Notwithstanding a substantial initiative several years ago to develop a chip-manufacturing infrastructure, the Chinese semiconductor industry is still small, consisting largely of foundries and joint ventures, which rely on offshore customers or partners to design the chips and set up the factories. Only one company generates more than $1 billion in revenue, and industry sales are only a few billion dollars, a small percentage of world sales. Standard & Poor's Ratings Services believes volumes are unlikely to grow dramatically over the near term, and current volumes are not sufficient to significantly affect overall market conditions, or affect the credit quality of most other industry participants. Over the intermediate term, Chinese manufacturers are more likely to produce chips for the local electronics market and for global
consumer goods, and enter the leading edge global markets over the long term.
An Attractive Market
China is now making efforts to expand its chip design skills, much as it had earlier encouraged the development of a chip-manufacturing infrastructure. An expanded intellectual property base, coupled with the continually-expanding internal demand for electronic products, and economic incentives offered to
local manufacturers, could lead to growth of the local chip industry in coming years. Still, it likely will take many years for China to bring its chip design skills to international levels--a key step, before China can become a major force in the global industry.
There are good business reasons for global companies to expand their Chinese presence. While only a small percentage of the world's semiconductors are manufactured in China, a significant part of the chip assembly and packaging industry--and a substantial portion of the electronics manufacturing services
(EMS) industry--are located in China: Therefore, a large percentage of the world's semiconductors pass through China at some point in the manufacturing process, either to be packaged into finished chips, or to be combined with other components in a phone, computer, television, or other electronic products. Local chip manufacturing could shorten distribution chains for many electronic products companies, and anticipated lower manufacturing costs and shorter working capital cycles could make these products even more competitive.
Perhaps the most intriguing long-term event would be a possible foray by Chinese suppliers into the world markets. This initially could take the form of Chinese semiconductors in expo rted Chinese branded electronic products. Subsequently, Chinese companies could design and make chips for a range of
applications, most likely for the rapidly evolving consumer and communications markets, which could be sold into the global electronics markets over the longer term. Credit implications for rated semiconductor companies could then become more significant, depending on how aggressively Chinese chip companies
enter the market.
China's initial efforts to develop a chip-manufacturing base began in 1996
Historically, the business risks of developing a wafer processing business for startup companies (like many of those in China) has been mitigated by their participation in joint ventures, where an offshore chip company supplies trained engineers, manufacturing processes and plant operating software. The offshore company buys the factory's output at cost--presumably less than its own direct costs, and trains the host company's engineers and production workers. At the same time, the offshore company must continue to protect its key intellectual property regarding products and processes, which is not ordinarily shared with the venture. The offshore company commonly gains a favorable local reputation, and may also gain preferential entry into the local market. Host-country governmental initiatives and subsidized construction
loans--well in advance of production--generally reduce the venture's financial risk. Still, the joint venture's revenues may lag capital investments by several years, and cumulative cash flow breakeven can take far longer.
Over time, employees of the ventures may migrate to other chip companies or teach at technical colleges. Eventually, a standalone industry can be established, as in Taiwan. China's industry could well follow a similar path.
Responding to governmental initiatives in the ninth five-year plan (1996-2001), HuaHong Group established a commodity DRAM factory with NEC Corp. (BBB/Stable/A-2) near Shanghai in 1997; 71% of the initial capital was supplied by HuaHong. In the next few years, additional Chinese chip companies and projects were announced, including Grace Semiconductor Manufacturing Corp., Advanced Semiconductor Manufacturing Corp. Ltd., Central Semiconductor Manufacturing Corp. and Shanghai Belling Co. Ltd. Many other factories were announced but could not secure financing and never took root. By 2005, only
Semiconductor Manufacturing International Corp. (SMIC) had revenues above $1 billion, and China's top five chip operations combined--all either foundries or memory joint ventures--had sales of only $2.2 billion, about 1% of global industry sales (see table 1).
| Table 1 | |
| Top 5 Foundries In China | |
| 2005 revenues (Mil. US$) | |
| Semiconductor Manufactuirng International Corp. | 1,459.0 |
| HuaHong NEC IC Design Co. Ltd. | 300.0 |
| HeJian Technology Corp. | 270.0 |
| Shougan NEC Electronics Co. Ltd. | 137.0 |
| Advanced Semiconductor Manufacturing Corp. Ltd. | 112.0 |
| Source: China Semiconductor Industry Association. | |
Additionally, the two large Taiwanese foundries--Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC, unrated) and United Microelectronics Corp. (UMC, BBB+/Stable/--), and some other global companies--own front-end wafer processing operations in China, the operations of which, had they been booked as revenues, could represent an additional few billion dollars of operations in China, although those operations are controlled by their offshore owners, not by Chinese entities.
Having established a base of both front- and back-end manufacturing operations in prior five-year plans, China launched initiatives to build a chip design industry in the tenth five-year plan (2001-2006). At the same time, Chinese companies, from small consumer electronics makers to large advanced communications companies such as ZTE Corp. or Huawei Technologies Co. Ltd., became volume users of semiconductors (mostly imported). According to the China Semiconductor Industry Association, about $4 billion of locally produced semiconductors were used in Chinese electronics products in 2005, although
this probably mostly refers to offshore-designed products made in foundries within China.
SMIC, which had purchased Motorola Inc.'s (A-/Stable/A-1) factory in Tianjin, was able to leverage the factory's existing core of trained employees and processes. Responding to stimuli in the government's ninth five-year plan, the company invested $2 billion in capacity expansion in 2004, including equipment
and facilities--a very large sum for a company of its size. Funding largely was provided by SMIC's $1.8 billion IPO. Industrywide equipment sales to all Chinese companies were $2 billion in 2004, double the 2003 level.
But following an industry down cycle, SMIC halved its spending to $1 billion in 2005; total Chinese industry spending was halved, as well. Still, SMIC's free cash flows were strongly negative in the period. The company's spending remains depressed, at $200 million in the March 2006 quarter, and only $168 million in the June quarter, although the company still plans to spend $1 billion for the full year. At these reduced levels, the company has achieved breakeven free cash flows, and plans to "expand in a financially disciplined manner." In March 2006, unprofitable Grace Semiconductor Manufacturing Corp.
again deferred its IPO, and reportedly sought to delay repayment on an estimated $680 million in loans. While SMIC generated revenues of about $900 per wafer in the June quarter, Grace generated about $600 per wafer (February 2006 data); by contrast, Taiwan Semiconductor Manufacturing Co. generated $1,450 per wafer in the June quarter.
Also reflecting current conditions, a Hynix Semiconductor Inc. (B+/Positive/--)- STMicroelectronics N.V. (A-/Stable/A-2) joint venture's planned 300 mm plant has apparently slipped into 2007, from 2006. Also, in November 2006, HuaHong NEC deferred completion of its 300 mm plant (also originally slated to be in operation before year-end), reportedly for lack of funding, and will instead expand its 200 mm capacity. The 300 mm fab had been part of a plan to become a vertically integrated chip company participating in the world market.
To date, only a limited number of chips have been designed in China. The top 10 design houses had sales of about $650 million in 2005, and the largest had sales of about $157 million (see table 2). Shanghai Fudan University estimates that there are only about 500 semiconductor design degrees awarded per year
in China; this low level could well be an impediment to industry growth over the intermediate term. However, as more engineers are graduated, the Chinese semiconductor content of locally designed electronic products will undoubtedly increase.
| Table 2 | |
| China's Top 10 Chip Designers | |
| 2005 Sales (Mil. US$) | |
| Datang Microelectronics Technology Co. Ltd. | 71 |
| Hangzhou Silan Microelectronics Co. Ltd. | 75 |
| Zuhai Actions Semiconducrto Co. Ltd. | 157 |
| China Huada IC Design Center | 46 |
| Beijing Vimicro Semiconductor Co. Ltd. | 96 |
| Shaoxing Silicore Co. Ltd. | 29 |
| Hangzhou Youwang Electronics Co. Ltd. | 75 |
| Wuxi China Resources Semico Co. Ltd. | 28 |
| Shanghai HuaHong IC Desogn Co. Ltd. | 46 |
| Tongfang Microelectronics | 29 |
| Source: China Semiconduction Industry Association. | |
Along with its burgeoning semiconductor design skills, China has established unique technology standards developed by Chinese companies. Whatever the relative technical merits of the various international and local standards may be, the establishment of local standards means that foreign companies will likely be paying royalties in order to sell their products in China, while any royalties paid by a Chinese manufacturer to a Chinese technology developer will at least not leave the country. High DVD royalties, for example, may have led to the prominence of the Video CD (VCD) format in China, and China is working on a digital television standard that differs from the three existing global standards. Although developed for the China market, TD-SCDMA has been ratified as an international standard. To the extent that TD-SCDMA is
implemented within China, it is likely that Chinese companies may have either a head start or become favored suppliers. Also, to the extent that TD-SCDMA is implemented outside of China, Chinese companies would likely earn royalties, as well as recognition, from global companies. Currently, China's 3G licenses have not been issued, and the technology remains unproven.
China supports the open-source Linux operating system, rather than Microsoft's Windows. In 2004, a disagreement between advocates of the international WiFi (802.11) wireless local area network standard, and the Chinese-developed WAPI (wireless authentication and privacy infrastructure) standard, nearly led
Intel Corp. (A+/Stable/A-1+) and Broadcom Inc. to withdraw their wireless LAN products, such as Centrinobased laptops, from the Chinese market, rather than share their intellectual property with Chinese companies the API standard would have required.
While royalty expense is an important issue, the ability to develop and control technology has additional benefits. When any company establishes the standards for a future electronic product, it earns a significant degree of global recognition, in addition to generating royalties for many years.
In an effort to further stimulate the domestic industry notwithstanding industry cycles, the Chinese government's 11th five-year plan, announced in 2005 for 2006-2011, included plans to invest an additional 300 billion yuan ($35 billion) to develop 15 large chip design companies, plus 10 200-mm and five 300-mm production lines. These production lines appear to be targeted towards producing leading edge chips. If these very ambitious plans come fully to fruition, China would have the capacity to meet a substantial portion of its internal demand for semiconductors, and the capacity to compete in the global markets, as well. China's foundries could play a significant role in this area, because these companies have had access to some level of global technology skills for several years.
It is likely that Chinese electronics companies would be the first customers for products from Chinese chip companies, as global electronics companies would likely be very cautious in relying on unproven suppliers. It is not clear that China's electronics companies can absorb the potential production volume of these planned 15 semiconductor factories, and the economic viability of these plans appears to be uncertain.
It remains to be seen how China will obtain the large quantities of capital equipment needed for these factories, as China does not have a native semiconductor capital goods industry (although the 11th fiveyear plan intends to look into this matter, as well). U.S. Commerce Department export regulations
generally allow sales only of two-generation-old equipment, unless a specific export license is granted. European and Japanese equipment suppliers' exports to China are only somewhat impeded by the "Wassenaar Arrangement," a U.S.-led voluntary multinational agreement that seeks to constrain the spread of several technologies, including semiconductor technologies, which could be used for military purposes. SMIC may have obtained its U.S.-sourced manufacturing equipment because it technically is a Cayman Islands company, listed on the New York Stock Exchange, although it appears to remain Chinese-controlled; its chairman is a government official. To the extent that these equipment export restrictions are relaxed, global semiconductor capital goods suppliers would benefit from the planned $35 billion in equipment purchases. Leading global equipment suppliers, significantly Applied Materials Inc. (A-
/Stable/--), Tokyo Electron Ltd. (BBB/Stable/--), ASM Lithography, Nikon Corp. (BB+/Positive/--), Canon Inc. (AA/Stable/A-1+), Lam Research Corp. (BB- Stable/--), and others likely would be among the leading suppliers to China (see table 3).
| Table 3 | |
| Capital Equipment Sales To China, 2004 | |
| (Est. sales, Mil. US$) | |
| Applied Materials Inc. | 610 |
| Tokyo Electron Corp. | 320 |
| ASML Holding NV | 290 |
| Lam Research Corp. | 130 |
| ASM International NV | 125 |
| Niikon Precision Equipment Co. | 110 |
| Novellus Systems Inc. | 90 |
| Advantest Corp. | 60 |
| KLA-Tencor Corp. | 50 |
| Dainippon Screen Manufacturing Co. Ltd. | 40 |
| Source: Gartner Dataquest. | |
A less-ambitious alternative could meet local needs
Rather than attempting to develop a world-class semiconductor infrastructure, a less costly and more cautious approach could be the further expansion of existing technology partnerships, presuming that the foreign partners have sufficient incentives to cooperate. The next most likely course of action would be for China to develop a mainstream semiconductor industry, rather than seeking to go head-to-head against industry leaders at this time; Chinese electronics companies would be a natural initial customer base for a Chinese chip industry. Such an industry could use older equipment, readily available at moderate price levels, and which would not run afoul of Wassenaar.
Again, Chinese-made chips likely would first find a market in Chinese-branded electronic goods. Once Chinese chip manufacturers establish consistent volumes, costs, support, and reliability levels, these chips could be sold to global electronics companies in volume. It remains to be seen whether China follows this path, how rapidly such businesses would enter the world market, and how aggressively they would compete against established global mainstream chip suppliers.
The global response
Global suppliers are also looking towards expansion in China to achieve:
. Proximity to the closely related chip packaging and testing industry;
. Proximity to the expanding presence of global EMS companies in China;
. Manufacturing cost advantages, including tax holidays and land subsidies;
. Proximity to China's increasingly affluent electronics-consuming populace; and
. Proximity to China's growing branded electronics manufacturing industry.
Several global chip companies already have chip operations or ventures in China. Currently, Taiwanese foundries Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC) and United Microelectronics Corp. operate 200-mm wafer foundries in China. A Hynix Semiconductor Inc.-STMicroelectronics NV joint venture manufactures memory chips on 200mm wafers in Wuxi, China. Any of these operations could expand over time.
Global suppliers could use their well-developed processes in China, and can bring in engineering-level employees for extended on-site assignments as they train local workers. However, to date, global companies have carefully protected their manufacturing process know-how, and are likely to continue
doing so, until they receive adequate assurance of China's intellectual property protection practices, which substantially lag Western standards.
Credit Implications Will Depend On Leading Companies' Next Steps
Over the near term, China's production levels are not large enough to materially affect global supply levels. Assuming that all product reliability and performance concerns are satisfied--no small accomplishment--and if Chinese suppliers aggressively ramp up mainstream capacity, they could create substantial price pressures on mainstream suppliers of commodity chips. However, this is a longer-term scenario, and still leaves most other global sectors unaffected.
If global chip companies expand their manufacturing capacity in China--which will require substantially increased comfort regarding intellectual property protection-- they stand to gain both from sales to Chinese customers as well as reduced manufacturing costs, while the underlying capital expenditures for Chinese expansion might be lower than equivalent expansion in more-developed areas. Assuming that the overall industry's expansion is proportionate to the rise in demand (or that global companies shutter enough higher-cost plants to prevent a capacity glut, it could be favorable for those companies' sales, profitability, and ultimately, for credit quality. Again, those decisions appear to be several years in the future; in the meantime, we do not expect any significant credit implications.
Sidebar: What It Takes To Manufacture Semiconductors
Semiconductor manufacturing involves three broad groups of steps: design, wafer processing (front end), and assembly and testing (back end). Chip design requires a core of university-trained engineers, plus workstations, software, and a customer base of electronics companies who want their product ideas to be converted into a set of chip designs. Chip design is largely an intellectual-property intensive business, with relatively low capital costs. Design entails the following main steps:
. Functional requirements for an electronic product are decided.
. System performance requirements, timing considerations, and interaction with other devices are established
. Engineers establish clusters of functional requirements.
. Requirements are divided among a group of chips, reflecting manufacturing process compatibility of the various functions (memory, processing, interface, etc.) placed on each chip.
. Chip functional requirements are combined with information about connections to other devices, temperature and voltage stability, electronic noise immunity and other design parameters, to determine the specific layout of each chip. Where possible, functional building blocks are used to speed the design process.
. Each chip is laid out by automated software, under control of an engineer.
. The layouts are translated into a set of masks (essentially, photographic patterns for each layer of the process). Masks are designed from the layout, and delivered to the wafer fabrication plant.
Front end
A leading-edge wafer fabrication plant (fab) is a highly automated factory meeting exacting safety and environmental regulations. Manufacturing is highly capital intensive and requires a substantial base of intellectual property, and a range of labor skills up through postgraduate engineers, resulting in high business and financial risks. Well-established chip manufacturing companies ordinarily spend around 20% of revenues on capital expenditures, averaged over the business cycle. Additionally, there are lower limits
on the capital spending required to establish even a small, trailing edge factory. Hence it is no surprise that China's capital expenditures in the past several years were well above the sustainable level.
Back end
Semiconductor back-end processing is moderately capital intensive and requires a fairly large number of moderately skilled employees. Back-end processing, also known as assembly or packaging, follows the fabrication process. Finished chips are sent to an electronics manufacturer, or manufacturing services
company.
Many global semiconductor packaging companies have established plants in China, to take advantage of low labor costs and for proximity to the PC and EMS industries, which are mostly located in Southeast Asia or China. Currently, about 9% of chip assembly and test is performed in China.
Key steps are:
. Wafers are shipped to the packaging factory where they are separated into individual chips.
. Wires are bonded to the chip surface, and to a mesh of external connection points.
. The chips are encapsulated in a plastic package; the external connections are exposed around the outside of the package.
. A final series of functional and performance tests are performed.
Packaged chips are shipped to electronics product manufacturers, where they are incorporated in the product and shipped to retailers, consumers or distributors.
As an example of the differences between the front end and back end sectors, Amkor Technology Inc. (B-/Positive/--), the largest packaging/testing company, had sales of $87,500 per employee in 2005, while Taiwan Semiconductor Manufacturing Co. Ltd., the largest wafer foundry, had sales of $378,000 per
employee. Reflecting the higher intellectual property content of wafer fabrication, compared to assembly services, TSMC's EBITDA margins were 63% in 2005, compared with Amkor's 16%). Manufacturers will gain economic benefits as the several stages of the electronics industry are located closer to one another; these benefits are especially important for lower-margin consumer electronics goods.
¬ Haymarket Media Limited. All rights reserved.
