Asian Wall Street Journal
September 19, 2000, p. 6
U.S. companies are selling countries the super computers they need to build the weapons America will have to defend against.
Much has been made lately of the U.S. proposal for a national missile defense system, and whether U.S. taxpayers should pay billions of dollars to counter threats from China and other countries developing nuclear and missile arsenals. But what American taxpayers don’t know is that U.S. companies could soon sell these countries the computers they need to build the missiles America will have to defend against.
Last month, after intensive lobbying by the U.S. computer industry, President Bill Clinton decided to lower once again the barriers that control the export of American-made, high-performance computers. As a result, arms makers in China and other countries will soon be able to buy computers up to 14 times more powerful than the ones they were able to get only eight months ago. Moreover, under the new regulations, which are scheduled to take effect in January, countries allowed to purchase such computers can do so without any security review by the U.S. government.
The decision pleased Silicon Valley, but it exposes the rest of the world to greater nuclear and missile threats. High-speed computers are the most powerful tools known for designing nuclear weapons and the missiles to deliver them. These computers can simulate the implosive shock wave that detonates a nuclear warhead, as well as virtually every force affecting a missile from launch to impact.
The Clinton administration decided to allow the export of high-performance computers with processing speeds between 12,500 and 28,000 MTOPS (millions of operations per second), claiming the new rules would “promote national security” while easing “the unnecessary burdens on both government and industry.” But according to recent studies sponsored by the U.S. Commerce and Defense Departments, computers operating at these speeds will give the biggest boost to foreign nuclear and missile makers.
Consider China. Beijing is now trying to develop smaller nuclear warheads to fit on new, longer-range missiles. Many of these will be aimed at U.S. cities.
But to achieve their goals, Chinese bomb designers will need computer simulations. Live nuclear tests are now taboo. What sort of computing power will be required? For “nuclear blast simulation,” the two studies say, China will need computers operating between 10,457 and 21,125 MTOPS – precisely the ones now being cleared for export.
China is also improving its missiles and submarines. To simulate the “aerodynamics of [a] missile at Mach 2.5,” the studies say, computing speeds of 17,503 to 20,057 MTOPS are needed. For a “3D simulation of [a] submarine,” and the “modeling of turbulent flow” around it, the studies list a speed of 21,125 MTOPS.
The boost to China’s military will be dramatic – and unprecedented. At the start of this year, the Chinese military could only buy computers rated at 2,000 MTOPS – just above the desk-top level.
The Clinton administration defends its decision by citing the increasing power of computer chips. More powerful chips, the White House says, will make more powerful computers too numerous to control for export. It is undoubtedly true that chips are getting faster. The computer industry has long argued that because computing power tends to double every 18 months, due to better manufacturing techniques, the export controls should be eased.
In a report on its decision last month, the Clinton administration told Congress that four of Intel’s new “Itanium” chips, if wired together in a single machine, would perform at speeds of about 24,000 MTOPS. The report predicted that these chips would be introduced in October 2000, resulting in “widespread commercial availability.” And it implied, but did not say, that computers using these chips would be available from abroad and too numerous to control.
But the administration is simply wrong. Chuck Mulloy, an Intel spokesman, stated last week that the Itanium chip won’t be available in volume to computer makers until mid-2001. Computers using those chips won’t come to market until sometime after that – a year from now at the earliest. And there is no proof that even then such computers will be available in large quantities from foreign manufacturers.
The U.S. government’s own General Accounting Office has greeted the Clinton administration’s reasoning with polite contempt. After investigating the administration’s decision to significantly relax computer export controls in 1996, the GAO found the decision “lacked empirical evidence or analysis.”
More recently, the GAO criticized the administration for “relax[ing] controls based on what computer manufacturers asserted would be their next mass-produced processors, not on actual sales.” As for the argument that China could buy the computers elsewhere, it said, “U.S. companies and their international business partners overwhelmingly dominate the international market for supercomputers.”
So why is the Clinton administration allowing American manufacturers to outfit arms-makers around the world? The most plausible answer resides in campaign contributions.
Since January 1, 1999, the computer industry has given more than $3.2 million to the Democratic National Committee, Mr. Clinton’s party machine, making it the committee’s fourth largest industry contributor, according to the independent Center for Responsive Politics. The industry also shelled out money to the Republican National Committee, but not so generously. It was the ninth largest donor with $1.9 million, according to the center.
This tide of silicon dollars coincides perfectly with Mr. Clinton’s latest decision to relax export controls. In mid-1999, the administration announced that it would lift export restrictions on computers operating between 2,000 and 6,500 MTOPS. The change didn’t take effect until January 23, 2000 because of a six-month delay imposed by Congress. Then in February 2000, the administration announced it would raise the level from 6,500 to 12,500 MTOPS by August. The last change takes the level from 12,500 to 28,000 MTOPS.
The new regulations are particularly troubling with regard to China. Of the hundreds of computers capable of 2,000 MTOPS or higher that were exported to China since 1996, only a handful have received the post-shipment security checks that are required by U.S. law. Under flimsy procedures established by China and the U.S. in 1998, the Chinese government has rejected almost every request to locate the whereabouts of U.S. built computers, claiming the checks violate its national sovereignty.
A number of dangerous Chinese buyers stand to benefit. One is the Harbin Institute of Technology, which makes rocket casings and other components for China’s long-range nuclear missiles. Another is Northwest Polytechnical University, which develops engines and guidance systems for those missiles. Both were blocked from acquiring U.S. computers under previous rules. Now both will be able to buy as many as they want.
In effect, America is competing against itself. Ironically, it is trying to base its security on advanced technology, while at the same time hawking that technology to the rest of the world. It simply won’t work. With the revolution in military affairs, the modern battlefield is electronic, and computing power now decides the outcome of wars.
In future conflicts, do Americans really want American soldiers to face powerful American computers in the hands of the enemy? That is the real question posed by export controls. The answer should be no.