U.S. Getting Supercomputer Competition Japanese Firms Now Bidding On Sale To U.S. Government

Americans make the best supercomputers in the world. Or at least they did. Two Japanese electronics giants, Fujitsu and NEC, are challenging the supremacy of Cray Research, the Minnesota supercomputer specialist that was recently bought by Silicon Graphics.

The three are bidding to sell a supercomputer to the National Center for Atmospheric Research, an agency of the National Science Foundation. Should Washington come to Cray’s rescue and lean on the foundation to buy American?

The Clinton administration, which has yet to express an opinion, is under pressure to view the question through a parochial political lens.

Supercomputing, after all, is a glamorous frontier of high technology, the sort of industry where American business must excel if President Clinton’s vision of global capitalism powered by Yankee ingenuity is to become reality. Besides it is an election year in America, and the folks who make Fujitsu and NEC computers don’t get to vote.

But economists see another issue at stake. How can we ask Tokyo to play fair on trade issues if Washington doesn’t, asks Robert Lawrence of the Kennedy School of Government at Harvard. “If we make an exception for protectionism in high technology,” he argues, “we’ll all be worse off in the long run.”

Supercomputers occupy a modest niche in the computer market, and a not very profitable one. But the prestige of making dazzling machines that can perform tens of billions of calculations a second has kept a number of major hardware manufacturers, along with a handful of specialized producers, in the supercomputer game.

And after years of Byzantine maneuvering to defend their home markets and penetrate foreign ones, America and Japan called a truce in 1990. They agreed to open both public and private procurement to the lowest bidders.

Japan’s commitment to a free market is a matter of debate. Kenneth Flamm, an economist at the Brookings Institution, suggests that public procurement is still uncompetitive in the sense that nationality counts a lot.

The Japanese government, however, has purchased seven of its last 13 supercomputers from U.S. makers, while Japanese companies have yet to make a sale to a government agency in America.

U.S. makers, among them Cray, arguably hold the technological lead in specialized supercomputers for military purposes like the simulation of nuclear explosions. But in peaceful uses, Japanese equipment is competitive on cost, performance and reliability.

Fujitsu and NEC have recently sold machines for weather simulation in third markets (Europe and Canada) where the buyers have no particular reason to play politics with supercomputers. That is why the stakes in the current case are so high: it apparently represents the first clear test of Washington’s will to honor the supercomputer agreement.

In theory there is a slender economic rationale for protecting the domestic American supercomputer market. The world may have room only for a handful of efficient supercomputer makers, just as there is room only for a handful of efficient jet aircraft manufacturers. Giving Cray a preference in the largest piece of the market may thus make it possible for it to survive, even thrive, among the more diversified computer makers.