InformationWeek: The Business Value of Technology

March 12, 2001

Printer-friendly

Reinventing Research

"As Wall Street ratchets up the pressure on high-tech companies to show growth, and customers see technology's potential to affect the bottom line, turning research into revenue has become a priority for IT vendors.

By Aaron Ricadela   (aricadel@cmp.com)

More on R&D:

sidebar: Partnerships And Spin-Offs Are On The R&D Menu Microsoft Gets Serious About Wireless (2/26/01) TechWeb Finance: Startup Will Exploit Companies' Neglected Inventions (2/19/01)

s it labors to unravel the fabric of the human genome, Celera Genomics Group is racing to analyze reams of chemical, genetic, and experimental data with systems the company says trail its accelerating needs by a mile. Can the world's biggest computer company keep up?

Scientists at IBM Research's Thomas J. Watson Research Center in Yorktown Heights, N.Y., are developing Blue Gene, a new breed of supercom-puter the vendor says will be 1,000 times more powerful than Deep Blue, the RS/6000 that beat chess champion Garry Kasparov four years ago by calculating some 200 million board positions a second. It's part of IBM's proposed answer to the life-sciences' computing crunch, a machine that can simulate the complex process by which proteins in the human body "fold" into their distinctive shapes. The goal: to shed light on how diseases form and which drugs can combat them.

But Celera, a unit of Applera Corp., which last month published what's regarded as the most accurate and complete version of the human genetic code--the sequence of DNA that occurs in each of the body's 100 trillion or so cells--wants more. Its Rockville, Md., data center is packed with 400 Compaq systems running Tru64 Unix that house a total of 100 terabytes of disk space and the capacity to compute 1.5 trillion calculations a second. That's about 10% of the storage capacity and 1% of the computing power the company estimates it will need in five years. Celera plans to conduct 1 million experiments a day to analyze tissue samples, with each day's output generating roughly 7 terabytes of data.

"Our biggest limitation right now is in computing," Celera chairman J. Craig Venter says. According to VP of infrastructure technology Marshall Peterson, IBM's Blue Gene is a highly specialized project. Its contribution to humanity could be great, but it doesn't meet all of Celera's requirements. "The one thing I try to get across to IBM and other companies is that we need a revolution, not an evolution," he says.

The seeds of revolutionary technology are being sown every day in the research labs of the world's largest technology suppliers and the IT shops and boardrooms of their most influential customers. Turning that research into revenue has become a priority for IT vendors, and with good reason. In recent years, Wall Street has ratcheted up the pressure on high-tech companies to show growth, while more customers have seen the potential for innovative technology to affect their bottom lines. But with tech companies' earnings foundering amid a cooling economy, there's greater urgency for vendors to invest in research that can pay dividends, sooner rather than later.

How do they do that? Talk to customers, says one expert. "IT vendors need to incorporate the voice of the customer into the agenda of the research organization," says Hank Chesbrough, a professor at Harvard Business School who's an expert on high-tech research labs. "You're still doing world-class research, but the problems you choose to work on are influenced by this direct, face-to-face communication with customers." Though the economic slowdown has caused users to curb spending on some less-critical IT projects, the door's still open for technology that delivers real business value (see "The Shrinking IT Budget," March 5, p. 22; informationweek.com/827/priorities.htm).

For IBM, it's a long way from the dark days of the mid-1990s, when the company slashed its labs budget, scientists left, and projects often languished in the pipeline. Under CEO Lou Gerstner, there's been more emphasis on projects the company can turn into revenue quickly. Recent examples include the new TransNote portable computer, which works with a digital notepad to integrate handwritten data; fast-running Silicon-Germanium chips that can be used in innovative wireless devices, such as collision-warning radar systems in cars; and the miniature Microdrive storage device, which holds up to 1 Gbyte of data, for the booming digital camera and handheld computer markets. Under new CEO Carly Fiorina, Hewlett-Packard Labs has also honed its customer focus. Meanwhile, newer labs at Intel, Microsoft, and Sun Microsystems pump out technology directly wired into products those companies are producing.

But troubles at other vaunted research institutions--chiefly Xerox Palo Alto Research Center (PARC) and Lucent Technologies' Bell Labs--have tech vendors thinking even harder about how to make research pull its weight. The challenge: Users are moving faster than ever in key sectors such as life sciences, personalized database marketing, and storage--often outpacing vendors' ability to keep up. "Technology is evolving in IT so fast that it devalues itself relatively quickly, especially if you sit on it," says IBM Research director and senior VP Paul Horn. "Our organization, our financial model, our measurement system--everything is about, 'How do you speed innovations to market?'"

That mind-set means most vendors aren't putting research projects on hold while waiting for the economy to turn around. "You never save your way out of a recession," Intel CEO Craig Barrett said recently. Intel will trim its R&D budget this year, but only slightly, by 2%, to $4.2 billion. The only way to emerge stronger, Barrett says, "is to have great new products." Sun last month warned that its third-quarter profits could be half of what analysts had projected, but the vendor said R&D spending will rise 20% for the quarter, which ends this month. It all translates into good news for customers, as research labs jump to meet their immediate needs while anticipating future requirements.

"We've always been known for delivering what our operators want three to four years down the road," says Sue Spradley, VP for network intelligence at Nortel Networks Corp., which is working with Sun Labs to tune Sun's Java software for high-availability phone systems. "But they want us to look at things coming 10 years down the road," like systems that automatically switch data among residential, cellular, and Ethernet networks. "Rather than pretend to have all the experts, the partnership with Sun is pretty invaluable," Spradley says.

Some 300 miles north of Celera's headquarters, in a simply-furnished office at IBM's Watson lab, Blue Gene's father, IBM fellow and VP of systems Mark Dean, contemplates the connection between research projects and customers' needs. "It's safe to say there's not a group at IBM that doesn't have an opportunity in the life-sciences field," says Dean. "We're an integral part of a bigger piece." Blue Gene is based on cellular multiprocessing architecture, which sprays requests to tens of thousands of chips at once--each containing its own CPU, memory, and input/output--in order to achieve the fastest-possible response times. If it lives up to IBM's expectations, Blue Gene will perform more than 1 quadrillion floating point operations per second when it's finished in 2003--giving it 2 million times more power than what resides in today's slickest PCs.

A former high-school jock who maintains his sturdy build at age 44, Dean has seen a lot during his 20 years at IBM. One of the industry's few prominent black computer scientists, he's a member of the National Inventors Hall of Fame for his pioneering work on the PC--he holds a third of its original patents. Blue Gene, Dean admits, is "out at the fringe of research." But because its architecture is so tightly coupled with the demands of life-sciences companies, it's also "key to the future of IBM." Last summer, IBM launched a $100 million Life Sciences division to target a market it sees as a huge growth area.

This month, IBM Global Services starts a consulting practice in the field. To help ensure the initiatives' success, Dean's researchers are talking to genomics, life-sciences, and pharmaceutical companies--including Celera--as well as national labs and universities in a quest for feedback. "The key to success is, we can't do it alone," he says.

But for the time being, Celera is leaving IBM out of its supercomputing equation. In January, Celera, Compaq, and Sandia National Labs signed an R&D agreement to build a life-sciences supercomputer based on Compaq's Alpha chip--a prototype is due in 2004--that's capable of 100 trillion floating-point operations per second. Compaq and Celera also plan to develop new research algorithms and visualization technologies for analyzing massive amounts of experimental data. Compaq won the bid by running Celera's benchmark fastest.

"We want to put together something that's an outstanding tool for life sciences," says Ty Rabe, Compaq's director of high-performance technical computing. The Alpha architecture excels at floating-point and integer operations, he says, and it performs well on jobs that require matching patterns of data, integral to genome-sequencing. "Biology is going to rip supercomputing out of the government labs and bring it into the commercial sector," says Celera's Peterson. "One thing we can't afford to do is ignore anyone's promise of technology."

Private industry spent $179 billion on R&D last year, an increase of 8.5% per year since 1994, according to the National Science Foundation. That's two-thirds of the $264.2 billion in overall R&D spending by companies, the government, and universities. Of that, 17.2% went to basic research, 21.5% to applied research, and 61.3% to development.

Corporate research labs are responsible for most of the inventions that define modern computing, and the home runs have been plentiful. IBM was first to market with magnetic hard disks, Xerox's highly successful DocuTech copiers sprang from PARC, and AT&T's Bell Labs research arm--now part of Lucent--developed the transistor, the laser, and cellular wireless technology.

But famous missteps are also common. Despite inventing the graphical user interface, Ethernet technology, and client-server architecture, Xerox never brought a commercially successful PC to market. IBM's advancements in RISC architecture and relational databases sat in the hopper until HP, Oracle, and Sun Microsystems capitalized on them. Lucent was slow to turn its inventions into money-making products, and a more recent decision to dial down investment in optical networking gear cost it the lead in high-speed data transmission to rival Nortel. Now companies, particularly those under financial pressure, realize that being fast--and first to market with new technologies--is critical. "Bringing [innovations] to market quickly is the No. 1 objective of my organization," says Lucent chief operating officer Jose Mejia. "To minimize the cost of development and maximize product opportunities is big for Lucent now."

Making research more agile is even more urgent at Xerox. With losses mounting, the company is selling assets, firing workers, and reevaluating its operations. Xerox plans at least 4,000 layoffs this year, though the number could reach 10,000, an executive said recently. As part of a top-to-bottom overhaul, Xerox is working to attract outside investors in PARC that can help the company commercialize ideas while shouldering some of the costs.

After a decade without substantial competition in the high-end, production-copier market, aggressive competitors including Canon, HP, and Ricoh have encroached on Xerox's turf. Meanwhile, Canon, IBM, and German printer Heidelberger Druckmaschinen have come up with a rival to Xerox's DocuTech that's more compatible with third-party software--long an Achilles' heel of Xerox. Associate center director Mark Bernstein says PARC technology can help Xerox compete on this front--if the company creates the capacity to absorb it. "PARC's view of the world has really been crafted by the view of 'The Document Company,'" he says. In that worldview, compatibility with other vendors' systems wasn't a priority. "We're ready to do something new and challenging."

Illustration by Rick Sealock
Photo of Peterson by D.A. Peterson
Photo of Horn by Melanie Weiner

continue on to page 2

E-mail To A Friend |  Printer-Friendly |  Send Us Your Feedback

Home | This Week's Issue | Workplace and Careers | Resource Centers | Research