has a real chance of ending Intel's server dominance.

Intel's outrageous profits
Since 2007, the average selling price of Intel's server chips has risen by 47%, from $429 to $629. This is the opposite of what typically happens in other chip markets, where there's at least some competition. It's obvious why Google is supporting an alternative to Intel's server chips.

The rising price of Intel's server chips has allowed the company to achieve margins in its data center business that are truly extraordinary. During fiscal 2013, Intel managed an operating margin of 46% in its data center segment, compared to only 36% in the company's PC segment. This enormous operating margin represents an opportunity for IBM, and if POWER8 servers can offer better value to enterprise customers, Intel may not be able to maintain this level of profitability in the long run.

Even if Intel doesn't lose very much market share, the competition may force the company to lower prices, and that would negatively affect the company's bottom line. While Intel derives more than twice the operating income from its PC segment as it does from its data center segment, server chips represent a larger growth opportunity compared to the stagnating PC market.

Final thoughts
It's difficult to predict whether IBM and its partners will be able to win a significant amount of market share from Intel. The level of support behind the OpenPOWER Foundation leads me to believe that IBM has a real chance at ending Intel's server chip monopoly. If that happens, the days of 46% operating margins for Intel's server business could be coming to an end.