IBM `Velociraptors' Modeling Monkey Brains in Quest for Sales
In a small, windowless room in the basement of International Business Machines Corp.’s Almaden lab in Silicon Valley, Dharmendra Modha is building a simulated monkey brain using images projected on flat-screen TVs.
Modha, the lab’s manager of cognitive computing, urged Big Blue to back his pet project, which combines advances in nanotechnology, supercomputing and neuroscience. If successful, the effort could help computers “think” like the human mind.
IBM green-lights proposals from Modha and other researchers at the San Jose, California, facility because some of the gambles can spawn significant business. Since the 1990s, lab advances in how data is stored, managed and analyzed have helped generate millions of dollars in sales. Yet even lab directors concede chances are slim that any one breakthrough will make a sizable addition to IBM’s almost $100 billion in annual revenue.
“We have no way of making money off of this -- unless a miracle happens and we figure out how to make better computers,” said Laura Haas, director of computer science at Almaden and the lab’s second-in-command, who wears a “Question Authority” button. “We’re different. It takes a different kind of craziness here.”
While IBM may move at a slower pace than smaller West Coast startups, it aims to give Almaden researchers as much leeway as their Silicon Valley peers, said Winfried Wilcke, senior manager of the lab’s nanoscale science and technology research.
“If you think of us as dinosaurs, then think of us as velociraptors,” he said, referring to the nimble predators of the Late Cretaceous period. “The freedom we have here is great.”
Silicon Valley labs have a mixed record of getting a payback -- a high return on investment -- from their discoveries. While IBM was able to capitalize on Almaden’s invention of the disk drive and relational database, many of Xerox Corp.’s early innovations at its Palo Alto Research Center -- such as the Macintosh-style graphical user interface -- brought more benefit to other companies.
“Making some bets that are unlinked to specific outcomes is important, but if it’s all kind of curiosity-driven research, the ROI on innovation investing may not be that high,” said John Kao, chairman of the Institute for Large Scale Innovation in San Francisco.
Some investors say IBM needs big deals more than research to increase revenue. While the company has boosted earnings per share through such steps as buybacks and cost controls, rivals Oracle Corp. and Hewlett-Packard Co. have eclipsed its stock- price performance in part through big purchases that added to sales.
Since Sam Palmisano became chief executive officer of IBM in 2002, it spent $25 billion buying publicly held companies. Compare that with at least $45 billion for Hewlett-Packard and $42 billion for Oracle. IBM stock has risen 25 percent in the Palmisano era, compared with 53 percent for Oracle and 128 percent for Hewlett-Packard.
Palmisano said in May that he plans to spend $20 billion more on acquisitions in the next five years. On Palmisano’s watch, IBM also has increased its research spending by 31 percent. Last year, it devoted $5.8 billion, about 6 percent of annual revenue, to R&D.
The Silicon Valley version of IBM fosters a freer attitude and more of a startup mentality than headquarters in Armonk, New York. While the century-old IBM is known for white shirts, dark suits and button-downed attitude, the staff at Almaden tackles oddball research projects and unwinds by holding sack races and tug-of-war competitions.
Located on a hill in a 1,600-acre wildlife preserve, the Almaden lab is painted a custom dark-green color derived from leaves and grass that employees gathered on site. The lobby boasts a 5-foot-diameter digital display globe that was purchased after researchers saw a similar device at a Denver museum and thought it would be “cool” to have one.
“New York was like, ‘You crazy scientists can do this, but come August, you have show us results,’” said Julia Grace, a social and collaborative computing engineer. The globe has been used to plot last year’s swine flu outbreak and more recently to show the location of people sending messages delivered via Twitter Inc. during World Cup soccer matches.
Almaden’s projects include an electric battery for automobiles that could run 500 miles on one charge, a filtration system for desalination and a program that shows changes in geographical data. Researchers can shake up an industry, even if they don’t make a fortune, Haas said.
“You’re not going to get rich, but you can change the world,” she said. “IBM is a big ship, and you need a certain amount of patience to turn the wheel. But if that ship turns, the wake in the industry is enormous.”
Employees at the lab consider themselves “wild ducks” for working on projects that don’t fit in with what headquarters in New York wants to pursue, she said.
Almaden has a “better sense of ‘the next big thing’” because of its location, said Mark Dean, IBM’s vice president of strategy, who works at the Watson lab in Yorktown Heights, New York.
“It’s the only place in IBM where we’re doing advanced battery research,” he said. “The only place we’re doing water desalination.”
Of IBM’s nine research centers, Almaden also is the only one to hold an annual Olympics-style event, where IBM Fellows -- the highest technical honor a researcher can receive at the company -- compete with interns in athletic events. Competition for the first-place trophy is so fierce that research team directors taunt each other in self-directed videos by breaking wooden boards with their hands or donning Star Trek uniforms.
“People expect more product impact from their research labs than they did 20 to 30 years ago,” he said.
In the 1990s, Almaden developed a method of encryption for media content. Used in memory cards and Blu-ray discs, this technology has earned IBM millions of dollars in licensing fees.
Giant magnetoresistive heads, introduced in 1997, helped hard disk drives increase their data-storage capacity 40-fold. A query system developed at Almaden in the late 1990s, named Garlic, integrates data from multiple sources and has earned IBM hundreds of millions of dollars in revenue. More recently, the lab developed a file-management system called GPFS that IBM now adds to its archiving, networking and cloud computing products.
Still, technology companies continue to fund research without knowing where it will lead. Mountain View, California- based Google lets its workers spend 20 percent of their time working on personal projects.
“We’ve found that people are more productive when they are working on projects that really excite them,” said Jordan Newman, a Google spokesman. Products including Google News and Gmail got their start that way.
At IBM, some projects don’t pan out. In the early 1990s, Haas and IBM’s database management team wanted to figure out a way of searching the World Wide Web, something that hadn’t emerged yet.
“If we did it right, we might have invented search before Google,” Haas said. “But we had the wrong model, and we totally missed the boat. We were thinking about librarians.”
Modha, the researcher working on the monkey brain, quit IBM a decade ago in order to work on his own startup. After a short time, he realized he didn’t have the financial acumen to run a business and returned to Almaden.
“We have a 100-year history,” he said. “The challenge is, how do we usher in the next 100 years?”