The year ahead will be no less challenging than the one we’re now exiting: We’re going into an election year, which is always tumultuous; there are signs of economic rough waters; and Silicon Valley is headed into a new boom that some are already comparing to the 1999 dot-com mania. And we all know how that ended.
Fortunately, Digg, Facebook, LinkedIn and the like also seem to remember the dot-com bust, and are learning lessons from the fate of Boo.com and The Industry Standard.
Just as the newest Silicon Valley pups are facing their own challenges, so do some of its biggest players: chipmakers AMD and Intel, which are looking to major updates in their designs to beef up their businesses and erase some of the missteps of the recent past.
But those two aren’t alone in their efforts to devise new technology. On the dark side, malware authors are growing ever-more savvy and show no signs of slowing in their own efforts at “innovation,” such as it is.
Still, technology innovation’s positive benefits may yet outweigh the bad: This year, server virtualization came into its own, offering myriad advantages for the enterprise. Not the least of which, luckily, is improved security.
AMD and Intel: All in with the chips
The biggest hardware challenge in 2008 undoubtedly will be the continuing fight between the two vendors whose CPUs power virtually all of our computers: Intel and AMD.
Until 2003, Intel was in a solid and somewhat complacent leading position, while AMD was a distant second with chips that were rarely competitive to Intel’s top of the line.
That all changed in 2003 with the introduction of 64-bit chips, and later dual-core chips on the desktop and the Opteron server processor. In four years, AMD went from having no major OEMs to all of them.
That’s quite a reversal. When the Opteron came out, not one server vendor was a licensee — and there aren’t a lot of server vendors out there. Today, you can buy an IBM System x, an HP ProLiant, a SunFire or Dell PowerEdge servers with AMD Opteron chips.
Veteran semiconductor market analyst Nathan Brookwood of Insight64 puts AMD’s market share at just about 50 percent of retail, with slightly more than half in desktops and less than half in notebooks.
“That’s a huge move from where they were, and it’s not all the low-priced stuff,” he said.
It was this success that proved AMD’s undoing. In short, it grew faster than it could handle, sort of like a teenager still trying to fit into its pre-teen clothes. It couldn’t produce enough product to satisfy demand when it added Dell as a customer in 2006 and the company got creamed in late ’06 and into ’07, and spent the year recovering.
Getting its production capacity in line has been AMD’s goal this year. Its problem wasn’t creating chips — it was making them. The massive Dresden, Germany plant had to be converted from 65-nanometer designs to 45nm, and at the same time upgrade from 200mm to 300mm wafers. Oh, and it also had to get the Quad-Core Opteron, a.k.a. “Barcelona,” out the door. A daunting plan for a company a fraction of Intel’s size with nowhere near its resources.
AMD has an ace up its sleeve with ATI. The $5.4 billion acquisition to date has been more of a drag on the company, but benefits are finally starting to show, beginning with the “Spider” platform. AMD has the advantage over Intel and Nvidia, its chief rival to ATI, in that it can put together a complete PC platform with all of the chips needed.
Going into 2008, AMD’s challenge then is chiefly about getting to 45nm chip designs and shipping “Shanghai,” the successor to Barcelona. Shanghai will supplant the quad-core Opteron CPU with a new 45nm process and 6MB of L3 cache, which is shared among its cores. Barcelona has only 2MB of L3 cache.
“They darn well better get their 45nm products up and running and out on a more timely basis than they did with Barcelona,” Brookwood said. “It’s important because AMD needs Shanghai to compete with [Intel’s upcoming] ‘Nehalem.’ If AMD does not have Shanghai to compete with Nehalem, then it will have a serious competitive problem.”
Like AMD, its daunting rival has not performed flawlessly company in the past, either — although most of Intel’s pain had been limited to 2006. For the world’s largest chipmaker, 2007 was a year of increasing momentum.
CEO Paul Otellini has cleaned house of all the bad old ideas, consolidated, cut and streamlined anywhere and everywhere to make Intel more efficient and very profitable.
The challenge in 2008 for Intel is rather simple: don’t screw up. With its scale and massive resources, the task may seem easy for Intel. But the company is taking a sizable risk with Nehalem.
The design, when it ships later in the year, will finally mark Intel’s dumping of the front-side bus — the external memory controller that all data must pass through when entering or exiting the CPU.
Abandoning the front-side bus should mean increased bandwidth. However, doing so requires a whole new architecture, new chipsets, new motherboards and a new way to handle memory.
“This means changes across the board, it’s not just a chip update,” Brookwood said. “Platform and processor change should not be underestimated.”
Next page: Innovations in viruses and cybercrime? You’d better believe it.