On the heels of the company’s financial conference, there is a buzz in the air about AMD and its new management team headed by Rory Read.
From the presentation material to the press reports, it would seem that AMD made all the right moves and said all the right things.
The company is re-enforcing its position in the x86 with both CPUs and APUs, continues to push the boundaries and it’s chief competitor Nvidia in graphics, and is targeting embedded applications like digital signage, medical, gaming, communications, and storage with custom SoCs. However, vague references to these new SoC solutions have sparked the rumor mill that AMD could be partnering with ARM or other processor IP vendors for future devices.
While I have been a vocal proponent of AMD seeking a new business strategy focused more on IP than chips, I still see many challenges ahead.
The customized SoC is positive in that it potentially leverages AMD’s graphics technology and design expertise. If these customized SoCs are based on the x86 architecture, then AMD also has the advantage of the CPU knowledge and software expertise. But, how will this translate to real-world applications?
Customized solutions are applications that either have very high prices, like military and aerospace applications, or very large volumes like handsets. While the x86 architecture can compete in the high-price applications, especially with the extensive software libraries and resources, the architecture has fallen short in many of the high-volume applications, particularly mobile devices.
Even the industry leader, Intel, has struggled to reduce the power consumption of the x86 architecture enough to compete with the ARM architecture. This leaves relatively small volume applications that typically require a considerable amount of support.
The alternative for those high-volume applications would be to use an alternative CPU architecture like ARM or MIPS. However, that requires an entirely different knowledge set for both the hardware and the software. Acquiring this expertise would likely have a high price in terms of financial investment and/or time.
Unfortunately, AMD is a small company relative to other processor vendors with limited resources. In the past, AMD has struggled with maintaining a consistent pace of new products and innovation, competing against its much larger competitor Intel, bearing the burden of large financial investments, and even maintaining a consistent marketing strategy.
This is a company that has launched new brands in the face of competition without the adequate budget to even support such a routine activity. And, in the past year, AMD’s revenue increased only 1.5% compared to 25.6% for Intel and 36.3% for Qualcomm.
Even though AMD claims to being a “fast following” in technology, the company has lagged the competition by a range of a few quarters to a several generations in implementing features like power gating, deep sleep and turbo modes, and even GPU integration.
So, can the reinvigorated AMD really compete not only with Intel, but with other significant competitors like Qualcomm, Samsung, and Texas Instruments that all have entrenched positions in the market? AMD won’t even have a single chip (processor + graphics + I/O hub) solution until 2013.
Despite these challenges, all hope is not lost. To be successful with such a broad strategy is likely going to require acquiring other resources or merging with another entity. With deep investment pockets from ATIC, both of these are potential strategies.
Building a broad strategy strictly on internal growth, however, would take too long to develop, especially in light of rising competition for the x86 architecture. This leaves AMD in a precarious position of needing to react quickly in the face of growing competitive threats from changes in the market, as well as a new company strategy.
So, where are these new resources going to come from?
If AMD is going to thrive or even survive, they will likely be through mergers or acquisitions by or of AMD. Look for 2012 to be a pivotal year for the future of AMD.
Jim McGregor is chief technology strategist at market research firm In-Stat. His commentary does not necessarily reflect the opinions of EE Times, its staff or its parent company, UBM LLC.
Kudos on a very well written article. AMD is in a very hard place. It's latest Bulldozer architecture is already inferior to Intels. AMD only did well when its Athlon and Opteron designs were better performing than the Intels designs and even then barely. Intel is also looking down the barrel with their finfet technology which AMD does not have an answer for.
So for now it will stay in the low margin space. If they decide to move to ARM will that mean abandoning the Bobcat line?
I would hate to be in Rory's shoes.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.