October 28, 2002
I.B.M. Weathers Tech Storm
ince he took over as I.B.M.'s chief executive eight months ago, Samuel J. Palmisano has moved swiftly to cut costs, jettison lagging operations and add to the company's big-services business.
When the slump in technology spending deepened, he shut down some of I.B.M.'s older semiconductor plants and sold off its hard-disk drive business to Hitachi. He has trimmed the work force by 5 percent, or 15,600 employees. He bought PricewaterhouseCoopers Consulting for $3.5 billion, in a deal that closed this month, to bolster I.B.M.'s services business.
The result has been that I.B.M. has weathered the technology storm better than most others. But the industry remains depressed and rudderless. What will start the industry moving again? What's next?
Reasonable questions, it seems, for the man who heads the computer company that spends by far the most on research and development. In a speech on Wednesday morning, before a few hundred corporate customers and industry analysts in New York, Mr. Palmisano intends to provide some answers. I.B.M. is not saying, but analysts who have been partly briefed on Mr. Palmisano's talk say he will articulate I.B.M.'s view of the future of computing as a utility-like service that companies can buy as needed, somewhat like electricity.
The utility concept, also called computing on demand, builds on work that has been nurtured in I.B.M.'s research labs, and elsewhere, for years and has been making its way into the marketplace recently. This work includes computer systems and networks of computers that use clever software and hardware to manage workloads and to detect and fix bugs on their own — an initiative I.B.M. terms "autonomic computing," after the human autonomic nervous system, which automatically handles basic functions like breathing and digestion.
Another ingredient is "grid" software, which shares computing resources and programs across networks. The grid software effort originated in the research community, led by computer scientists like Ian Foster of the Argonne National Laboratory and Carl Kesselman of the University of Southern California. The grids were initially designed to help scientists collaborate on complex projects that require massive amounts of firepower, like climate modeling and high-energy physics. But grids are being used in efforts including breast cancer research and oil exploration.
I.B.M. is by no means alone in its pursuit of utility computing or computing on demand. Microsoft, Hewlett-Packard, Sun Microsystems, EDS, Accenture and other companies are working on the software, hardware and services for utility computing. And industry experts recognize this as a potentially important direction for computing.
In a recent report, Steven Milunovich, an analyst at Merrill Lynch & Company, wrote: "The next big thing is the Internet as a distributed computing platform. Grid computing is distributing applications over the Internet, which allows companies to share not only computing resources but data." The report added, "Utility (or on-demand) computing is coming."
Next big thing, perhaps, but not really new. The utility concept has been around for four decades, first advanced by Herbert Grosch, an industry pundit who once worked for I.B.M. And computer time-sharing — the hot idea of the 1960's, both in the research community and on Wall Street — was a form of computing on demand. But the time-sharing systems of the 1960's proved unable to scale up beyond a couple of hundred users.
They were projects that staggered under the weight of their own complexity, given the hardware and software tools of the time. A renowned flameout of the late 1960's, the Berkeley Computer Corporation, employed some brilliant computing minds including Butler Lampson, Charles Thacker and Charles Simonyi. After the time-sharing failure, they went on to become part of the team at Xerox PARC during the 1970's, where so much of the fundamental technology of personal computing was developed.
Yet the modern advocates of utility computing point out that decades of advances in processing power, storage, networking and software make a huge difference. The Internet itself, they note, originated in the 1960's, but it wasn't until the 1990's, when all the technical ingredients fell into place, that it moved into the mainstream.
A similar phenomenon may be under way for computing on demand, as continuous progress in a range of related fields have enabled engineers to divide and conquer complexity in new ways. This means that grid-like distributed computing has a chance of moving into the mainstream.
"All these advances — faster and faster processing, better systems administration tools and the maturity of software communications standards — have really turned utility computing into a much more doable proposition," said Glen Salow, the chief information officer of the American Express Company. Earlier this year, American Express agreed to a seven-year computing-on-demand deal with I.B.M., after Mr. Salow became convinced that the pay-for-use, utility model would afford his company both reliability and cost-saving flexibility.
I.B.M., to be sure, faces plenty of competition in the emerging market for utility computing. Yet the company has some obvious advantages. I.B.M. seems to have the most ambitious approach with the most pieces of the technology in place. The implicit sales pitch — don't worry, we'll take care of all that computing complexity for you — is the message of I.B.M.'s big-services business. It echoes I.B.M.'s past as well, though in the mainframe era that meant buying the closed world of I.B.M. hardware and software.
"The model is kind of a variation of the mainframe days," observed Tom Bittman, an analyst for Gartner Inc. "But in the new world, I.B.M. is not locking customers into its proprietary technology as it did before."
In recent years, I.B.M. has displayed a deft hand in explaining unfamiliar new technology to the corporate establishment, which is I.B.M.'s customer franchise. Big Blue exploited the rise of the Internet in the 1990's by telling its corporate customers that the significance was not in viewing Web pages or "new media." No, I.B.M. insisted, the real payoff from the Internet was as a low-cost communications medium for reducing transaction costs of all kinds for business. Mr. Palmisano's predecessor, Louis V. Gerstner Jr., termed the dot-com startups "fireflies before the storm." I.B.M. dubbed its Internet idea "e-business" and marketed it hard and successfully.
In championing utility computing, I.B.M. and Mr. Palmisano face a similar challenge, but in a very different environment. In the latter half of the 1990's, the economy was booming and technological euphoria was in the air. Today, with the economy weak and technology viewed skeptically, I.B.M. runs the risk of marching off in a new direction only to find it has few followers.