SANTA CLARA, Calif. -- ARM this week sketched out a vision of
moving methodically into servers and deeper into embedded, expanding
its reach from a dominant position in mobile systems.
There's an architecture, a vision, a market need and an evolving
ecosystem. A lot of great generals throughout history had similar ambitions
but overreached and outran their supply lines. What's different today?
History is littered with powerful companies that dominated a market
and pushed their technology into new areas, thinking their brand,
technology or manufacturing would carry the day. Companies often overreach, mistakenly thinking applications will
embrace the technology rather than the other way around. A decade ago, Intel was poised to crash the networking communications party.
Didn't work. Different considerations, different sales cycle than
the PC industry. More competition too. In addition
to hubris, companies simply can lose focus or suffer from
in-fighting and tech turf battles.
I'd wager that ARM's expansion into embedded and
push into servers won't blow up in its face. Embedded design
considerations aren't too far afield from mobile phones and tablets,
and ARM already has a presence in many embedded systems.
Servers are a different animal altogether, but the reason ARM probably
won't experience an Intel moment is structural. ARM is a step
removed from the sturm und drang of a given market in that it
defines a relevant architecture and lets partners run with the ball
(and assume the risk).
It has no capital equipment skin in the game, and it
studiously builds an infrastructure community (standards, software,
peripherals) to raise the odds of success. Management and
engineering seem conservative but not overly cautious.
For its drive into servers, it's enlisted a more-than-willing
partner in AMD, already has success with Calxeda and is actively
managing the design conversation through things, like HSA (Heterogeneous System Architecture) Foundation and
TrustZone. This isn't to say Intel isn't great at seeding a market
and building community; it is. But Intel's manufacturing exposure is what makes
any expansion a higher-risk endeavor.
Servers is a gamble for ARM. They have very little invested in it. They are counting on their partners - Calxeda, AppliedMicro, AMD and others to invest millions and take the risk. All they have to do is to define the architecture, work on enabling the ecosystem and stand back and watch the battle.
I agree Rick...I think it is going to be long and uphill road for ARM to conquer server space...despite Peter's response I am still not convinced why they don't just focus on mobile which is a huge market on its own to propel them into the top 10 semi vendors shortly...kris
I'm no microcontroller guru, but my sense is ARM is pretty far along in surrounding that market in which a few remaining proprietary architectures have circled their wagons.
As for servers, ARM still has some heavy lifting to do in ecosystem software and the key 64-bit chips won't even arrive until 2014, so we are just in the preface of this book.
Businesses, like sharks, have to keep moving forward.
I remember reading ARM's revamped "vision statement" about six years ago.
It read something like this: "ARM intends to be the preferred digital architecture in everything."
At that time ARM was big in mobile and trying in other sectors and vision statements were all the rage.
I did a double take at the time but i have got used to the idea of "first mobile, then the universe."
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.