AMD Announces Lackluster Q4; Questions Loom For Analyst Day
"AMD shipped more than 30 million APUs in 2011, resulting in record annual notebook revenue," said Rory Read, AMD president and CEO. "The unmatched combination of computing and graphics capabilities in our low-power "Brazos" platform has made it our fastest ramping platform ever, paving the way for continued growth in key segments and geographies. Our server business has regained momentum, delivering two consecutive quarters of strong sequential growth."
Record mobile shipments and server growth weren't enough to keep AMD in the black for Q4; the company reported a net loss of $177M for the quarter. This loss was due to a substantial write-down in the value of the company's GlobalFoundries investment and restructuring charges related to AMD's decision to fire ~1,000 people in late October.
Foundry Friendship Foundering:
AMD's decision to slash the estimated value of its GlobalFoundries investment by nearly 50%—from $478M to $286M—caught investors offguard. Thomas Seifert, AMD's Chief Financial Officer, offered this explanation: "based on certain indicators, their future growth outlook, they have changed the time line on the new fab in Abu Dhabi. Their business plans have changed, and we reevaluated our investment in the company." Seifert's explanation doesn't satisfy. GF's decision to postpone breaking ground on a fab that wouldn't start shipping products in 2015 is no reason to slash current valuation.
Since GlobalFoundries is privately held, investors have very few metrics by which to judge the company's performance. By voluntarily hacking off half its investment value, AMD has indicated what it thinks of the foundry's ability to execute long-term. CEO Rory Read noted that AMD's 32nm shipments grew by 80% this quarter, and stated that 32nm parts "now represent a full third of our product mix."
Asked if AMD was partnering with TSMC for 28nm APUs and beyond, Read said: "What we're really focused on is making sure we have the execution capability and to be able to deliver that consistently." That's CEO-speak for "Yes," and we expect the company will unveil those plans at its Analyst Day next week.
Big Questions, Inarticulate Answers:
The company has a lot of explaining to do come February 2. AMD's Brazos faces potential competition from 28nm ARM Cortex-A15 and Qualcomm's Krait CPUs later this year, particularly given that it's a 40nm chip going up against new cores built on a 28nm process. AMD's own plans for an "ultrathin" segment are a weak echo of Intel's ultrabook, and Trinity's CPU core will likely lose even more ground compared to Intel's Ivy Bridge. Sunnyvale will fight the latter trend by positioning Trinity in lower-end systems, which should help the core gain traction.
Read often emphasized the company's double-digit growth in servers, but such gains have little impact on the company's bottom line--AMD's share of the server market is too small to yield substantial short-term gains. The investor Q&A also illustrated just how badly Read flounders. Here's the question, asked by Steve Eliscu of UBS Investment Bank:
Q. Can you give us a sense as to the success you're having selling Opteron, the Interlagos into the data centers?Being articulate isn't at all the same thing as being smart, and there are people who simply don't answer questions well on the fly. That said, putting one of those people in charge as company CEO at a time when your ability to communicate clearly is of critical importance is a really bad idea. Read continues to claim that AMD's future is built around "low power, emerging markets, and the cloud." Hopefully next week, we'll get a strategy unveil that turns a string of buzzwords into a coherent product strategy.
A. I think server is another one of these key areas that we're seeing a change in the marketplace. We're approaching an inflection point. There's no doubt in my mind that we're seeing a breakdown in some of the proprietary control points that have dominated the market for years. And I think in the server space, based on what's happening in the cloud, what's happening around convergence, the types of workloads that are going to be driven through these mega data centers, these cloud data centers, the kinds of solutions that are going to be created going forward are going to be tailored to very unique almost server appliances that allow us in dense kinds of solutions to introduce, I don't know, disruptive, really disruptive server plate that allows us to win.
You'll see us continue to focus in the high-performance computing market, because that's always a forerunner of where next technology is going. We've got a strong foothold there. Continue to build on the database work that -- I mean, the traditional data center work that we're doing with the kinds of work around virtualization and around certain workload segments where our type of architecture really plays. And you'll see us get really aggressive around this disruptive play as we move forward.