|
Looking to the bellwethers: Cymer posts earnings, perceives softening market, reveals nuggets |
|
|
|
Jan 29, 2008 at 09:28 PM |
One of the bellwether companies in the semiconductor equipment arena is Cymer, the market-share monster of the DUV lithography laser-light source sector.
With the stepper/scanner companies as its direct customer companies and the chipmakers as the other customers, the San Diego-based firm has a unique vantage point on both the state of semi manufacturing's most important (and lucrative) process module and the overall health of the fab industry.
The company announced its quarterly and annual earnings on Tuesday, and then elaborated on the details in an analyst conference call. The short version finds 4Q07-vs-3Q07 revenues up ($139,122,000 vs. $132,125,000) though net income was down a few hundred thousand (to $21,437,000), while the annual revenues ($521,696,000) fell below 2006's total ($543,855,000), with net income also down a bit in FY07 ($88,362,000 vs $95,048,000). Expectations for 1Q08 look bleakish, with Cymer forecasting about a 15% hit in revenues because of soft industry conditions (that's a double-digit demand decrease in system sales), although system average selling prices (for those tools the company does manage to sell) should remain robust.
Nonsystems revenues (service, consumables, etc.) shot up to a record 52% of total revenues, and are expected to continue making up a larger chunk of the company's cash flow. One telling indicator of Cymer's business is its tracking of worldwide annualized average light source utilization, better known as pulse usage. The number grew about 20% in 2007, indicating that the fab companies continue to find ways---and Cymer continues to offer solutions---to squeeze more productivity out of those hard-workin' laser modules.
The annual usage rate is now up to some 10 billion pulses per tool, but a breakdown by chipmaker type finds memory users averaging more than 13 billion pulses, foundry just under 10 billion, and logic houses right around 6 billion. Put another way, those 10 billion pulses equate to roughly 27,397,260 pulses per day and a mere 1,141,552 pulses per hour, if the units ran nonstop 24/7/365. The company said it has an installed base of 3250 systems (about 85% of the total number of excimer lasers used in semi production), so if each unit ran at 10 billion pulses per annum, that would amount to 32.25 trillion bursts---downright cosmological numbers.
Cymer revealed some other nuggets during the conference call. Flash memory makers have already started placing purchase orders for EUV litho tools, so Cymer expects to ship its first EUV source by the end of 2008. (ASML has already named the company its EUV source supplier.) It also will send out its first TCZ evaluation systems later this year, as the flat-panel production equipment joint venture with Carl Zeiss starts to garner a wee bit of traction. The initial second-generation double-patterning/immersion source boxes, the XLR600i (targeted for 32-nm half-pitch apps), will get out the door to customers by mid-2008.
One item not mentioned at all in the financials or during the call was the effect (or even occurrence) of the San Diego County fall wildfires on Cymer's Rancho Bernardo HQ, plant, operations and personnel. It was if the fires never happened, as if the company never came perilously close to a direct hit, as if the vast majority of Cymer employees were not evacuated from their homes. A bit surprising perhaps, but apparently since there was no substantive impact on the company's financials from the near-miss disaster, it was not enough to warrant any comment.
In a nice, compact summary of the state of mainstream semiconductor litho, head honcho Bob Akins listed the findings reached by panelists and attendees at Cymer's recent symposium, held at Semicon Japan. "The most significant conclusions reached...were:
- "First, with immersion lithography beginning production for high-volume manufacturing at the 4X-nm node, defects and overlay at this CD are considered resolved or resolvable issues.
- "Second, for the 32-nm half pitch, ArF immersion double-patterning is the most viable solution. When EUV is ready for high-volume manufacturing, however, some double-patterning critical processes could potentially be replaced by less process-intensive EUV.
- "Third, the productivity impact of implementing double-patterning may be compensated for in part by effective process integration an utilization improvements.;
- "And fourth, the key to EUV's success is source development, not only because it will be key to determining tool throughput and hence economic viability, but also because it will enable resist and mask challenges to be faced and resolved."
If Cymer's view of things can be taken as a harbinger (and it's pretty much in sync with industry forecasts), the first half of 2008 will be a challenging business environment for the semi tool sector and some of its customers, especially the DRAM players, but the second half may have a more optimistic flavor.
|
|
Blogs 
