AMD foresees strength in server CPUs, big market opportunity for ATI
In the PC industry, it's the OEMs that are the first to feel the effects of a recovering global economy, as demand for PCs from four of the world's top five producers has picked back up. It could actually be a good holiday quarter for HP and Acer. So if Dell or anyone else is still feeling negative impact, analysts are coming to the conclusion that the economy isn't to blame this time.
For CPU manufacturers, the economy works a little differently. With parts inventory having built up already, PC manufacturers have to sell through their existing product lines before they start ordering new replacement parts. Those "new" parts may actually be a little old, which is why AMD has a few more 65 nm parts on hand than it had planned. Getting rid of those parts means cutting prices, which is one reason why gross margins -- today's key indicator of a producer's relative health -- came in at lower than expected for AMD's second quarter of 2009.
"We have not given up on our goal to be break-even, cash-flow-positive in the back half of the year," said AMD CFO Bob Rivet yesterday. "I'll be very disappointed if we don't exit the year at a 40% gross margin." Later, Rivet clarified: He doesn't believe the company's Q4 gross margins will come in at 40% spot-on, but he does believe the needle will be on the "40" come December. That could mean good news for Q1 2010.
Revenues for AMD for the last quarter came in just over 13% below the company's Q2 2008 numbers, now at $1.18 billion. That correlates almost precisely with analysts' estimates. Like an amateur figure skater competing under the old ten-point scoring system, Intel had the bad luck of performing first with its quarterly earnings report last week. Those numbers were largely positive, but they set a kind of benchmark for AMD by helping analysts to establish a baseline for the bad economy's effect on the entire IT sector. When AMD's revenues sank just 13%, they sat right at the baseline, effectively affirming for analysts that its fundamentals were actually pretty sound.
At this time last year, AMD's gross margins were at 38% -- the part of those revenues that AMD divvies up to become profit, after cost of sales is subtracted. In the last quarter, they rose to 43%, which should have been a good sign. But factory utilization played a factor here: Since fewer PC manufacturers were ordering parts, AMD was spending less to produce new ones. "We actually didn't turn wafer starts on in a material amount of difference until June," Rivet told one analyst.
So now, margins are back down to 37%. What will AMD rely upon to make that needle hit 40% in six months' time? All of a sudden, we recall that AMD's not just a CPU company any more. Here is where its ATI division could play a very critical role, with the rollout of new GPUs that support the DirectX 11 drivers premiering with Microsoft Windows 7.
CEO Dirk Meyer said that DX11 will help ATI bring out its upper-level, 40 nm GPUs. "Clearly that'll give us an opportunity to regain the high ground, even more substantially, and hence help improve our mix and ASP [average selling price], and enhance margins." Toward the end of the call, in response to one analyst pointing out that ATI is still losing money for AMD, and that Nvidia is making it hard for AMD to respond by keeping prices low, Meyer added that he didn't expect price points to change toward the end of the year.
And then he threw in this line: "Where we compete, I'll say line-on-line, I think the new dynamic in the back half of the year, we're going to be able to offer things that they [Nvidia] won't be able to provide an answer to.
Three years ago at this time, Windows Vista was being relied upon as a catalyst for new hardware sales, especially from ATI and Nvidia, both of which contributed their own resources to developing Vista's new driver model. They had both hoped for a payoff either from Vista owners who needed system upgrades for better performance, or from new PC buyers who acquired Vista for the better performance. That payoff was less than predicted; so with Windows 7 this time around, AMD was clearly less enthusiastic about embracing it wholeheartedly.
"Windows 7, awfully hard to predict," Meyer at one point admitted, in response to an analyst's question as to how he believed the new operating system would impact the back-to-school market. "Good product, well executed, I'm enthusiastic particularly about the [DirectX 11] technology that's being dragged along at the same time, so I'm hopeful that that's going to motivate people to buy a little bit richer mix of graphics technology in their systems. I think one of the wildcards is the degree to which the consumer channel stalls a little bit, waiting for Win7 machines to show up on shelves. We're not modeling that as a big factor, and a lot of those machines are going to ship in advance of Win7 availability with upgrade coupons, so I'm hopeful that that won't stall the channel."
But in contrast to how CPU makers initially embraced the onset of Windows Vista, Meyer actually downplayed the importance of Windows 7 as a positive factor in re-energizing the PC market: "We're not modeling Win7 as being a huge contributor to overall unit demand," the CEO said. "But...I hope at the very least there's some ability for us in the industry to upsell to more richly configured machines, particularly around the GPU."
This season, AMD is looking for sales of its new six-core server CPUs to help gain back valuable market share lost to Intel, and to help put its profits back in black-ink columns. Desktop CPU sales, including for its new Black Edition Phenom IIs, could still be better, both executives admitted.
Yet there remains some real questions as to whether the netbook form factor is solid enough to be considered a niche unto itself. Clearly Intel now believes it is, with the success of its Atom CPU carrying it through the worst part of the economic mess; and certainly reports from DigiTimes regarding Intel's likely delay of its newest Atom N450 chip, code-named "Pine Trail," are being taken seriously. AMD does not have a chip to counter either Atom from Intel or Ion from Nvidia, so it may be in its own best interests to downplay market demand in that division.
But now, the demand for what could at least be described as "smaller notebook PCs" has measurably increased; and meanwhile, competitors in the embedded space such as ARM are demonstrating a potential latent market in handheld computers that may or may not be phones. So how exactly does AMD plan to carve out a market space in this area, regardless of what it chooses to call it?
"First, we're aggressively working on more power-efficient, lower-cost components purpose-built for lower price points in the marketplace," said Meyer, "like those associated with netbooks. We continue to see kind of a gap between a netbook on the one hand and smartphones on the other. We're clearly targeting devices with reasonably useful keyboards and 10-inch screen sizes, and the continuum of PC-like devices above that, independent of what operating system happens to be loaded on those machines. The smartphone space, little bit of a different story. As you know, we sold our handset division, or at least some of the technology assets, a couple of quarters ago; and while our technology is applicable to some of those form factors, that's clearly not a market focus for us at this time.
"We're clearly focused on 10-inch screen size and above," Meyer added, effectively setting a "floor" on the minimum size of the market in which AMD feels it can compete.
AMD's cross-licensing agreement with Intel over x86 processors may yet impact its business model with respect to Global Foundries, the former production end of AMD that it spun off earlier this year in a desperate restructuring move. That cross-licensing agreement prohibits AMD from sharing Intel's x86 instruction set technology with anyone other than a company subsidiary, and AMD had plans on spinning GF off entirely. Now, CFO Rivet admitted, it may have to consider a modified business arrangement whereby AMD maintains some small subsidiary stake in the foundry company, though executives would not go into details. They did say that GF may be ready to take on its first non-AMD customer by the end of this year, and murmurs between Rivet and Meyer seemed to translate to "Bull Technologies," though perhaps that name wasn't supposed to be spoken aloud just yet.