Blogvesting

Cymer is the leading company in the semiconductor photolithography laser industry, with the technological lead and a market share of close to 50%. Its light sources are installed in more than 3250 fab factories worldwide. Cymer sells light sources to ASML, Canon and Nikon, who incorporate the light sources into their wafer steppers and scanners, which are in turn sold to major chipmakers. Cymer also directly maintains and services its light sources at chipmaker fab plants, and its major customers include Intel, AMD, Samsung, TI, NEC, Fujitsu and others. The only competition for Cymer is Gigaphoton, a joint venture between the Japanese firms Komatsu and Ushio started in 2000. Gigaphoton claims to have an installed base of more than 500.

The manufacture of today’s chips requires DUV (deep ultraviolet) lasers which emit light pulses with precise intensity, wavelength and duration, and can do so hundreds of thousands of times without variation. Cymer pioneered the shift from KrF lasers (248 nm wavelength) to ArF (193 nm wavelength) years ago, allowing the production of 65 nm and 45 nm chips. (The current newest generation of CPUs uses 45 nm chips made mainly using ArF lasers, with KrF lasers for non-critical layers.) A few years ago, Gigaphoton successfully reverse-engineered the process, and began to manufacture ArF lasers as well. Gigaphoton has always focused its sales in Asian chip foundries, thus investors were shocked in early 2008 when Intel revealed that it has switched over its purchases of all KrF lasers from Cymer to Gigaphoton. (This is only natural, as businesses hate to rely on a single supplier, and where possible will try and help a second supplier grow.) This represents the first time that Gigaphoton has ventured into the US market, and analysts predicted market share erosion for Cymer, leading to a sharp drop in Cymer stock.

Cymer’s current ArF lasers, combined with immersion lithography technology (where water is inserted between the final lens and the wafer to compress the light pulse), can already produce 32 nm chips. Intel has officially committed to a 2 year cadence in chip production, moving from 65 nm in 2005, to 45 nm in 2007, to 32 nm in 2009. Further advances to 25 nm chips will require new lasers utilizing EUV (extreme ultraviolet), which Cymer has been working on since 1997 and is scheduled for production in 2011. Gigaphoton has recently ventured into immersion and double patterning technology using its own ArF laser, and has some success in selling its ArF lasers in Asia (although Intel has committed to buying only KrF lasers from Gigaphoton). Gigaphoton is believed not to possess EUV technology at present, and there is some doubt as to whether its parent companies will be willing to plough cash into researching this technology. Therefore, in the near-term, Gigaphoton may prove a viable threat to Cymer, but in the long-term, Cymer holds a technological lead.

In 2007, sales of new light sources (mainly ArF immersion) brought in some $270M in revenues. Replacement parts for its 3000+ installed sources brought in another $197.5M, while service and maintenance brought in $48.8M, for a total of approximately $517M in revenues, and an EPS of $2.50. 2008 is shaping up to be a down year for Cymer, which faces a combination of a weaker semiconductor industry combined with fiercer competition from Gigaphoton. For the 2 quarters of 2008, revenues totaled $248M, and net income came in at $27M, or $0.88 per share. Analysts are expecting full-year net income of $1.80-$1.95 EPS, which gives $18 share price if one uses a PE multiple of 10. As the company currently has $240M excess cash on hand, that adds another $8 per share, giving a share price of $26. Note that this multiple is very low for a company of this quality, and that the base earnings are in many ways also unusually depressed.

However, the semiconductor cycle is reaching the end of the current cycle, and the next round of chips will cause a spurt of spending. Already, Cymer is reporting a pickup in sales of its latest XLR ArF immersion system, and Cymer holds a commanding market share at the higher end of the market (something like 90% market share for the ArF immersion systems), and its ArF immersion double patterning systems are also ready to ship. I expect 2009 to be a breakout year for Cymer. Intel certainly will be chafing to administer the knockout blow to AMD by moving rapidly to 32 nm chips. I believe that Cymer has minimal downside since approximately half of its revenues are derived from its installed base, and there is a good chance for upside as the semiconductor companies move towards their next generation chips, which will require the more advanced lasers, an area where Cymer has a good reputation and track record. I expect this to be similar to the Intel-AMD situation, where Gigaphoton takes the less profitable lower end of the market, and Cymer continues to capture larger profit margins from the higher end of the market.