ARM: A Trojan Horse Strategy - Microservers

The Microserver: Three Key Players, Three Different Views
By

an Ferguson, Director of Server Systems and Ecosystem at ARMan Ferguson, Director of Server Systems and Ecosystem at ARMJust like Intel, ARM knows of the strengths of its competition and is not making overly aggressive statements about its competitive strategy in the microserver market.

"They won't just let us tickle their bellies," said Ian Ferguson, Director of Server Systems and Ecosystem at ARM.

He told us that ARM has been working on the idea and production of server-oriented processors for at least five years when "first tier vendors" approached them about the possibility of such a chip. This initial idea has evolved into an actual product, which we expect to see late next year in prototype form and in customer implementations in 2014.

However, ARM will not go head-to-head with Intel. Instead, it will focus on those areas that especially benefit from its low power chip architecture. "This will not be the chip to throw gigahertz at. We will not be making claims of being the best thing since sliced bread," Ferguson said.

ARM's strategy is to go after the opportunities in which the company has established credibility through its low-power achievements.

That does not mean that Ferguson does not see performance for ARM processors improvements down the road, as he hinted toward GPGPU leverage in ARM SoCs that may see a boost via partners such as AMD. However, it would be wrong to see ARM as a direct Intel competitor in the microserver arena initially, as the two will not clash immediately: ARM is very conscious about its performance proposition and is unlikely to challenge Intel on these grounds anytime soon.

It is storage, networking and security appliances in web 2.0 and cloud environments where ARM believes it can be most compelling, initially. According to Ferguson, the code base in these applications is less important to data centers than the potential power savings and integration opportunities.

"These are areas in the data center that are based on high level code. Keeping the existing code do not always provide a massive benefit against the potential power savings," he said.

However, even if the current ARM strategy is not designed to go head-to-head with Intel, the approach of the company is to use the expertise of each of its partners such as Marvell, TI, Qualcomm and Nvidia to create customized chips for very specific purposes. Ferguson conceded that Intel has a clear advantage in available cash, marketing and technology resources. "But, for example, Marvell and Broadcom, are much better in developing networking products than Intel," he said.

Ultimately, ARM's initial opportunity lies in specialized products in the data center "where performance is not key" - and in products that are not in the immediate focus area of Intel. "We understand power better than Intel does," Ferguson confidently said.

Comments