AMAT: Foundries are the key to dimmed outlook

August 15, 2007 – Applied Materials said fiscal 3Q07 profits slid about 7.5% to $473.5M on almost-flat sales of $2.56B, which was better than analysts expected — though the firm cast some cloudy doubt on prospects for business through the rest of this year, particularly foundry spending that has not rebounded as expected.

New orders were down 14% both Q-Q and Y-Y to $2.28B, with memory making up 74% of that (42%/32% DRAM vs. flash), logic/other 20%, and foundries just 6%. Gross margins were a tad better than fiscal 2Q at 47.5%, impacted slightly by a $50M charge related to ceasing development of beamline implant tools.

Counted-on capacity expansion plans from both flat-panel display and foundry segments didn’t materialized as expected, according to company execs discussing results in a public conference call. “Logic and foundry utilization did not increase at the anticipated rate,” noted company CFO George Davis. Meanwhile, slow flat panel display orders means revenue in that part of the business won’t improve for at least several quarters, added president/CEO Mike Splinter. Solar contracts, however, have exceeded the $400M goal set during the previous quarter and are now expected to exceed $600M for the year.

Despite pricing pressures, memory customers are still confidently investing in advanced production capacity, and spending in this segment should remain strong, while logic investments should be basically flat, according to Splinter, adding that memory will still drive the bulk of future investments. He listed off factors such as strengthening end demand due to both seasonality and new applications (e.g. the iPhone), plus the “longer tail” of Vista adoption from corporations finally jumping on board the new OS, which “is going to drive an awful lot of DRAM.” Also, as NAND makers head to sub-65nm production a lot of 200mm capacity — he estimated a ballpark of 200,000 300mm equivalent — will need to be replaced, a trend he sees stretching “next year and into 2009.” Probably something less than that, but still “a significant amount of capacity,” will have to be replaced for DRAM as well, Splinter added.

One analyst pointed out that several DRAM customers (representing abut a fourth of total memory spending) have said their 2008 spending plans will be well below this year’s level, suggesting total 2008 memory capex could be lower by 30% or even 50%, but Splinter explained that memory makers will continue to invest in capacity as long as bit growth keeps going up. “They know that if they don’t their share and their position in the market is going to deteriorate relatively rapidly,” he said. “You can see that already with how the share has shifted around in the various memory players in the last year or two years, with some people used to have close to 20% now are under 10%. They didn’t invest and their share has dramatically dropped off.” Some individual memory firms may indeed cut back investments, he acknowledged, “but when you look at the major companies who represent the other 80%, I think they are still looking at trying to meet a very aggressive bit demand growth.”

“Bit demand growth has been so aggressive in this first half this year,” Splinter said, “and I thought it would initially slow down in the second half… and I don’t see it slowing down.”

In the end, though, the company’s business outlook, and that of the industry in general, “gets down to what is going to happen in foundries,” which are closing in on the 90% utilization rate threshhold that has historically triggered new capacity investments, Splinter acknowledged. A year ago foundry orders kicked in once utilization rates were in the mid-90% range, but Splinter pointed out that over the last couple of process generations, foundries have watched their utilizations tick up to 95%, “and it’s getting I think now maybe close to 100% utilization before they add capacity.”

In the meantime, instead of adding capacity, foundries have found ways to extend productivity in existing lines. Plus, ramp-ups of 65nm processes are taking longer than expected due to “cost crossover” — products that “absolutely need 65nm” are already being made with that process node, Splinter noted, and chipmakers are gauging cost and performance benefits as well as yields before shifting any other products to the next node. As a result, he cautioned that he still expects “very low’ orders and revenue from foundries in Applied’s current quarter (fiscal 4Q07).

Asked by an analyst during the call whether the bearish outlook qualifies 4Q as a “trough,” Splinter responded that it would depend on how strongly seasonal factors like back-to-school and holidays come back — though positive feedback from Taiwan suppliers for computers and consumer products suggests better-than-normal seasonality, he pointed out. “But again, it is going to be whether our customers have confidence to invest going into 2008,” he cautioned. “At this point, we think they should, but I don’t think they are going to decide even until October.”

Looking into its final fiscal quarter of 2007, Applied projects bookings to be flat to down 5% sequentially, with 5%-10% lower sales (roughly $2.30-$2.43B) and EPS falling to $0.26-$0.29, vs. $0.34 in fiscal 3Q.

“Overall, this is going to be our best revenue year of all time,” Splinter said in the call. “If the things that we think are going to happen, we think that the growth is going to be relatively similar to 2007 — but boy, it’s way too early to make a solid projection on that.” — J.M.


Easily post a comment below using your Linkedin, Twitter, Google or Facebook account. Comments won't automatically be posted to your social media accounts unless you select to share.