AMD Sheds GlobalFoundries; Sets Q1 Revenue Record

Q1 was the first quarter AMD reported its results entirely separate from those of GlobalFoundries, and the company's numbers were quite good. Revenue for the quarter was a record-setting $1.57B, the company reported an operating income of $247 million, and an operating profit of $182 million. That's an excellent shift from a year ago, when the company lost $298 million on revenue of just $1.18 billion.

The company's gross margin of 47 percent (up two percent from Q4), thanks to a richer product mix. Revenue in both AMD's CPU and GPU businesses fell sequentially, in line with standard seasonal trends. AMD didn't break out its estimated market gains by segment, but believes it gained ground in server and mobile CPUs and in mobile GPUs as well. When asked about the TSMC situation and the current state of 40nm production, AMD declined to comment other than to confirm that yes, Radeon 5000 series GPUs are still production constrained. Although GPU revenue declined slightly from Q4, volume shipments grew, primarily thanks to a record number of discrete mobile chips.

This is AMD's Magny-Cours. The company says it's already seeing good uptake from its recent server launch.

Q1 2010 was one of AMD's best quarters ever, but the company will have to thread a very narrow needle to keep its nose pointed upwards. One of the things Sunnyvale's executives discussed on the conference call is that ASPs are expected to be flat; gross margin improvements, if they appear, will be the result of improved utilization or more profitable product mixes. That's a tacit admission that AMD's ASPs are effectively capped by Intel's ASPs at the moment, and while we don't expect any major upsets, Santa Clara is ultimately calling the shots.

Llano, AMD's first "Fusion" part, will begin production in the back half of 2010 and ship for revenue in Q1 (implying we could see a late Q4 launch), and AMD is moving Radeon production to GlobalFoundries beginning at the 28nm node. The company wasn't willing to commit to when that might happen specifically, but GF's own roadmap shows the company beginning 28nm production at the end of this year. Without commenting on specifics, the CPU manufacturer implied that Llano is looking good—functional samples of two different Fusion products are running in AMD's labs, with one chip already sampled to a few customers.

This is what you get if you Google Magny-Cours without the AMD part.

The one concern we have with AMD's results is whether or not the company can sustain its momentum. After covering AMD's performance for a decade, it's painfully clear that the CPU manufacturer has had trouble maintaining consistent gains in performance, market share, or revenue. There are a lot of theories as to why that's been the case, but there's no denying that AMD has consistently leaped ahead, only to slowly lose ground to Intel as time goes on.

Don't get us wrong, we're thrilled by the company's latest reports, and the CPU manufacturer has been on an upward trend since it launched Shanghai. At this point, however, AMD is effectively out of second chances. If it wants to continue to make back the ground its lost to Intel since the Core architecture debuted in 2006, Llano, Bulldozer, Bobcat, and Fusion all need to hit dead-on target. So far, things look good—here's hoping it stays that way.