How China Is Quickly Capturing The World's Semiconductor Industry
In a few short years, the semiconductor industry has decided to change its plans for building dozens of major new fabrication plants in Taiwan and instead locate them in China. In 2000, half of the fabs under construction in the world were in Taiwan, with confirmed plans to build about 30 fabs over the next eight years at a cost of $70 billion.
Today, those plans have been severely curtailed, down to only four or perhaps six 12-inch fabs planned for Taiwan. There are now as many as 19 fabs either under construction, operational or planned in China. With a majority of U.S. high-tech companies having outsourced the production of chips to Taiwanese companies, in the near future many of those semiconductors will be coming from China.
"The plan they envisioned for Taiwan is being implemented in China instead," says Thomas Howell, a partner at Dewey Ballantine LLP in Washington, D.C. There are numerous reasons for the abrupt change. China has initiated a program to replicate Taiwan's success in microelectronics "on a much larger scale, drawing heavily on Taiwanese and other foreign capital, management and technology," says Howell, citing China's Tenth Five-Year Plan for Information Technology.
But more than anything else, China is successfully attracting investment through its 17 percent value-added tax (VAT) on imported semiconductors (as opposed to a 3 percent VAT on those made in China). This 14-percentage point differential provides Chinese semiconductor manufacturers with an "extraordinary -- enormous -- pricing advantage," says Howell. "With this VAT, it's amazing how quickly the investment patterns changed."
The 14 percent surcharge on foreign chips entering China is enough to completely skew the balance and force those companies operating in China to seek indigenous sources of supplies. "The Taiwanese ran the foundry model over the years by saying they could serve anybody in the world from Taiwan. The question is, why did they jump across the straits when it's only a few miles? Why couldn't they serve the Chinese mainland from Taiwan?"
The answer was simple: Taiwanese semiconductor companies cannot compete in the Chinese market with the differential VAT. "It's a huge difference," says Howell. "Once you're there, you have an enormous advantage because the imports can't compete. The effect on profits is very dramatic."
But there is more at play than just the VAT. China has adopted a strategy of industrial development that stresses tax holidays and the creation of industrial parks. It has also adopted a strategy of "close followership" that seeks foreign investment and encourages absorption and diffusion of foreign technology. China views the United States as a collaborator, a source of technology, training and capital. And it is aggressively recruiting talent.
This is very different from the "leadership" development model being pursued by the European Union and Japan that depends on large-scale government-backed R&D projects, technological and market leadership and a view of the United States as a rival. Countries pursuing the "leadership" model of semiconductor industry development are trying to surpass U.S. capabilities.
Countries pursuing the "close followership" strategy (Taiwan, China, Singapore, Malaysia, Thailand and Israel) are successfully capturing a greater share of the market. This strategy "is the big challenge over the next 10 years" for the U.S. industry, says Howell. "They are drawing the manufacturing function to their countries, capturing investment and a rush of people to Asia."
China has created a tax-free environment for semiconductor plants. It has created "very powerful financial incentives to draw in talent from abroad," and it has all the capital and technology needed to put into place a vibrant semiconductor industry.
No new fab will pay taxes in China for 10 years. Employees are being given stock that is not taxed at market value. They can sell the stock and not pay any capital gains taxes. "The perception is you can get rich in a short period of time and there is a stampede of people now going to China," says Howell. Individuals are receiving tax rebates. They are being given houses and cars. "People are flocking from Taiwan to Shanghai -- hundreds of thousands of people, draining Taiwan of capital and people," says Howell.
Speaking at a recent congressional gathering under the auspices of the National Academy of Sciences, Howell said that every country in the world that has a semiconductor industry has active government programs nurturing the sector. If there are no government programs, there is no industry "and that includes the United States," he said. Programs being put in place abroad "are those that were implemented in the U.S."
Most countries wanting to develop an indigenous semiconductor industry have been studying the U.S. model and implementing best practices that encourage close working relationships between industry, university and government researchers. Other countries use the words "ideal" and "perfect" to describe the U.S. system of industry/university/government interaction. Other countries are trying to replicate the consortia model that was created in the United States with Sematech, as well as the venture capital culture that exists here. They are trying to attract the world's best talent and are investing heavily in their educational systems to train a new generation of technically adept individuals.
Yet the United States government has not recognized the importance of any of these programs and while the level of spending on semiconductor R&D has risen in most other countries, in the United States it has declined over the past five or six years.
Moreover, other countries are nurturing their industries by providing them with tax incentives, infrastructure improvements, direct financial aide, incentives for talent from around the world and new research programs on the cutting edges of technology. "A tax-free environment allows income to be made available for R&D investment," says Howell. "No new fab will pay tax in China for the next 10 years."
The United States needs to respond to the challenges posed by overseas competitors, says Taffy Kingscott, who works on public policy issues for IBM in Washington, D.C. The United States needs to insist that the World Trade Organization challenge China's 17 percent VAT, she says. The U.S. government should require that China provide full protection for intellectual property and increase enforcement of IP "because lack of enforcement enables the piracy of semiconductor designs," she adds.
The United States can also implement policies for accelerated depreciation of capital equipment from five years to three years. "The reason it's never been done is because it's expensive and the tax writing committees say, 'Why should we do it for one industry and not another?' "
The R&D tax credit should be made permanent and the methodology of what applies under the credit should be improved.
There needs to be additional spending on education, particularly in math, chemistry, materials, physics and engineering. Those disciplines have been starved recently and are in need of investment. "There is no recognition that investment in these areas is important but it is truly, truly the only way you have to achieve the basic understanding [needed] to keep this industry growing," says Kingscott.
More Americans need to be enrolled in advanced engineering and physics degrees, and the United States should address the issue of a high proportion of foreign students studying these disciplines in the United States only to go back to their country of origin. A proposal to accelerate funding at the undergraduate level for math and science education was attached to the NSF budget-doubling bill that passed into law. "It's a great idea, but those are authorization bills and not appropriations bills and they need to be funded," she says.
The Focus Centers Research Program supported by the semiconductor industry also needs to be funded by the federal government. These programs at universities include the financial backing of the Semiconductor Industry Association and equipment suppliers. The Department of Defense has been asked to fund a 25 percent share of these centers, but has no desire to do so. "It's a lousy $12 million and they zeroed it out" in their latest congressional budget request, says Kingscott. "It's a small number below the radar screen -- $12 million isn't even a pimple on the DOD budget, but it's hard for it to get any attention."
Economic incentive programs should also be encouraged at the state level. New York State has been active in providing economic incentives for the growth of its semiconductor industry around its universities and IBM over the past five years. To keep the semiconductor industry in the United States, the federal government should consider providing incentives that states could then use to build infrastructure that benefits the manufacturing sector. "Other countries do this very dramatically and with the state budget crisis it becomes even more difficult to make these kinds of choices," says Kingscott.
Please provide us with feedback on this article, or suggestions for people to interview.
To receive notification when new articles are posted, please sign up for a free e-mail newsletter.
You can receive more stories like this in every issue of Manufacturing & Technology News by becoming a regular subscriber.
Scan Back Issues | Reports & Analyses | Comments | About Us | How To Order