Comment With its expected reversal of fortunes from last year's awful first quarter, Intel has set its sights on having a second quarter of 2010 that is better than normal seasonal trends and perhaps a much better second half as all of its PC and server chips hit their market strides.
In a conference call Tuesday with Intel execs announcing the company's first-quarter 2010 results, Wall Street analysts learned that they had undershot Intel's performance - even though they had suspected things were on the rebound. They also learned that Intel is optimistic about the rest of the year.
During the call, Intel's chief financial officer Stacy Smith said that although CPU shipments were down, PC buyers went for performance on the new Westmere family of Core processors for PCs, which boosted the company's average selling prices (ASPs) for chips. This boost smoothed a decline that was less steep than chipmakers normally see moving from the end of one year to the beginning of the next.
With Intel's new six-core Xeon 5600, eight-core Xeon 7500, and four-core Itanium chips ramping up volumes in servers, the same phenomenon is expected to repeat itself in the server racket, giving the chip maker a healthy bump-up in sales - and driving profits.
Part of the reason Intel could be aggressive with the Westmere Core chips is that its transition from 45 to 32 nanometer processes is moving along more smoothly than any wafer-baking shift in Intel's history, according to Paul Otellini, Intel's president and CEO. Intel now has two factories baking 32nm chips, and will have two more up and running by early in the fourth quarter.
Intel shipped the Nehalem-EX Xeon 7500s for two weeks during the quarter, so their revenue didn't help all that much to juice the balance sheet. But these high-end chips, which are being used to make fat memory two-socket and four-socket servers, and even more scalable machines with from 8 to 256 sockets, will contribute to what Otellini called "that better-than-seasonality picture we are painting."
Because everyone knew that the Xeon 5600s and Xeon 7500s were coming out in March, the server processor mix was a little thinner in the first quarter of 2010 than it was in fourth quarter 2009. Otellini said that sales of uniprocessor Xeon chips used in entry servers aimed at SMBs helped pick up some of the slack. He added that demand for these entry Xeon processors was higher in the first quarter than it was throughout the second half of 2009. This uptick in entry servers is also one of the reasons why Intel is anticipating a better-than-average pickup in the server-chip business in the second quarter of this year.
As for the PC and server refresh cycle that is currently gaining momentum after two years of brakes being slammed on spending, Otellini lectured that he "would like to think that people are looking at these things with a clear economic eye," and that they would realize they should refresh their aging servers first and then worry about the PC fleet.
But what no one called him on during the call is that Intel refreshed its PC chips first, and maybe should have released its server chips first if it wanted a ramp in Q1. A very clear economic case based on ROI and TCO can be made to get rid of servers using single-core x64 processors from either Intel or Advanced Micro Devices.
But, then again, a server that has long since paid for itself can be cheaper to keep on the short term, particularly if you wanted to wait for the good technology - like Xeon 5600s and 7500s and Opteron 4100s and 6100s - before spending. It's a wise customer that saves the budget for a quarter or two, gets by, and then gets even better ROI and TCO.
Once the Lisbon Opteron 4100s are in the field by the end of the second quarter, everyone will know where the products all stand in terms of performance, capacity, and price. The price war will get well underway, and we'll see how much demand there really is. If server demand is high, processor ASPs should stay steady or even rise a bit for Intel - and given the aggressive pricing AMD has, ASPs are probably going to fall as volumes pick up.
AMD needs to have a greater market share in servers to command more attention from server makers, ISVs, and customers, and it seems hell bent on getting the kind of share it enjoyed a few years ago when Intel's product line had some issues.
Unlike some industry observers who caution that the long-term effects of server virtualization and now cloud computing, which respectively increase server utilization within organizations and then across them, Otellini says he is not worried about a virtualization or cloud downdraft on Intel's server business: "If you plot out the growth in data traffic and network traffic and the kinds of things modern servers are doing, that growth curve is faster than the refresh rate for old versus new equipment," he explained. "We see a very robust scenario for servers going forward." ®