Testimony of George Scalise
President, Semiconductor Industry Association
Before the U.S. - China Security Review Commission
January 17, 2002
On behalf of the Semiconductor Industry Association ("SIA"), I would
like to thank you for the opportunity to testify before the U.S.-China Security
Review Commission. I would like to address some of the elements of U.S.-China
relations as they pertain to national security and do so from the perspective
of the information technology ("IT") industry. In particular, and
as requested, I will focus on export controls.
SIA is the leading trade association representing the U.S. semiconductor industry.
Its member companies comprise 90 percent of U.S.-based semiconductor production.
SIA member companies manufacture a wide variety of semiconductor products, including
chips used in computers and related equipment, cellular phones, telecommunications
systems, automobiles, appliances, and many other items.
Security Dimensions: Information Technology And U.S.-China Relations
Information technology contributes significantly to all htmects of U.S. security
related to China -- political, economic and military. Information technology
has had a particularly important role in moving China to a more open, market-based
society that is becoming more broadly integrated with the rest of the world
community.
SIA began many years ago to seek to bring China into the world trading system
through multilateral trade agreements. SIA has also worked hard to support and
shape the basis for China's accession to the World Trade Organization as it
relates to information technology.
SIA welcomes China's accession to the WTO. As a part of its WTO commitments,
China has agreed to join the Information Technology Agreement and eliminate
tariffs on semiconductors, computers, telecommunications equipment and many
other IT products. China has also made numerous commitments to reduce and eliminate
trade barriers and to adhere to the same rules as other WTO members. China's
accession will contribute to further opening of its market and to the development
of IT in China. WTO membership will serve to make China's economic potential
an economic benefit to the community of nations and strengthen the rules-based
international trading system.
The United States should not take these developments for granted. While SIA
supports China's WTO accession, it needs to be accomplished pursuant to the
rule of law and China's obligations must be fully and carefully enforced if
China's impact on the world economy is going to be constructive rather than
generating major disruptions and distortions. SIA recognizes that the challenges
of WTO enforcement are substantial. For example, to circumvent high tariffs
and taxes, today up to 80% of semiconductors imported into China are traded
through indirect channels that are conducive to extensive smuggling. Further,
intellectual property protection, which is vital to our industry, is at best
very fragile in China.
The advancement of information technology also contributes to meeting the United
States' defense needs. Agility in utilization of information technology is
helpful to all sectors of activity: from consumers to students to soldiers.
But it should be emphasized that there is nothing inherently military about
IT. On the contrary, information technology is a product of, and has come
to be dominated by, the civilian sector. U.S. industry's ability to keep pace
with innovation, and therefore remain a reliable source of critical technology
to the U.S. military, is directly correlated to its ability to produce and market
its products on a global scale.
United States national security interests are dependent on how effective the
United States develops advanced technologies, not how the United States purports
to restrict technologies. To maintain its leadership in advanced technologies
-- leadership which is crucial to all dimensions of security -- the United States
must be able to:
• Provide an environment that attracts, and is conducive to, technology
development. The United States should welcome research and development within
our shores and value and foster greater technology advancement within the United
States.
• Maintain a strong education base including a vibrant university system
that attracts students from around the world. The United States Commission on
National Security in the 21st Century, co-chaired by former Senators Gary Hart
and Warren Rudman, concluded in its January 2001 Report, “Road Map for
National Security: Imperative for Change,” that “the inadequacies
of our systems of research and education pose a greater threat to U.S. national
security over the next quarter century than any potential conventional war that
we might imagine.”
• Maintain a regulatory system that does not drive technology development
elsewhere. The United States needs to make sure that the regulation of IT is
managed in a manner that allows technology development to flourish in the United
States.
Export Controls: A Critical Challenge for Relations with China
The United States should do what it can to better relations between China and
Taiwan so that the likelihood of aggravated political tensions or conflict is
reduced. Like elsewhere in the world, economic development in China and economic
integration with Taiwan can provide a stabilizing effect in that region.
One of the central features of the United States' security relations with China
is the unsettled policy area of export controls. The United States currently
places restrictions on a wide variety of commercial information technology exported
to China, including semiconductors and related equipment and technology. While
these Cold War-era controls are imposed in the name of national security, they
are increasingly becoming counterproductive to U.S. policy goals, most notably
the democratization of China and its integration into the world economy. Commercial
information technology will be a driving force in China's future modernization
and economic and political stability. Delaying the adoption of this technology,
which has already proven to be an important agent for change within China, will
work against U.S. interests and will be increasingly detrimental to the global
leadership position of many U.S. technology industries.
In addressing export controls and their impact on U.S.-China relations, I would
like to address the following topics:
• The nature of semiconductor products and other widely distributed IT.
· The unique opportunities and challenges presented by the Chinese market
for semiconductors and related products.
· Problems presented by the current export licensing system.
· Recommendations for changes in U.S. export control policy
I. The Nature of Semiconductors and Related Information Technology
Over the past few decades, the IT industry has undergone a dramatic transformation.
The manner in which IT products, including semiconductors, are manufactured,
marketed, distributed and serviced is radically different than it was even ten
years ago.
The growth and impact of IT driven by Moore’s Law has created change at
an exponential rate. For decades, computing power has doubled roughly every
18 months, while the cost of computing has been cut in half during the same
periods. As a result, we have moved far away from the world in which IT systems
were large, few in number, hard to build, and geographically constrained. We
have entered a fundamentally different domain where IT products are small, easy
to assemble, globally produced and distributed in huge volumes, and connected
to intelligent networks and the Internet.
The process by which IT products are produced and distributed has flipped from
a vertical "control" model to a horizontal "dispersal" model.
In the computer industry of the 70s, each company largely controlled the development,
production, sales/marketing and maintenance of its products. A company generated
its own designs, components, subassemblies, platforms, operating systems, applications
software, sales channels and product service. In the 1980s and 1990s, this vertical
model changed to a globalized horizontal model in which computers are built
(increasingly locally) by assembling mass-produced chips and other standardized
IT building blocks from vendors around the world. This change occurred because
companies saw that a horizontal approach would increase innovation, drive cost
reduction, and foster ease of assembly through uniform global standards. This
horizontal model is now being expanded to the Internet itself.
Within the last thirty years, information networks have changed from conventional
telephone systems and a few localized computer networks into a vast information
infrastructure. This transition has resulted in media-rich exchange of information
over long distances – from text to graphics to voice to streaming video.
It has also enabled computing to be conducted on a remote-access basis since
network capabilities remove the need for a physical presence in a given location.
Furthermore, networks are migrating beyond standard wired “PC-to-server”
connections to include cell phones, hand-held information appliances connected
on both a wired and wireless basis. Remote information exchange and computing
are thus becoming “the fabric of our daily lives” -- developments
that allow individuals and organizations always to be on the network and communicate
with anyone or anything anywhere. This is fundamentally changing the way in
which businesses and governments operate, as well as the way people conduct
their lives.
The ability to cluster computing power has allowed supercomputer performance
from networks of standard personal computers. The systems software that allows
cluster technology is readily available on the Internet. The emergence of cluster
technology has led the Office of the Undersecretary of Defense (Science and
Technology) to conclude in February 2001 that MTOPS controls for high performance
computers were ineffective and should be scrapped in favor of better security
of military applications software. A similar analysis led the Center for Strategic
and International Studies to recommend scrapping MTOPS controls on microprocessors.
Special links between the IT industry and military systems have ended. Thirty
years ago, the major U.S. semiconductor companies were, in large part, defense
contractors. They worked closely with the Defense Department and followed its
lead in developing advanced IT systems specifically for military applications.
Today, commercial applications completely dominate the markets served by all
major semiconductor producers. IT technology development is led by advances
in the commercial arena, and the vast bulk of IT products is inherently commercial.
Military systems rely principally on the same off-the-shelf, commercial IT products
used by civilian customers.
Given these drastic changes, applying the current Cold War-era export control
regimes to the commercial IT industry merits careful reassessment.
II. Opportunities and Challenges in China
China has been gradually but irreversibly embracing the global IT revolution,
and development is now moving at an incredible pace. The country's deliberate
and concerted efforts to promote the broad adoption of most types of information
technology raise fundamental strategic issues for the U.S. IT industry, and
the semiconductor sector in particular.
Although China remains in the early stages of IT modernization, expansion of
China’s IT infrastructure over the past decade has already resulted in
China becoming the world's second largest PC market and third largest semiconductor
market. The rate of growth has been among the highest in the world. By 2005
-- in just 36 months -- China expects to have 70 million computer owners
-- up from about 19 million today. China is predicted to be the world's second
largest market for semiconductors, behind only the United States, by 2010.
With regard to production, the Chinese Government's 10th Five-Year Plan aims
to increase Chinese semiconductor output from $2 billion in 2000 to $24 billion
in 2010, at which time Chinese producers would satisfy 50% of their domestic
demand compared to 20% today.
In short, the success of U.S. IT companies, including chipmakers, will be inextricably
tied to success in the Chinese market. To remain world leaders, U.S. companies
will have to be leaders in the Chinese market.
United States IT companies will not be able to secure and maintain leading positions
in China without manufacturing there. In general, IT companies need significant
production in critical markets to succeed in these venues. This is part of being
close to the customer – a cardinal rule in any business.
Moreover, it is clear that China will be a center of semiconductor manufacturing.
The Chinese semiconductor market has been supplied predominantly by foreign
chip manufacturers through exports of finished products. China still satisfies
only 20% of its demand for semiconductors through domestic production.
Increasingly, however, joint ventures have been formed to mass-produce semiconductors
in China. In 2000, for instance, plans were announced for several multi-billion
dollar Chinese semiconductor fabrication plants with advanced production technology.
One Sino-Japanese producer in Shanghai recently announced that it will be using
leading-edge, .18 micron process technology by early 2002. Over the past decade,
semiconductor capital spending in China has risen from a mere $20 million in
1990 to $3 billion in 2001. And analysts predict that semiconductor capital
spending in China will be $7 billion in 2003, more than 300% higher than 2000
levels.
In the past two years, there has been a surge of submicron fab activity in China,
with at least twelve fabs either under construction or in their planning stages.
Most of these facilities will utilize .25-.18 micron process technology and
will be capable of adjusting to even smaller feature sizes. These plants are
concentrated in industrial parks near Shanghai and Beijing.
Over the next several years, significant growth in Chinese semiconductor production
capacity is expected to be fueled in large part by Taiwan-based semiconductor
producers, who are transferring some manufacturing functions to the mainland
while retaining "high end" functions in Taiwan. Semiconductor firms
from Europe and elsewhere in Asia are also making strategic commitments to chip
manufacturing in China.
Beyond the cost-related reasons for expanding chip production into China, growth
in domestic productive capacity has become a priority for the Chinese government,
which desires a level of self-sufficiency in semiconductors.
Accordingly, it is important for U.S. semiconductor companies likewise to participate
in Chinese semiconductor manufacturing. As a related matter, it is important
that U.S. makers of semiconductor production equipment and materials or “SEM”
be able to sell into the Chinese market. United States chip producers rely heavily
on U.S. SEM companies to supply their fabrication facilities. And, like U.S.
chip makers, U.S. SEM companies must be leaders in China to be worldwide leaders.
Finally, with the continued growth in China, much of the focus for future research
in China will be in the area of IT. The opening of markets within China also
means that intellectual talent will grow, thus presenting an additional opportunity
for U.S. companies, particularly in an industry driven by intellectual talent
and creativity. United States semiconductor companies already utilize highly
skilled workers from the PRC, and this trend is expected to continue. It is
therefore imperative that the U.S. export control system not unilaterally and
arbitrarily place restrictions on individuals working for U.S. companies and
residing in the United States based on their non-U.S. nationality.
III. Problems with the U.S. Export Control System
United States export controls have long been a complicated issue for U.S. companies
competing in the Chinese semiconductor market. Despite periodic liberalization
of controls on semiconductors, and the overwhelmingly commercial nature of the
semiconductor industry, existing constraints continue to affect a wide range
of devices, technology and equipment and threaten to capture mass market, general-purpose
items.
SIA is concerned that the export control system has become disconnected from
IT realities, including the growth and impact of Moore's Law, the ubiquity of
commercial IT, the ability to cluster computing and foreign capability.
In many cases, export regulations represent much industry effort and restraint
without resulting in purposeful or effective control. Even if a license is granted,
export license application reviews cause damaging and undue delay and uncertainty.
The problem is essentially the result of a collision between centralized controls
and the decentralized nature of IT.
Policy-makers are unable to identify a connection between mainstream, commercial
IT products and national security that warrants maintenance of export controls
in this area. Most IT products are purely civilian items, which cannot, for
these purposes, be distinguished from civilian applications such as automobiles
or automobile production equipment. Consequently, restricting shipments
of these products is akin to an economic embargo, not strategic export controls.
There are three specific areas of the export control system that SIA believes
need immediate attention:
Performance-based controls on IT hardware components, including microprocessors,
digital signal processors and other microcircuits;
Controls on chip manufacturing capability and SEM; and
Controls that focus on the individual employee rather than company behavior.
Performance-Based Controls On Components. Propelled by Moore's Law, commodity-level
technology is perpetually on a collision course with performance-based export
controls. In 2000, 500 million 32 bit and above microprocessors and microcontrollers
were produced. Meanwhile, microprocessors used in business computers have increased
in performance over 22 times over the last five years.
The government has tried to keep pace with this accelerating technological advance
by raising MTOPS levels for processors and other microcircuits on an ever more
frequent basis. Since 1998, the government MTOPS level moved from 500 to 1200
to 1900 to 3500 to 4500 to 6500. Earlier this month, the United States announced
plans for yet another increase -- this time to 12,000 MTOPS.
The problem is that it has become an ever more challenging and unpredictable
exercise to update component MTOPS levels to keep pace with technological advances.
This timing "collision" can subject high volume products to export
restrictions that cause shipping delays, administrative burdens and potential
export denials. For example, in 1998 and 1999, low MTOPS limits subjected scores
of China-bound shipments of Intel's Pentium® processors to export licensing
delays that lasted for many months. Only when the MTOPS levels were raised could
Intel ship these items. If the levels had not increased, the licensing caseload
would have jumped to hundreds of applications, since the microprocessors in
question were only beginning to be released into the market place when the MTOPS
increases occurred.
This year, the current level of 6,500 MTOPS will be exceeded by the performance
of PC, workstation and server processors. Even with an increase to 12,000 MTOPS,
the government will have to continue adjusting the threshold in the future in
order to avoid a collision between high volumes of uncontrollable components
and the export licensing system. Future adjustments will require greater
increases in the threshold at even more frequent intervals. This process will
only increase the likelihood of a regulatory breakdown, without enhancing national
security in any measurable way.
Recognizing the futility of MTOPS controls, nearly every member of the 33-nation
Wassenaar Arrangement supported the elimination of these controls during the
organization's 2001 negotiation round. The United States was and remains
the principal opponent to this reform. As an alternative to MTOPS elimination,
the United States supports the perpetuation of the MTOPS adjustment process,
and has considered new controls that would impose a new, more complex tiered
approach to MTOPS controls.
Controls On Chip Manufacturing Capability and SEM. Manufacturing
or otherwise investing inside a major market, particularly in the early stages
of market development, is a fundamental business tenet since it can bolster
a company's status as a key supplier. In SIA's view, the strategic importance
of the market in China, coupled with major tax incentives and highly skilled
labor, will drive ever-increasing investments and technology transfers which
will quickly enable that nation to close its technology gap with other countries.
Today's export controls on chip manufacturing capability present a significant
competitive disadvantage in this regard. First, the controls seek to keep China
behind in the acquisition of semiconductor manufacturing know-how, despite readily
available foreign sources of supply. Second, SEM are tightly controlled by outdated
regulations, even though U.S. competitors have less stringent control regimes.
As a result, even if a license is ultimately issued, licensing delays can represent
a decisive factor leading Chinese buyers to source SEM, for example, from non-U.S.
sellers since Japanese and European SEM producers do not face the same licensing
constraints.
Controls on SEM are a particular challenge for U.S. competitiveness in China.
The adverse impact of these controls has grown as the demand for semiconductor
manufacturing equipment in China has continued to expand. SEM items have been
subject to very little export control liberalization since the end of the Cold
War. Most modern tools for chip fabrication facilities require a license when
exported to China.
As already noted, foreign companies are engaging in China-related technology
transfers, capital investments, building of wafer fabs, and joint ventures.
Other IT-producing countries have not committed -- through the Wassenaar Arrangement
or otherwise -- to administer any particular export licensing policy regarding
products such as semiconductors and SEM. Japan and European countries have made
it clear that they do not consider chips or SEM exports to China to represent
a security threat. Consequently, their licensing policies are far more liberal
than are U.S. licensing policies.
While the United States still has a large semiconductor production equipment
base, China can obtain all major types of semiconductor production equipment
from non-U.S. sources in Japan and Europe. For example, Hua Hong/NEC --
China’s most advanced 8”, .25 micron semiconductor fabrication facility
-- incorporates primarily non-U.S. production equipment. As a result of the
ready availability of SEM globally, the U.S. policy objective of using export
controls to keep China's indigenous semiconductor production two generations
behind the state-of-the-art is not being met, yet the controls remain, hampering
U.S. SEM suppliers and their ability to remain global leaders.
There is no justification for restraining U.S. exports to the extent that
the items to be exported are freely available from other sources. Doing
so gratuitously aids non-U.S. semiconductor and SEM development and production
at the expense of U.S. production and development.
Controls That Focus On Individual Employees Rather Than Company Behavior.
SIA member companies experience significant problems in the area of deemed exports,
i.e., transfers of controlled technology to foreign nationals working or visiting
in the United States that are from controlled countries. Many of these foreign
national hires are from China.
The deemed export licensing process is often unduly complex and time consuming.
It can take three to six months to process and may impose burdensome restrictions
on technology flows. The effect of these restrictions is to complicate intra-company
endeavors, significantly retard the development of new products and thereby
impede the advancement of technology in the United States.
Unlike other controls in the United States, the deemed export rule crosses beyond
the company level and seeks to unilaterally single out select individuals on
the basis of their non-U.S. nationality. As companies within the United States
-- particularly in the IT industry -- become increasingly reliant on talent
from abroad, administration of the deemed export rule becomes more difficult
and the detrimental impact it has on company operations and technological development
is greater.
IV. Recommended Improvements
The United States should generally replace its policy of controls on exports
of commercial IT products to China and elsewhere with a policy of staying ahead
technologically. This policy shift should embrace the positive correlation between
technological leadership and the broad dissemination and absorption of IT in
China and around the world. This will, in turn, lead to major political, economic
and, ultimately, security benefits for the United States.
In sum, the balance has shifted so that the benefits to the United States of
diffusing IT in China and worldwide far outweigh potential risks. The national
interest is advanced by promoting commercial IT exports rather than restricting
them.
Eliminate Performance Based Controls on Widely Available Commercial Items.
Geographic containment of components and computing power generally has become
unworkable in an era of IT globalization marked by wide availability of IT products,
worldwide IT manufacturing capability, and pervasive computing driven by networks
and clustering technology. For example, mass market microprocessors and other
microcircuits with rapidly changing performance are, by definition, so voluminous
and widely distributed that they are no longer susceptible of MTOPS controls.
It is therefore self-defeating to impose MTOPS export restrictions on widely
available commercial chips and computers. Such unnecessary controls only serve
to create a competitive disadvantage for U.S. companies, thereby thwarting their
ability to advance technologically.
Rather than perpetuate the flawed MTOPS control process, the United States should
seize an important opportunity and join with the rest of the Wassenaar Arrangement
in agreeing to eliminate the MTOPS parameter as a basis for controlling general
purpose commercial chips.
Remove Outdated Export Controls on Chip Technologies and Semiconductor
Equipment and Materials. Inevitably, China will be a center of semiconductor
manufacturing. Chip technologies and SEM are widely available to China through
non-U.S. sources and semiconductor firms from Europe, Japan and South Korea
are currently positioning themselves through strategic commitments to chip manufacturing
in China in order to achieve a strong foothold in this crucial market. Rather
than imposing outmoded Cold War-era export controls on these items, U.S. export
control policy should focus on determining with greater precision the specific
types of technologies that are truly sensitive.
Focus on Company Behavior Rather Than Individual Employees. United
States export control policy historically has not sought to look behind a company's
internal operations. The principal exception, promulgated in 1994, is the deemed
export rule. This rule has served to impede technology transfers within a U.S.
company operating in the United States, made it increasingly difficult to attract
the most talented scientists and engineers, and stunted the pace of technological
advancement in the United States. All of this has occurred with no discernible
benefit to U.S. security.
The deemed export rule should be eliminated. Consistent with other U.S. controls
and regulations, the onus should be put on companies to ensure that technological
know-how and information is kept within the confines of their operations.
* * *
During the spring of 2002, the Congress is expected to continue its consideration
of a renewed Export Administration Act, the authorizing legislation for U.S.
dual-use controls. As a general matter, new export control authorizing legislation
should reflect changing U.S. security interests and the dynamic growth of technology.
It should address the current shortcomings of the export control systems as
well as establish a basis for an effective and enduring regulatory regime. To
this end, an improved export control system should have the following generic
features in addition to the specific recommendations enumerated above.
Flexible. The pace and proliferation of modern technology development means
that an export regulatory system must have a variety of ways to adjust controls.
Authorizing legislation should mandate a focus on unique differentiators of
military performance and dismiss items that do not merit control or are not
controllable. United States export controls should account for rapidly changing
product performance levels, mass market products, foreign availability global
markets and manufacturing, and electronic transfers of information. One hallmark
of a flexible system would be reliance on measures other than license requirements,
such as mandatory notification for sensitive exports.
Timely. New authorizing legislation should establish a control regime that operates
with the speed of modern information technology. Decision-making delays measured
in weeks and months are unnecessary and intolerable.
Balanced. For dual-use items, perceived national security and foreign policy
interests should be weighed against economic and technological realities. Restricting
civilian exports has an economic cost to the nation and a competitive cost to
the affected industry. These are costs that can undermine the technological
leadership on which the U.S. military depends. The overriding national security
goal should be continued and expanded U.S. technological leadership, which depends
largely on global competitiveness.
Simplified. Any modern export control system must be based on voluntary compliance
and therefore must be clear and understandable to businesses large and small.
Imposition of complex requirements works against this principle and will ultimately
decrease the effectiveness of export controls.
Multilateral. Where controls on information technology or other products are
warranted, they should be implemented on a multilateral basis. A multilateral
approach is the only way that controls can work in an era of global production
and distribution. Unilateral controls should be avoided at all costs. They are
self-defeating because they provide advantage to foreign competitors while failing
to realize the objective of control.
These improvements would help maintain the U.S. leadership in cutting-edge technology,
rather than isolating U.S. industry from commercial and investment opportunities
in China and elsewhere. In the long run, current export policies that isolate
U.S. industry serve to diminish U.S. interests -- not strengthen them. Industry's
ability to keep pace with innovation in information technology -- and therefore
remain a reliable supplier to the U.S. military -- is directly correlated to
its ability to produce and market its products on a global scale. Removing costly
and unnecessary export controls on information technology is crucial to achieving
this end.
For the same reason, SIA also encourages this Commission to include in its final
report specific recommendations related to education and university-based research.
We believe that in the next 10 to 15 years, the fundamental semiconductor process
on which the industry has depended for the last 30 years will approach its physical
limits, and that university research in fields such as materials sciences and
physics must be increased so that the understanding of nanoscale materials will
exist to develop electronic devices to replace our current semiconductor circuits.
If semiconductors reach their physical limits and we have not adequately invested
in the science necessary to develop a replacement technology, our nation will
quickly lose its technological lead over China as well as other nations in the
critical field of microelectronics. We would thus ask that this Commission
include in its recommendations that Congress appropriate the funds necessary
to implement the math and science education initiatives that were included in
the "Leave No Child Behind" Act recently signed by the President,
and that Congress and the Administration commit to substantially boost funding
for information technology related research at our nation's universities over
the next five years.
* * *
In sum, export controls represent a major, unnecessary impediment on U.S. participation
and success in the Chinese semiconductor market. Outdated and unjustifiable
restrictions on the export to China of U.S. finished products, as well as on
manufacturing equipment and related technology, currently produce significant
competitive costs to U.S. companies, particularly as foreign competitors vie
for market leadership. Without changes in current U.S. export control policy
towards China, U.S. companies will increasingly fall behind in this crucial
market, and, by extension, the global market.