Profile: Prith Banerjee

Some baseball players are known for being consistent, if not flashy, hitters. They get on base and keep the game going. Then there are the sluggers, who bring the crowd roaring to its feet every time they step up to bat. You know they’re going for that home run.

When Prithviraj Banerjee arrived at HP Laboratories, in Palo Alto, Calif., as the organization’s new director in August 2007, he was intent on pushing the researchers to swing big. HP Labs, a sprawling enterprise with some 500 researchers spread over seven locations worldwide, is the company’s advanced research arm, spending about US $150 million annually. The first commercially available LED, the pocket scientific calculator, thermal inkjet printing, and reduced- instruction-set computer architecture were all born in the labs. The labs got the company into digital printing, and HP is counting on those researchers for future wins.

In recent years, company executives had grown concerned that the labs had lost their focus. When Banerjee arrived, there were 150 projects under way, most of them involving just a couple of researchers and having no overarching goal. He didn’t know yet which ones had home-run potential; he just knew he had to focus on those that did and forget about the rest.

And he had to move fast. Since his arrival he has flattened out the labs’ hierarchy, restructuring the organization into 23 laboratories, each with a leader and a single research focus. He asked researchers and lab directors to come up with 20 “big bet” projects that would advance the state of the art of technology and have a significant impact on HP’s businesses now and in the future.

This bold strategy didn’t immediately win everyone over. Under Banerjee many longtime employees were forced to set aside projects they had spent years on. Many people wondered why he had been tapped for the job at all.

HP has a history of promoting its management from within. Not only had Banerjee never worked for Hewlett-Packard, he had never worked for any large corporation. He’d spent nearly his entire career in academia, with a brief detour to start a software company. Banerjee was the ultimate outsider.

A year into his tenure at HP, Banerjee has won over some skeptics. Crawford Del Prete, executive vice president of worldwide research at International Data Corp., a market research and analysis firm, says that Banerjee has “brought a level of focus to HP Labs that before didn’t exist. I see a number of focused teams working away, and far more cooperation between product groups and labs than I ever saw before.”

And even his critics can’t question Banerjee’s energy, dedication, or passion. Whether those qualities and the changes he’s set in motion will be enough to put the labs in a position to score big for HP is not clear.

Of course, Silicon Valley is a culture of outsiders, drawn from all over the world and all walks of life, united by their common passion for technology. In that sense, Banerjee fit right in. Born in Sudan, he spent much of his childhood in the Himalayan foothills of Bhutan, to this day one of the most isolated nations in the world. The nearest school was a 2-hour hike through steep hills, so Prith studied at home. In his middle school years, the family settled in Calcutta, where Banerjee began more formal schooling. He was always tinkering with electronics, and in high school he won the top prize at a national science fair with a widget that measured the flow of water in a pipe using an external transducer. From an early age, he also liked to be in charge. His older brother, Sanjay, remembers 8-year-old Prith once shoving his father onto a train, impatient with his father’s extended farewells.

Prith entered the India Institute of Technology, in Kharagpur, as an electronics engineering major in 1976 and then went on to the University of Illinois at Urbana-Champaign for a Ph.D., where he focused on fault tolerance for multi processor computers. After getting his doctorate in 1984, Banerjee entered a booming job market, with more than a dozen job offers from the likes of Bell Labs, IBM Research, Carnegie Mellon University, and the University of Michigan. He chose to stay at Urbana but changed his research thrust to parallel algorithms for computer-aided design. Within eight years he was a full professor.

In 1994 Banerjee got his first taste of management. At the time, the university housed one of four national super computing centers. The College of Engineering wanted a graduate program to take advantage of this resource. Two early attempts had failed. Banerjee wanted to take another crack at it. He rallied support among students and faculty and within two years had 10 departments involved, 100 graduate students signed up, and several million in grant money committed.

“I learned in doing this,” he says, “how to build support for an idea from the bottom up, because if it comes from the top down, people in the academic world will hate it.” Even as he nurtured consensus, Banerjee didn’t hesitate to step in at key moments and take charge. He was passionate about seeing the project through. In fact, “passionate” is typically the first word that comes to people’s minds when describing Banerjee. In recent years, he says, he has tried to moderate his more passionate moments, at the urging of his wife. “She thinks I sometimes come off as a dictator,” he says with a chuckle.

Banerjee left the University of Illinois in 1996 to take on a new challenge at Northwestern University, in Evanston, Ill. At the time, Northwestern’s graduate engineering program ranked so low, Banerjee recalls, “it wasn’t even on my radar. But the undergraduate school was fantastic.”

As a newcomer, he didn’t feel he had the knowledge or the right to simply make sweeping changes. Instead, he began holding regular meetings with faculty members to hammer out a strategic plan. “It was the first time I tried doing strategic planning,” Banerjee says, “so it was not a very well-thought-out thing.” He learned that Northwestern professors tended to work alone. “They’d go off and do something with one or two graduate students, and no one would bother them.” Having autonomy was great for the researchers, but their results hardly made a ripple. An important idea emerged: To make an impact, the department needed to do a few big collaborative projects instead of lots of individual ones. Banerjee also began recruiting star faculty from other universities.

“It was a fun time,” he says. “We had all these top people coming in, we got five large DARPA grants, and our [graduate engineering] ranking, according to U.S. News & World Report, moved up to 15th or 16th.” (Currently it ranks 21st.)

Banerjee was directly involved in one of the Defense Advanced Research Projects Agency efforts, the creation of a software tool that would greatly simplify chip design. It would take a file created using the popular Matlab design software and spit out the corresponding design. Without such a tool, design engineers had to manually convert the Matlab chip simulation into the hardware design. Banerjee, working with another faculty member and two graduate students, developed a prototype and demonstrated it to DARPA. He called the effort MATCH (for MATlab Compiler for Heterogeneous computing systems).

It was 1999, and “everybody and his dog were doing a dot-com,” he recalls. “I thought, ‘I have a real technology, I should be able to do something.’”

Starting a company was something Banerjee knew nothing about. But after a childhood spent educating himself, his first impulse was to hit the books. He spent six months studying and then wrote his business plan. He sent it out to a long list of venture capitalists, and he quickly got an offer of $2 million from an East Coast firm. The venture capitalist told Banerjee to start hiring people, lease an office, and buy computers, because the money was on its way. Banerjee did all that and also retained a Chicago lawyer to represent the company. Toward the end of 2000, he booked a ticket for the day after Thanksgiving so that he could fly East and sign the paperwork.

The day before Thanksgiving, the firm pulled out of the deal. The dot-com bubble had burst. And Banerjee had just committed $200 000 that he didn’t have.

After spending what he recalls as the worst Thanksgiving in his life, he decided to go ahead with his start-up anyway. He took out a second mortgage to pay the bills and began shopping the product around to customers, who told him it was a good idea but off target. The MATCH prototype supported a type of field-programmable gate array used in defense but not in commercial applications. Banerjee and his team started modifying the prototype. He also began making regular trips to Silicon Valley to talk to companies that might be interested in buying MATCH if it ever got to market.

If he was worried, he didn’t show it. Says David Zaretsky, then a graduate student at Northwestern who worked on the start-up, “You couldn’t tell that anything was bothering him. He remained cool and calm.”

Finally, in February 2001, Banerjee got an order from a customer that promised to pay $2 million over three years. With that in hand, he was able to attract a new venture capitalist to fund the company, which he named AccelChip. He took a one-year leave from Northwestern to get the company off the ground. The AccelChip Compiler came out in 2002; early customers included Elixent, Motorola, and Quicklogic. Four years later, Xilinx purchased AccelChip for an undisclosed amount rumored to be around $21 million, most of which went to investors.

Banerjee returned to Northwestern in September 2002, although he continued to serve as AccelChip’s chief scientist until the Xilinx buyout. Straddling business and academia is common for professors in Silicon Valley; it wasn’t in Chicago. “I was the guinea pig,” Banerjee says. Since then another 9 or 10 engineering professors at Northwestern have followed in his footsteps.

By 2004, Banerjee was restless and began applying for engineering dean positions. Three schools made offers: Iowa State University, Johns Hopkins University, and the University of Illinois at Chicago (UIC). He accepted the job at Johns Hopkins, but his family balked at moving during his son’s high school years. Banerjee instead went to UIC, another school without a top-ranked engineering program.

Again, Banerjee got his faculty to work on a strategic plan. For many of them, it was an eye-opener. “Most colleges have strategic plans that are 5 or 10 pages long,” says Peter Nelson, who was a department chair at UIC under Banerjee and is now dean. “Prith locked 20 or 25 of us into a room for a few hours a day, several days a week, and we came up with a 256-page detailed plan.”

Banerjee again pushed the ideas of doing small numbers of large-scale research projects. Meanwhile, he was getting calls from other universities. In 2007, with his son out of high school, Banerjee began to consider the offers seriously. “I wanted to move up the academic ladder,” he says. He was thinking provost, maybe even president.

Then in November 2006 he got a call from a search firm hired to identify candidates to take over HP Labs. The headhunter knew Banerjee was looking at academic posts but asked if he’d consider the HP job. “Absolutely not,” Banerjee replied.

The recruiter persisted. He flew out to Chicago, and once there he asked Banerjee to meet him at the airport. Banerjee said he’d give him an hour.

“I’m not interested,” he said after hearing the pitch. “I am going to academia. My dream is to be a college president.”

Perhaps he wasn’t interested, but the headhunter badgered him into agreeing to come to Silicon Valley for one more meeting—with Shane Robison, HP’s chief strategy and technology officer.

Robison says he was looking for someone with academic credentials, start-up experience, and “a fresh perspective.” Banerjee had all those qualifications—but so did a number of other candidates. The CTO scheduled just an hour with Banerjee at Hobee’s, a popular brunch spot in a strip mall across from Stanford University. “Two and a half hours later I had to finally break off to go to another appointment,” Robison recalls. “He was that interesting.”

When Banerjee returned to Chicago, he asked his wife, Swati: What if he got an offer from HP—would she move to California? To his surprise, Swati was ecstatic. She had always dreamed of living in California, she told him.

Still on the fence himself, Banerjee then called his brother. Sanjay was happy with an academic career himself—he’s at the University of Texas at Austin. But just after grad school, he’d spent a few years in Texas Instruments’ corporate laboratories and felt his brother might thrive in such an environment.

“The HP job was an absolute godsend,” Sanjay says. “Of course, it was financially lucrative, but more, it’s the kind of position in which you can do something real. In academia, without the market push, we publish paper after paper. In corporate R&D, you can do cutting-edge research that influences society.”

On 1 August 2007, Banerjee started work. As he’d done at Northwestern and UIC, he started with developing a strategic plan. But this time the discussions, which involved 80 people from the labs and 40 from HP’s business units, bogged down. “The process drove me and the entire organization crazy. Mark Hurd [HP’s CEO] told me I was wasting too much time,” Banerjee says.

Meanwhile, much of the research ground to a halt. Banerjee had expected as much. “I told Mark and Shane that for a transformation the size of which I’m trying to accomplish, we had to stop and really think about what we should be doing,” Banerjee says.

After two months of back-and-forth during which little was settled, Banerjee stepped in and announced his vision, which included sweeping away layers of hierarchy. Under the director were four centers, under the centers were 12 labs, under the labs were 24 departments, under these 24 departments were a total of some 150 projects. Banerjee envisioned a flatter, more focused organizational chart with 20 projects at most. He didn’t call for layoffs. (By the time Banerjee started, the labs were already smaller than they had been in the past, with about half the number of researchers as there had been in 2000.)

Distinguished Technologist Parthasarathy Ranganathan, who worked as part of the planning group, said the labs’ previous philosophy was to “let a thousand flowers bloom. But Prith said, ‘My garden doesn’t have room for all these roses,’ and once we got the sense of what kinds of flowers he liked, we started pruning.”

The final list of laboratories came to 23; the structure went into effect in December 2007. Bob Tarjan, one of the researchers, wrote what he called a stable-marriage algorithm to match lab directors and researchers, who had ranked their preferences. By 15 March 2008, the new labs were in place, but with only general missions. Projects were to be selected from researchers’ proposals by a review board that included representatives from the business units as well as the labs; each year 20 percent of the projects would be replaced by new ideas.

To observers, this move optimized the research structure to focus on existing businesses and was not that shocking. Lab operations, points out a former researcher, have always been cyclical. “Sometimes,” he recalls, “Hewlett and Packard would tell us to focus on supporting the businesses, that HP had enough on its plate and wasn’t looking for new lightbulbs. Other times they would look to the labs for new footprints.”

Banerjee thought it a fine plan. Many of the researchers, though, hated it. In principle, most were willing to give up their individual projects and focus on a smaller number of large projects. But in practice, there were a few pet projects and research paths they were loath to see go.

He realized that the researchers wanted greater autonomy. Google, he recalled, lets its researchers spend 20 percent of their time on whatever they want. He decided to follow suit, and in addition he permitted 20 percent of the people in each lab group—6 out of a typical lab’s 30—to work full-time on whatever they wanted. (Interestingly, it was HP founder William Hewlett who was an early champion of this type of policy. He called his plan “schedule-free Fridays.”)

By October 2008 the review board had selected its list of projects. Banerjee had been looking for about 20 big bets; what he got were 22 large projects and 40 medium ones. Some were entirely new, and some encapsulated projects that had been canceled. “We came a long way, but we were not quite there,” he says. Now, a year later, after some more judicious trimming, the projects number 21.

Banerjee explains: “It’s as if we are in an auto company, and we have efforts of tire research, body research, and so on. I’m asking people to focus on the vision. What is the car you are trying to build? A fast car? A fuel-efficient car? I don’t know. They can tell me what they want. I just want them to focus.”

For the first time in a long time, the labs have that focus, says analyst Rob Enderle, of the Enderle Group. He gives Banerjee an A– for his work so far. “He’s made a difference. He got costs under control; he got the labs focused. The groups I’ve met with seem to have a higher morale than before; they know that no ax is going to fall soon. He isn’t going to get a higher grade with me until we see products in the market, but that takes time.”

Banerjee’s radical restructuring may have been tumultuous, but the labs needed it, says HP CTO Robison. “The most effective way was to do it quickly. A big bang approach.”

It’s too soon to tell if Banerjee’s big bang approach will pay off—advanced research organizations measure their results in decades, not months. But Banerjee is confident. “I am going for the big bet, the home runs,” he says. “There is a huge risk, but I just keep swinging—boom, boom. A few of them will make it.”

This article originally appeared in print as “Go Big or Go Home.”