Advanced Micro Devices (AMD 1.42%) has for long lagged Intel (INTC -2.27%) in the high-performance computing space. Its limited grab over the retail and enterprise chip segment has hampered its revenue growth over the recent years.
AMD may have an ace up its sleeve, however. The chipmaker intends to aggressively capture the server market with its ARM (ARMH)-based Opteron offerings. In my opinion, this product direction has the potential to propel AMD's growth going forward.
Breaking conventions
It's worth noting that AMD and Intel have historically used similar x86 and x64 instruction sets in their respective 32 and 64-bit chips. Intel's repeated use of relatively advanced fabrication processes and chip architectures made sure that its offerings were much faster than their AMD-based counterparts, though. For this reason, AMD has struggled to catch up with Intel in recent years.
Gartner: 2013 server stats | ||
---|---|---|
Market share |
Revenue share |
|
Intel |
92% |
75% |
AMD |
7% |
5% |
IBM, Oracle etc. |
1% |
20% |
In this race to dominate the high-performance chip segment, Intel ignored the low-voltage smartphone and tablet segment -- perhaps for too long. This allowed power-efficient ARM-based offerings to gain mass popularity, even though they were much slower – and offered less hardware-level features – than Intel's full-fledged laptop chips. Impetus on power efficiency led to ARM's rise.
Aiming to also capture the prospective market base of power-efficient servers, ARM has begun designing 64-bit enterprise-scale server chips. AMD has licensed some of these designs to challenge Intel Xeon's hegemony in the server segment, albeit with its expected power efficiency.
Potential market
Regarding prospective clients, web-based companies that develop their own code – like Google and Amazon – may be willing to switch to the ARM architecture. Software and hosting companies like Microsoft and Rackspace would likely want to retain their current server architecture, however.
The current market for ARM-powered servers therefore appears to be very small. Gartner expects ARM-based servers chips to represent a 5% revenue share by 2017. That doesn't mean ARM has no future in the server segment, though.
ARM-based offerings aren't designed to compete with their high-performance x86-based counterparts in the first place. Having a niche target market should in theory allow ARM-based server manufacturers to prosper without competing directly with established high-performance Intel, IBM, and Oracle server chips.
It's also worth noting that AMD's upcoming SkyBridge framework – supporting both x86 and ARM chips – is an innovative step forward in the field of ambidextrous computing. To an extent, this will eliminate the need for Android virtualization on x86 platforms – something that developers often rely on – and expand AMD's total addressable market.
These chips will most likely be priced substantially lower than Intel's Xeon as well, allowing them to compensate for their moderate performance. This discounted pricing may be another contributing factor in the pick up of ARM-based server demand.
Why it matters
ARM's server business may not involve high margins, but it should contribute in giving a much-needed boost to AMD's top line. In fact, AMD management expects to capture a 25% volume-based server market share by 2019. How this figure will translate into a profit-based market share, however, will depend mostly upon AMD's execution, benchmarks, and product roadmap.