The government views creation of TMC as the only way to save Taiwan’s DRAM industry, but the existing players see it as an unwanted competitor.  

BY PHILIP LIU

The government is moving ahead with plans to support the establishment of the Taiwan Memory Co. (TMC) as a means of fostering indigenous DRAM (dynamic random access memory) technology and assuring the survival and development of Taiwan’s DRAM industry within an acutely competitive and tumultuous international market.

TMC will have projected capital of NT$60 billion (US$1.8 billion), half of it from the government, and is to be led by a corps of Taiwan semiconductor industry veterans recruited by John Hsuan, honorary vice chairman of United Microelectronics Corp. (UMC), who was brought in by the government to supervise the project. The company plans to establish a team of 800 engineers, half for product design and the other half for manufacturing-process technology, drawn from the island’s ample pool of semiconductor industry talent. It also plans to enter into a technology alliance with Elpida of Japan, the world’s fourth largest DRAM firm, for joint development of next-generation DRAM technology. Plans to cement a similar alliance with Micron of the United States are pending Micron’s obtaining the consent of Nan Ya Technologies, with which it currently has an exclusive contract for technological cooperation.

In an April 4 press release, Elpida announced that it will sell a 10% stake to TMC and capitalize on the cooperation with the new company to gain global market share. As TMC will share Elpida’s R&D costs – the two companies plan to set up an R&D center in Taiwan in 2011 – the cooperation will improve Elpida’s financial position, according to the release.

TMC has enlisted the commitment of a number of leading local IC design houses, including Elite Semiconductor Memory Technology and Etron Technology, to join its camp. They will contribute to TMC’s goal of developing mobile DRAM and other niche DRAM products, enabling it to steer clear of the highly competitive standard DRAM sector dominated by Korean firms, notably Samsung, the world’s largest DRAM maker.

TMC expects to be formally established in the third quarter of next year and start to do business in the fourth quarter. By outsourcing to contract manufacturers, it will seek to capitalize on Taiwan’s large DRAM-production capacity, which accounts for 40% of the global total (including 40% of the worldwide capacity for manufacturing advanced 12-inch DRAM wafers).

TMC’s formation is seen as filling the technological vacuum in the local DRAM industry, permitting it to shed the existing near-total reliance on foreign owners of patented technology. Taiwan manufacturers have been paying some NT$10-20 billion (US$303 million to $606 million) in royalties annually, equal to some 7-10% of their revenues. In the eyes of government policy-makers, TMC will therefore be critical to the long-term viability of the local DRAM industry.
For Taiwan’s six existing DRAM makers, however, the way the government is proceeding with the TMC project has come as a major disappointment. They had anticipated receiving a massive injection of government funds to help them weather their serious financial plight, but instead fear that they will be gaining a strong local competitor.

The industry’s difficulties derived from over-supply in the global DRAM market starting from the first quarter of 2007. Those problems were then aggravated by the impact of the global financial tsunami, which prompted Korean DRAM makers to slash prices to bolster business – aided by the sharp devaluation of the Korean won. Global DRAM prices have plunged 85% over the past two years, hitting a level below the cost of labor and materials and bringing nearly all makers to the brink of insolvency. Perhaps the only exception has been Samsung, due mainly to its diversified operations. Qimonda of Germany, originally the world’s fifth largest firm, went bankrupt in early 2009.

Burdened with huge excess capacity, Taiwan’s four largest DRAM makers incurred a staggering combined loss of NT$152 billion (US$4.6 billion) in 2008, including a NT$57.5 billion (US$1.7 billion) deficit for Powerchip, NT$36.7 billion (US$1.1 billion) for Nan Ya Technologies, NT$21.8 billion (US$660 million) for Inotera Memories, and NT$36 billion (US$1.1 billion) for ProMos Technologies. Most of these companies are still able to keep their heads above water only because of bailout support from the government.

Financial reprieve

On February 16, a consortium of eight government-run banks agreed to extend a fresh loan of NT$3 billion (US$90 million) to ProMos, helping it meet redemption demands on NT$11 billion (US$333 million) worth of ECBs (euro-dollar convertible bonds) on concessionary terms and thereby remain solvent. This was the second time the banks had come to the rescue of the beleaguered firm, after agreeing to extend a repayment deadline for NT$50 billion (US$1.5 billion) in loans last year.

Powerchip, the island’s largest DRAM maker, has also obtained agreement from domestic banks to extend the repayment deadline for its loans, which exceed NT$60 billion (US$1.8 billion). Together, the island’s six DRAM manufacturers owe NT$400 billion (US$12.1 billion) to local banks, mainly for funding their expansion two years ago. (Besides Powerchip, ProMos, Nan Ya, and Inotera, the other two players are Winbond Electronics and Rexchip Electronics.)
The domestic DRAM makers therefore urgently called on the government to come to their rescue, mainly by inducing industry consolidation and the injection of fresh funds, so as to ensure the survival of an industry with some NT$800 billion (US$24.2 billion) in investment and employment of 20,000 workers. Along with the foundry sector, it is one of the two major pillars of Taiwan’s semiconductor industry. Besides the potential impact of the industry’s failure on employment and financial markets, the companies also cite the risk that a Korean monopoly of global DRAM production would lead to much higher DRAM prices, impeding development of the local electronics industry, a mainstay of the island’s economy.

“Should Taiwanese DRAM makers bite the dust, U.S. and Japanese DRAM firms are likely to follow suit, due to the loss of the competitive contract-manufacturing service [that Taiwan provides], paving the way for Korean firms to monopolize the global market,” warns Frank Huang, chairman of Powerchip.

In the first quarter of 2009, Korean makers commanded almost half of the world market, including a 26.6% share for Samsung and 22.3% for Hynix, according to DRAMeXchange, an industry research body. Although Taiwan companies held only a 17.5% market share, the other two leading makers – Micron and Elpida, with shares of 15.3% and 12.1% respectively – rely heavily on Taiwanese manufacturing capacities for contract production. 

In view of the serious current plight of the DRAM industry, Frank Huang argues that it makes no sense to set up another Taiwanese DRAM company. “At a time when [other] governments worldwide are busy bailing out their beleaguered enterprises, the Taiwanese government is engaging in a battle with the nation’s enterprises,” he says. He calls the government a “vulture,” accusing it of hoping to buy up existing DRAM firms at rock-bottom prices in two years after they go bankrupt.

A ranking official at the Formosa Plastics Group, the parent of Nan Ya Technologies and Inotera, is equally suspicious of the government’s intentions. He blames the authorities for “opportunism,” saying they intend to reduce all the existing DRAM producers to the status of contract manufacturers for TMC. 

At an April 1 press conference, John Hsuan responded to the criticism by describing TMC’s purpose as the restructuring of the DRAM industry, rather than spurring industrial consolidation or providing a bailout. “The goal is to establish indigenous technologies, narrowing the technological gap with archrival Samsung,” he told the media, adding that “this is the only way to assure the survival of the DRAM industry” in Taiwan. He was echoing remarks made by President Ma Ying-jeou on March 18 at an industry forum sponsored by Merrill Lynch in Taipei, in which Ma said the aim of the TMC project is “to rescue the industry, rather than individual enterprises.”

Hsuan estimated that the government will eventually have to sink some NT$30 billion (US$909 million) – half of the projected paid-in capital – into the TMC project, with the remainder coming from the private sector. He pledged to bear the responsibility and close the company if it incurs losses of more than NT$15 billion (US$455 million). 

Spending money to try to rescue the existing firms would only postpone the inevitable, and promoting consolidation among the current players would lead to over-capacity, a serious liability during a period of market over-supply, says Hsuan. In addition, consolidation would pose the complicated challenge of integrating different corporate cultures.

Using a “fabless” model

But Hsuan pledged that TMC would not compete against the existing local DRAM makers by building up its own production capacity, but rather operate as a “fabless”   semiconductor company, relying on the other firms for manufacturing. The business model would be to seek technology licensing from foreign partners on optimal terms, complemented by technologies developed in-house, with a focus on DRAMs designed for innovative applications, notably mobile DRAM. The products could be marketed either under its own brand or those of foreign partners.
Hsuan said TMC will produce a business plan capable of convincing private investors of its profit potential, adding that the project has already aroused the interest of many potential investors, including private equity firms, IC design houses, and companies from the semiconductor module and assembly, mobile phone, and GPS (global positioning systems) industries.

Responses to the TMC plan have been mixed. On April 24, the budget center of the Legislative Yuan released a report disputing the ability of the project to upgrade the local industry or, in the absence of consolidation, to improve its structure. The center therefore suggested that the government’s National Development Fund refrain from investing in the project, as the Fund’s statutory mandate is to invest for the purpose of promoting industrial upgrading or structural improvement.

Many industry specialists, though, have applauded the project. At the Merrill Lynch forum, for instance, Daniel Heyler, chief semiconductor industry analyst for Merrill Lynch, noted that the policy of “rescuing the industry, rather than individual enterprises” is correct, since it can boost industrial competitiveness and improve the gross margin of the industry. A shakeout of the local DRAM industry during the current crisis, said Heyler, could speed up the industry’s restructuring. He also noted the difficulty of industrial consolidation, citing the examples of two mergers that took place in 1997 during the Asian financial storm: the creation of Hynix through the merger of Hyundai Electric and LG Semicon and of Elpida by comining the semiconductor arms of Hitachi, NEC, and Mitsubishi. Both transformations took five years to complete. 

Topology Research Institute of Taiwan also supports the project, saying that it can provide the critical components missing in the local DRAM industrial network, namely indigenous technology and branded marketing, without which the industry will be unable to compete with Samsung.
The existing DRAM makers are aware that they will have to rely on themselves to weather the storm. To cover its staggering NT$35 billion (US$1.1 billion) loss in 2008, the board of directors of Nan Ya Technologies resolved on April 17 to carry out a capital reduction of NT$31.1 billion (US$942 million), cutting its paid-in capital by 66.4% before issuing 4 billion new shares at a par value of NT$10 per share for a capital increment worth NT$40 billion (US$1.2 billion). The move will jack up the company’s book value to over NT$10 per share, from NT$5 in the first quarter, thereby greatly strengthening its financial position. Parent FPG has pledged its full support for the initiative. The fresh funds will be used to finance conversion of the company’s manufacturing technology – with assistance from Micron – from the trench process to the stack process needed to upgrade to the sub-50-nanometer level.

At Powerchip, on the other hand, Frank Huang hinges his company’s hopes for survival on a market upturn. In a March 25 press conference, he predicted a shortage of supply in the third quarter, thanks to inventory depletion and the effects of production cuts by major makers worldwide. As a result of those cuts, global output in March already came to less than 580 million chips, far below the demand of 820 million.

Huang says he remains unshaken in his confidence in his company’s future. “Since first getting involved in the DRAM sector, I have seen the price plunge from US$6 per unit to US$0.6 and then soar to back to US$8,” says Huang. As long as the price can remain above US$2 per unit, over the production cost of US$1.5, the industry can survive, he says.

Continuing to express doubts about TMC’s viability, he says the new company will have major difficulty obtaining manufacturing capacity in the wake of a market upturn and that technological strength, without manufacturing capability and sound finances, is not enough to prosper in the DRAM market. He cites the inability of TI, IBM, Toshiba, and Hitachi to succeed in the DRAM business, despite their technological prowess.

In reply, John Hsuan says that if a market uptick makes it impossible for TMC to obtain manufacturing capacity from other makers, it will instead focus on product design. Should existing local makers out-perform TMC even in the design sector, he says, then TMC should fold up, as it will have lost its raison d’etre.