The most interesting data point from Lam Research's (LRCX) fiscal 1Q11 results call may have been the observation that foundries are buying new 300mm equipment for trailing-edge processes (0.15μm-0.10μm) -- the last time that happened, the industry saw a capex surge. Also: Mixed results for CYMI, and an interesting nugget from the EUVL Symposium about EUV source power.
Deconstructing LRCX: The rise of trailing-edge 300mm
The most interesting data point from Lam Research's (LRCX) fiscal 1Q11 results call? Foundries are buying new 300mm equipment for trailing-edge process work (0.15μm-0.10μm), and the reasons behind this trend are worth consideration, says Credit Suisse's Satya Kumar. If the manufacturing economics for 300mm/trailing nodes are now even lower than for 200mm wafers, chipmakers can retire 200mm assets and replace with 300mm equipment -- a trend last seen in 2007 with DRAM, which "drove a strong capex cycle," he notes. Or, the cost-case could be from the other direction, with demand for trailing-edge chips spurred by increasingly broad-based semiconductor usage.
Meanwhile, LRCX blew out its 1Q numbers (EPS $1.52 on $805.9M sales, vs. Street's $1.37 on $793.9M), and is guiding higher for 2Q11 (sales ~$825M vs. Street $811M, shipments $865M vs. $838M, EPS $1.55 vs. $1.42). Execs expect 1H11 to "look a lot like 2H10," as shipments plateau, with some caution in 2011 planning due to PC/consumer softness though broad trends are healthy. Tradition Equity's Peter Wright noted that LRCX management sees tablets at ~40M shipments in 2011 and "largely cannibalistic to netbooks."
And more broadly, LRCX's results and comments help Kumar's argument that, assuming semiconductor capital intensity is only going up, "then we are still in year 1 of a 3 year up cycle." Deutsche Bank's Peter Kim, though, sees growing foundry spending as "looking more and more like a bubble," with LRCX's 4Q projection of >40% shipments to foundry and well beyond previous highs in terms of dollars.
Sizing up CYMI, and an EUV horserace?
Cymer eked ahead of Street estimates in its 3Q10 results ($141.7M sales and EPS $0.70 vs. $137.2M and $0.63), but its outlook for 4Q is low: $140M-$145M and $0.66, vs. $150.8M and $0.72, blamed on lower immersion and EUV units. Positives out of the call: CYMI has shipped 2× more immersion sources in the past two quarters than competitor Gigaphoton; the company should gain share as memory capex wanes slightly (and CYMI indeed thinks it has won new memory biz, perhaps Samsung); and there's increased customer interest in TCZ. Still, Kumar points out, margins are a bit soft, and there's news from the EUVL Symposium that DPP lasers have higher throughput than LPP for EUV (CYMI uses DPP, Ushio has DPP), though both are still <10wph throughput; this is something that could make EUV production a bit more of a horserace, he notes.
XLNX: How to close a 40nm gap
Xilinx has its work cut out for it: rival Altera has enjoyed nearly double the growth in the past two years, largely in part to its higher mix of 40nm process technologies, writes Credit Suisse's John Pitzer. "While XLNX held the lead at the 65nm process node, it conceded the lead at 40nm where it is ~3-4 quarters behind ALTR," he writes. ALTR's 40nm sales mix was 13% in 3Q10, vs. just 5% for XLNX, which translated into $69M vs. $31M. "The challenge for XLNX is to close the gap while ramping a new foundry partner in TSMC," he says.