It's hard to believe it, but there was a time when Intel (INTC 5.51%) was the small-fry trying to challenge server chip giants such as International Business Machines (IBM +0.90%) and Sun Microsystems (later acquired by Oracle) on their turf. The strategy was actually quite cunning: leverage the economies of scale afforded to it by its mainstream PC processors to build high margin – but still cheaper than the competition – server processors that leveraged much of that same design work. While much has been made about the potential threat that ARM (ARMH +0.00%) and its licensees pose to Intel's server chip dominance, it seems that there's much more hype than substance here.
If ARM vendors go cheap, Intel can go cheaper
A big part of why Intel was able to "win" in the datacenter back in the 90's was due to the fact that it could provide a significantly more cost-effective solution that still offered solid performance. Of course, many have taken this lesson and have used it to extrapolate that since ARM based chips are typically cheaper than Intel's high end PC processors that the same fate awaits Intel. This couldn't be farther from the truth.
While there is still debate as to how much of the broader server market will be serviced by the "wimpy" cores versus the "bigger" cores that Intel develops (although in theory, there is nothing stopping an ARM licensee from developing a "big" core), the point is largely moot. The ARM vendors' 64-bit server chips have yet to hit the market, Intel is already attacking this market with Atom-based system-on-chip parts codenamed "Avoton" and "Rangeley." These are 64-bit, very low power, sport a very set of 2-8 processor cores, and offer full compatibility with the suite of x86 software.
From a purely economic standpoint, Intel couldn't be in a better position. There's an incredible amount of R&D leverage to be had here, from reuse of processor cores and other IP that will be spread across multiple different product line within Intel – the majority of the ARM server start-ups are betting their entire companies on this one effort and the volumes in servers simply aren't the same as tablets or smartphones. Going a step further, note that most of the ARM server vendors are paying both the external foundry margin of the likes of TSMC, packaging and test costs to the likes of Amkor, or Advanced Semiconductor Engineering as well as license and subsequent royalty fees to ARM.
But will Intel's margins come under pressure?
Another major concern is that if Intel is able to successfully defend its market share, will it be able to do so without sacrificing its gross margin profile? It seems likely. Intel has always had competition – from IBM and Oracle in the big iron space, from Advanced Micro Devices in the traditional X86 server space, and now from a handful of ARM vendors such as Calxeda and Applied Micro – and yet it has consistently commanded high gross (and operating) margin for its server division.
The key is that the server market isn't quite like other markets. Sure, the upfront cost of the hardware is an important consideration, but an even more important consideration is total cost of ownership. A company that can offer chips with better performance-per-watt and easier maintenance and lower software costs (porting X86 software to ARM and then validating it is hard work) can charge a premium for its chips. While Intel could engage in a price war with the ARM vendors in the near term and still be immensely profitable, it most likely won't need to if it can continue to leverage its process lead to maintain total cost of ownership leadership.
The Foolish bottom line
Intel was able to defeat IBM, Oracle, and countless other big RISC architectures back in the 90's, and it is likely to do the same to the smaller ARM vendors of today. If a larger, consumer-levered name such as Qualcomm or Samsung decided to play in the ARM server market, then it could stand a better chance from a sheer economics standpoint, but they still wouldn't have the years of experience of developing server solutions, the process technology lead, and the partnerships that Intel has. This is why Intel is likely to continue to dominate the server space for the foreseeable future.
