HP, like many other big-name tech companies, is bent on wringing more revenue out of its state-of-the-art research-and-development labs--a collection of 500 researchers scattered across seven countries, including China, India, and Russia. Created in 1966, the Palo Alto, Calif.-based operation was once responsible for such landmark inventions as the pocket calculator and inkjet printing. But its output has disappointed of late, with projects dragging on for years without producing any meaningful discoveries. So last May, HP kicked off an overhaul of the labs.
One year into that effort, HP Labs provides an illuminating case study of the difficulties companies face in trying to inject discipline and accountability into the innovation process.
To spearhead the transformation of its labs, HP brought in a new director from outside the company. Since taking the reins last May, Prith Banerjee, a former engineering professor and founder of two chip-design companies, has worked to rid HP Labs of some bad old habits, such as funding projects based on personal relationships. He introduced a formal review board that is responsible for greenlighting projects.
And while the labs' budget has remained steady at around US$150 million a year, Banerjee has slashed the project count from 150 to 22. The ones that made the cut were those that promise big payoffs--either in terms of internal savings or sales. "We are not running a charity here," says Banerjee. "We're looking at an impact of no less than US$1 billion [in market opportunity] per big project."
One iniative that survived the shakeout is CloudPrint, an HP technology that lets business travelers print out e-mail attachments on any printer without the hassle of having to fiddle with cables or having to install drivers. On May 4, Research in Motion announced that it has licensed CloudPrint for use with its BlackBerry devices. The software will be available as a free download for BlackBerry users or via corporate IT departments.
Although terms of the deal were not disclosed, it's a good bet that licensing revenues from CloudPrint will fall well short of the US$1 billion blockbusters Banerjee is targeting. Nonetheless, the technology has promise and may eventually also find its way to places such as Africa and India, where many more people own cell phones than PCs. "We're trying to remove friction to print," says Patrick Scaglia, chief technology officer at HP's printing group, which generates 21 percent of HP's revenue, and 36 percent of operating profits.
Improvised innovation
CloudPrint is the baby of Bernardo Huberman, one of Hewlett-Packard's top computer scientists and director of its social computing lab. The project, which got under way before Banerjee landed at HP, is a good example of the sort of improvised innovation that was the hallmark of HP Labs. Says Huberman: "We did this almost the way we would do a startup."
Huberman's team began work on the project without first getting the approval of a committee or doing lots of market research. And when it came time to show off their results, the crew bypassed HP's technology transfer office, another Banerjee innovation, and went straight to the top brass. Working his connections at headquarters, Huberman secured an audience with Scaglia. It wasn't a very polished presentation, either. Jerry-rigging a printer salvaged from the trash to a laptop surrounded by foam core, Huberman & Co. assembled a printing kiosk modeled after a photo of a Wells Fargo ATM that one of the team members had snapped.
To hear Scaglia tell it, though, the makeshift setup didn't detract from the demo. "It's one of those ideas where once you see it, you say, 'yeah, of course,'" says Scaglia, who was present at the demo.
Managing the tension between conducting corporate research by the numbers and recognizing the value of "a-ha" moments will be key to HP's goal of increasing technology transferred out of its labs. Shaw Wu, a senior analyst at Kaufman Brothers , notes that while licensing revenue from products like CloudPrint is a drop in the bucket for a company with US$118 billion in annual sales, technologies that emerge from HP Labs can give the company advantages in PCs, printing, and other key markets. "If they're able to out-engineer Dell and Acer, not only does it drive out costs, but they can change their strategy," he says.
Banerjee is bullish on a new user interface for PCs being tested in HP's Bangalore (India) lab that lets computer novices use hand gestures and voice commands to control machines. "It's extremely cool," he says. The interface could find a home in the PC division's "customer experience group," which is crafting new ways of interacting with computers. Another team is working on thin, flexible displays the company believes it can use to supply future screens for laptops, which would allow HP to control component costs and source materials more cheaply than rival Dell.
And mathematical operations research born in its labs has helped HP's personal-systems group optimize the way it buys PC components, saving US$300 million in inventory and supply-chain costs over three years. Banerjee's goal now is to increase by an additional 50 percent the number of tech transfers from HP Labs to product groups, technologies licensed to other companies, and academic papers published by HP scientists. He concedes that transforming the formerly slow-moving labs will take time. "It's not like tomorrow we flip a switch," he says.
Banerjee has already hit some roadblocks. A year after announcing a partnership with Foundation Capital intended to connect HP with startups that could commercialize technology from its labs, the two entities are still trying to reach an agreement on a formula for revenue-sharing. And Banerjee's results-driven approach to research is bound to vex some HP engineers. Indeed, as Huberman's work on CloudPrint shows, the new boss may have a tough time yoking some lab veterans. Looking back over his CloudPrint project's ad hoc trajectory, Huberman mused: "I don't know if this subverts the whole process."
Play video
There are currently no comments for this post.