Under the lead management of Goldman Sachs, ProMos Technologies priced a 300 million new share offering last Friday at a 20% discount to a spot close of NT$21, raising proceeds of $146 million.
The deal was said to have closed marginally oversubscribed, with a fairly even distribution split between the three regions and a book that comprised virtually all new investors to the company. However, no-one doubts that the transaction was an extremely difficult one and many might wonder why the company proceeded at all when the stock price is off 53% from its NT$41.7 February high and trading on a price to book value of only 1.7 times 2002 earnings.
For the lead manager, the main challenge was combating market conditions that deteriorated markedly over the course of roadshows and the fact that rival DRAM manufacturer, Powerchip, had already sucked out investor demand the previous week with its own 350 million share offering via Deutsche Bank at an 11.76% discount to spot.
Comparing the pricing achieved by the two deals is difficult. Were, for example, ProMos to have priced on May 9 at the same time as Powerchip, it would have come at an equivalent 18.9% discount to spot, based on the absolute price investors receive after applying a 20% discount - NT$16.8 - and the price ProMos was trading at on May 9 - NT$20.7.
But as some observers point out, this argument since is flawed since the market got worse rather than better over the week that ProMos roadshowed its deal. For example, while the first week of May saw DRAM spot prices crash right down to their December lows again - 128Mbit SDRAM chips fell back to $2.50 - the second and third weeks have been characterised by contract chip prices following suit.
Experts say that DRAM prices are down 24% since the beginning of May, because manufacturers (and particularly Hynix) have been putting further pressure on spot prices by offloading inventory in the belief that slower than expected growth in the US economy will dampen demand for new chips over the next few quarters. Analysts have consequently responded with a series of sector downgrades, led by UBS Warburg on May 10.
Secondly, it is also the case that Powerchip had an inherent advantage because of its existing GDR. "Investors would have felt more comfortable accepting a slimmer discount when there is an outstanding GDR to hedge against," one banker comments.
Instead, some bankers have compared the all-in cost for the two companies, taking into account stock price movements between the beginning of roadshows and pricing (including the final discount). Under this scenario, ProMos ends up with an 18.8% discount since its stock moved up very slightly during roadshows, while Powerchip ends up with a 40.5% discount, since its stock price moved sharply downwards.
Again, however, the argument is flawed, since most of the most recent DRAM stock price crash occurred as Powerchip was embarking on roadshows towards the end of April. The company also undertook a much lengthier series of presentations spanning 13 days compared to the six undertaken by ProMos.
Both companies are said to have ploughed ahead with their deals because they want to establish their 300mm or 12" fabs ahead of the competition. ProMos leads the Taiwanese sector, since it has already begun to produce new chips from its 300mm fab and will use proceeds to secure additional equipment for the plant.
Analysts estimate that the capacity of a 300mm fab should be about 2.5 times that of a 150mm fab and production costs about 30% lower, once yields fully ramp up. Larger chip sizes and smaller process geometries (0.13 micron) will increase output and reduce costs.
Yet some analysts argue that there is still a question mark over the return DRAM manufacturers will make on their 300mm investments. Partly this is because of the huge costs of the fabs plus their associated equipment. But mainly it is a result of witnessing short market upturns and wondering how demand will be able to match the huge increase in supply which the 300mm fabs are capable of.
Back in 1982, when the industry made the jump from 50mm wafers to 100mm, the capital outlay for a new Fab amounted to $150 million. By 1989, when the industry made the jump from 150mm to 200mm, the cost was about $400 million. The latest jump, which has been 11 years in the making, will bring in 300mm wafers at a cost of $2.5 billion per fab.
Alongside the lead, there were five co-managers on the 30 million GDS deal numbering ABN AMRO, Bear Stearns, Chinatrust, Nomura and Sun Fund.