IBM's Research Cutbacks
Now Seem to Be Brilliant

By BART ZIEGLER
Staff Reporter of THE WALL STREET JOURNAL

IBM's latest research breakthrough -- that it can boost the power of computer chips 40% by implanting them with microscopic circuits made of copper instead of less-conductive aluminum -- rocked the industry. But the discovery, announced two weeks ago, was also sweet vindication for the company's combative chairman, Louis V. Gerstner Jr.

When Mr. Gerstner swept into International Business Machines Corp. 4 1/2 years ago as the first outsider to run the computer giant, the Armonk, N.Y., company was hemorrhaging money and losing market share. Mr. Gerstner shuttered plants and laid off thousands. But one move was especially shocking: He ordered a $1 billion cut in IBM's once-sacred research-and-development budget.

Some said that Mr. Gerstner, who previously ran companies selling credit cards and cookies, had no business tinkering with a technological treasure. They warned of a threat to U.S. competitiveness and feared the Nobel Prize-winning IBM Research Division, which invented such seminal computing devices as the hard-disk drive and memory chips, would be irreparably harmed. "If IBM is getting out [of basic research], who's going to do it?" an alarmed official of the National Science Foundation asked at the time.

But now it appears the cost-cutting and refocusing at IBM Research wasn't such a bad thing after all. Instead of resulting in a demoralized, damaged operation, the changes have energized many at IBM's three major research labs, in Silicon Valley, New York state and Switzerland.

Gone is the heavy emphasis on research for its own sake. Today what matters is getting the fruits of that research to market -- and fast.

The changes are helping solve an age-old problem for IBM: Though its labs turned out groundbreaking technology for decades, too often those advances showed up first in the products of competitors. For example, IBM invented a breakthrough computer design called RISC, for reduced instruction-set computing. Unable to protect it with patents, IBM sat back for years while Sun Microsystems Inc. and others turned it into commercial fortunes. IBM missed other industry-altering inventions entirely. Rivals developed the microprocessor and the "graphical user interface" -- read Windows -- that drove the personal-computer industry and led to huge profits.

This sorry history frustrated Mr. Gerstner. He fumed as Microsoft Corp. and Intel Corp. won much of the industry's adulation and market share, even though they did little basic research and in many instances built their businesses largely on inventions made elsewhere.

Customer Service

And so the new chief executive ordered his company to get its research into its own products. First.

That fiat forced an attic-cleaning. IBM Research abandoned unpromising areas of inquiry, emphasized projects with greater potential and got rid of lower-ranked staff. Scientists were directed to spend much more time with product developers and even customers, something unheard of in the IBM of old.

"This used to be a country-club atmosphere," says Hans Coufal, who works at IBM's Almaden Research Center in San Jose, Calif., where he is trying to perfect a device that one day might replace computer hard drives by storing data in holograms. Under the new regime, Mr. Coufal says, "I really enjoy solving real-world problems, having the feeling that I'm really needed and appreciated."

The big unknown, however, is whether IBM has lost the serendipity factor. "Eureka!" moments in the lab can't be scheduled into business plans like the launch of a new line of PCs. Instead, they require years of patient, long-term research.

Goodbye Blue Sky

Hundreds of scientists left IBM as it cut research-and-development spending to $5 billion a year from about $6 billion and imposed an even steeper cut of 37% in just the Research Division's $650 million annual budget alone. Many who stayed turned their focus from blue-sky efforts to more mundane product matters. Gone was IBM's quest to be the first to measure the mass of the smallest subatomic particle, the neutrino, in hopes of furthering Einstein's theories. In its place were efforts to develop a foldable keyboard and to improve the little "eraser-head" used to control the cursor in IBM laptop computers. "The long-term outlook went very sour," says Richard Webb, a respected physicist who departed in frustration amid the downsizing in 1993. "Almost everything became short-term directed: solutions and applications."

Mr. Webb, who had been exploring the physical behavior of atoms in very small devices, laments that by cutting his team from a dozen to just two, IBM lost "some brilliant young people... . All the infrastructure I had built up over 15 years was basically destroyed." Now running a research program at the University of Maryland in College Park, he argues that the cutbacks could come back to haunt IBM and other corporate labs. "How long is this going to go on before the technological superiority of this nation is in danger?"

But a close look at IBM's global research effort shows it has hardly abandoned basic science or long-term projects, at least not in the areas it regards as most important to its business. Instead of spreading itself too thin, IBM now tries to target areas where it has both extensive background and the best minds. The Research Division's budget has slowly grown back to about $600 million this year, and its staff is up to 2,785 from fewer than 2,500 three years ago. IBM estimates that 10% to 15% of its researchers are engaged in long-term work.

Paul Horn, IBM's research director, says the overhaul was about more than just cost-cutting. Fifteen years ago, IBM had an iron grip on the mainframe-computer business and controlled the speed with which new technology entered the market. "Now, winning is taking ideas that are very far out and being the fastest in converting them into significant technological advantage," he says.

The IBM Ethos

Any concern about speed-to-market was unheard of in 1945, when Thomas J. Watson Sr., IBM's founder, started the company's first formal science operation by opening a lab at Columbia University in New York. Wallace Eckert, the lab's first director, said at the time that the center's mission was to carry out "scientific research where the problem is dictated by the interest in the problem and not by external considerations."

That ethos continued for many years. In 1961, IBM opened the Thomas J. Watson Research Center in Yorktown Heights, N.Y., an imposing, semicircular building with sweeping glass-and-stone walls. The Research Division became best-known for its basic science involving such esoteric but fascinating areas as fractal geometry and superconducting materials.

Generous research funding continued until the early 1990s, when the mainframe market soured and IBM faced financial crisis. The first cuts began even before Mr. Gerstner arrived, as IBM brass ordered James McGroddy, then its research director, to slash $50 million from the division's spending in late 1992.

'Consuming Capital Like Crazy'

"We had been a very rich and happy corporation," says Mr. McGroddy, now retired. "And we were consuming capital like crazy. A lot of things could easily be done much better." He pushed to have 20% of the Research Division personnel working on projects that could directly lead to products and customer services, up from just 2% in 1990. He held meetings to explain why the overhaul was needed, telling scientists "we're going to make a big turn. This change is real." Some were outraged. "They didn't buy the story we were telling," Mr. McGroddy says now.

The pressure intensified after Mr. Gerstner's arrival on April 1, 1993. By 1995, the Research Division budget had plunged to $475 million from $650 million in 1991. Employment at the labs fell to fewer than 2,500 by 1994, down about 1,000 jobs from 1990, with most of the cuts coming from the research ranks rather than the support staff. The division vacated rented buildings and consolidated at IBM-owned sites, shrinking its operation in upstate New York to two buildings from a half dozen. The cuts in the number of buildings and in support staff (to 300 from 600) produced annual operating savings of about $30 million.

Consolidating Brains

IBM also combined research efforts that had been spread out among various labs, in an effort to end overlap. All its work on disk drives, for instance, was moved to its Almaden center in Silicon Valley. Mr. McGroddy killed projects that no longer looked fruitful because of changes in technology, or were ones in which IBM didn't have the best skills, among them "magnetic bubble" memories, astrophysics and chips made from an exotic substance called gallium arsenide. Some researchers in these areas moved to surviving projects inside IBM's labs, while others left to pursue their projects at universities.

At the same time, IBM expanded promising programs that could help customers sooner. These included voice-recognition systems, Internet-security software, data-storage technology and biometrics, which is the use of biological signatures to identify people.

"The difference that the Gerstner regime introduced was, it wasn't enough anymore to just move ideas into the product world. What was important for us in Research was to understand what the market wanted" and then work from there, says Inder Gopal, who left IBM Research last year to join Prodigy Inc.

One way IBM Research responded to Mr. Gerstner's call is a program called First of a Kind, which pairs research projects nearing completion with an IBM customer to help solve a real-world problem. The intent is to come up with a solution that can be replicated for many customers.

Taking Dictation

In one such pairing, researchers developed a voice-response system for radiologists by working with Memorial Sloan-Kettering Cancer Center in New York and Massachusetts General Hospital in Boston. Previously, doctors examined patients' X-rays and dictated their findings into tape recorders for later transcription, causing a delay. Now they can dictate to a PC, which automatically turns the diagnosis into text. The resulting IBM product is now sold to hospitals across the nation.

The research arm's closer interaction with IBM product developers also is paying off. This year alone, work at IBM's labs has found its way into the processing chips inside IBM's popular new mainframes; a wireless modem that links PCs through the airwaves; and a tiny disk drive for laptop computers that can store five billion bytes of information. The Research Division's most-hyped success this year had little to do with business computing: It was Deep Blue, the supercomputer that whipped chess-master Garry Kasparov.

IBM's copper-chip breakthrough, another collaboration between researchers and product people, is likely to have the most long-term impact. The goal was especially elusive: how to form the microscopic circuits on the silicon base of computer chips using copper instead of the traditional aluminum. Copper is a much better conductor of electricity, which means more electrons can move across even tinier circuits, allowing more circuits to be etched into each chip. Added circuitry, in turn, means a more-powerful chip.

Atoms Askew

But copper atoms are unruly. Rather than stay put, they tend to seep into the silicon, contaminating it. "I was starting to think the problems were insurmountable," recalls Randall Isaac, the IBM Research vice president who heads the scientific part of the team. Coating the copper with a barrier to keep the atoms in place was one potential solution, but that required making the copper lines so thin that their advantage over thicker aluminum circuits was negated.

"In the early '90s, one of the breakthroughs was to discover a barrier that was thin enough and sticky enough," Mr. Isaac says, declining to discuss the material, which IBM regards as a trade secret. "The ability to find that was one of the outgrowths of our strength in materials science." The resulting technique "changes one of the fundamentals" of chip making, says John Kelley, a vice president at IBM Microelectronics, the chip division.

Despite the emphasis on pushing research into products, some IBM scientists still do work that is years away from leaving the lab or may never see the light of day. One such scientist is Don Eigler, who gained fame as the first to manipulate individual atoms with a special microscope, spelling out the letters I-B-M with the tiny particles.

Like a Video Game

In his lab at IBM's Almaden center, a mountainside complex surrounded by a 690-acre wildlife preserve, Mr. Eigler moves the mouse attached to a PC. His commands are carried out in tandem in an adjacent room by a needle-like mechanism on a machine the size of a refrigerator. As the microscope's tip passes over a thin layer of copper and manganese, a representation of the manganese atoms appears on the computer's screen as a series of gray blobs. By applying a tiny electric current to the tip, Mr. Eigler can pick up an atom and then set it down nearby. It is an amazing achievement, but one that his exotic setup handles as easily as if this were a video game.

His research into the movement of atoms on surfaces might one day be important in making hard drives and chips. Or it might never help out with products.

Yet even Mr. Eigler, working on some of the "purest" research at IBM, worries about the bottom line. The scientist, who favors black T-shirts and jeans, proudly gives tours to delegations of customers. "It's always a crack-up to me to let them move atoms around. Everybody loves it," he says. His purpose is more than mere show-and-tell, he says; he hopes it "helps build a relationship." Why bother? "I'm an IBM stockholder."

Copyright © 1997 Dow Jones & Company, Inc. All Rights Reserved.