CSMC launches pre-marketing

Chinese foundry hopes earnings consistency will win over investors despite poor market conditions.

CSMC Technologies began pre-marketing an IPO on Monday May10 via lead manager Citigroup, with DBS and UBS as co-leads. The all new share deal is scheduled to price towards the beginning of June, raising up to $100 million including the greenshoe.

Because the company has fixed the exact amount it hopes to raise from the flotation, the percentage of share capital it issues will be subject to the valuation it can achieve. Fund managers say preliminary reports suggest a final freefloat of about 27% and a price to book valuation of roughly 1.4 to 1.6 times 2004 earnings. At the bottom end of the scale, it would price in line with Chartered Semiconductor.

Pre-marketing valuation ranges are always pitched fairly wide to facilitate price discovery among investors. But in this instance, the heightened volatility of the foundry sector and China related stocks necessitates more flexibility than usual. Movements in the freefloat and valuation are, therefore, likely to be closely aligned to further sharp corrections, or an upswing in the foundry sector.

CSMC has a number of hurdles to overcome, not least of which is the negative overhang of SMIC's recent IPO and poor trading pattern of global foundry stocks such as TSMC, UMC and Chartered Semiconductor. All three have fallen about 15% to 18% over the past three weeks despite strong earnings results and widespread buy recommendations across the analysts' community.

Yet CSMC is only a small deal and if it can uncover a core group of investors who believe in the story, it could end up giving the whole China market a much needed boost. A positive reception to a second Chinese foundry stock would go someway towards banishing the ghost of the first: SMIC, whose $1.8 billion IPO of mid-March, heralded the most recent collapse in sentiment towards China stocks.

It would also pave the way for a third listing by a Chinese foundry, since ASMC is still on schedule to launch a roughly $150 million IPO via BOCI and Goldman Sachs towards the end of June/beginning of July. Specialists say ASMC and CSMC have broadly similar business models, although the former is currently larger in terms of output, producing about 45,000 wafers during 2003 compared to CSMC's 26,000.

Specialists say one of CSMC's chief tasks during marketing will be to disassociate itself from SMIC, against which it will almost certainly be benchmarked even though the two companies operate at opposite ends of the foundry market. SMIC's guiding strategy is to break the monopoly of the three global foundries and it operates leading edge technologies for a global roster of clients.

CSMC, on the other, hand specialises in trailing edge, otherwise known as mature technologies and produces chips for a predominantly local client base of fabless IC design houses. Most of its wafers are used in "white box goods" such as refrigerators and microwaves.

Because of their different strategies, specialists say SMIC and CSMC have very different earnings cycles and patterns. "The global foundries are obviously the best starting point to value CSMC since they do operate in the same space," says one. "But they're proxies at best. Accounts will adjust their valuation taking into account CSMC's different growth pattern, before applying a discount because of its size."

Post IPO, CSMC is likely to be at best a $370 million cap stock versus $39 billion for a global giant like TSMC. CSMC is currently 65% owned by CSMC Cayman, a company, which is in turn owned by China Resources Logic and CSMC management. Chartered Semiconductor owns 10% and the nature of its ongoing relationship with CSMC will be another key determinant of the IPO's success.

Under an agreement forged last summer, Chartered is providing CSMC with equipment and technology transfers in return for a mixture of stock and cash. It has also been touted as a possible technology partner for CSMC's second prospective 8" fab.

The company currently operates one fab based in Wuxi, which produced about 14,000 5" wafers in 2003 and 12,000 6" wafers. It specialises in process geometries of 0.35 to 0.5 micron, compared to the more sophisticated geometries of 0.13 micron and below adopted by TSMC and co.

One of its key selling points of the deal is that the company has not been sucked into the highly capital-intensive capex cycles of the big foundries and consequently has a far steadier earnings stream. The global foundries can see dramatic peaks and troughs in average selling prices, earnings and stock valuations, because of the huge costs of developing new technologies and supply/demand imbalances created by sudden gluts of new capacity.

Yet whereas an 8" fab can cost upwards of $1.5 billion, a 6" fab can cost as little as $300 million since all the equipment is re-cycled from bigger foundries as they move onto the next stage of technological development. Specialists say trailing edge foundries like CSMC stand or fall by their ability to transfer and re-install equipment. There are currently about a dozen such foundries with plans in China, but only a couple have been able to source enough capital or find the right technology partners to succeed.

CSMC started life as a state-owned IC enterprise in the 1980's and was originally known as Huajing Electronics. In the 1998 it formed a foundry JV with Taiwanese expertise. Its current CEO Peter Chen formerly worked for chipmaker Mosel Vitelic.

Since 1998, CSMC has been continually profitable, whereas its technology partner Chartered Semiconductor has only just turned a profit after 12 consecutive quarters of losses. Specialists say CSMC will be a less attractive play during a cyclical upswing as its margin expansion will be lower. However, in bear market conditions, investors may feel more attracted to its consistency.

Fund managers say syndicate research shows net margins of 13% in 2001, 25% in 2002, 10% in 2003 and 13% projected for 2004. By contrast, Chartered has swung from minus 52% in 2003 to 4% forecast in 2004 and 6% forecast in 2005. At the other end of the scale, TSMC's net margins are forecast to top 35% this year.

However, while CSMC's margins have been relatively stable, its revenue and net profit growth has been strong. The company recorded revenues of $42 million in 2003 and this is forecast to double to $85 million in 2004 and $135 million in 2005.

These forecasts do not take account of plans for an 8" fab and the company will have to show what impact a more aggressive move up the technology chain will have on its margins. However, tech specialists say completion of the 8" fab is dependent on finding the right technology partner.

Over the course of 2004, capacity is forecast to expand from 26,000 to 60,000 wafers via three 6" lines.

In terms of valuation benchmarks, the cheapest comparable is Chartered Semiconductor, which is trading on a price to book valuation of about 1.4 times 2004 earnings. It produced 153,000 chips during 2003, of which about 41% were 0.25 micron and higher.

SMIC, which priced its IPO at about 2.3 times, is now trading at about 1.57 times, whereas UMC has come down to 1.7 times and TSMC has slipped back below three times to trade around the 2.85 level (all UBS estimates).

TSMC had traded up to above four times book just prior to Presidential elections in Taiwan, but has slipped badly since then. Initially, the freefall was driven by fears that a Chen victory would hamper the company's plans to develop foundries on the Mainland. More recently, the prospect of interest rate hikes in the US has prompted heavy selling by foreign funds in particular.

Many analysts expect the stock to spike back upwards, but there is some dissension about how much upside is left to the current cycle. In a recent research report, CSFB argued the current cycle may be peaking.

It warned that inventory build up at major fabless houses posted its sharpest increase since 1998 during the first quarter (37%). It said that unless global demand continues to accelerate during the second half of the year, the industry will run into trouble, since capex will increase by 47% during 2004.

It wrote, "We expect economic growth to slow down in 2H04 and coupled with increasing supply, we believe we are in the late stage of the semiconductor cycle."

Others, however, believe the semiconductor cycle is still coming out of a recovery phase and TSMC management say they expect the cycle to extend to 2006. Where CSMC is concerned, growth is highly dependent on China and while the government is putting the brakes on an overheating economy, the fabless industry is expected to continue growing strongly.

Tech specialists say the industry is showing a CAGR of 55% and is still in its nascent stages of development, with about 350 or so companies of five to 10 employees producing revenues of about $500 million per annum. Six out of the 10 largest houses currently commission CSMC chips.

Most of these companies are still seeking trailing edge technologies and analysts point out that they are cheaper to develop in China than elsewhere. Labour as a percentage of cost of goods sold (COGS), for example, is 14% for mature technologies versus 7% for cutting edge technologies.