Hewlett-Packard is banking on a handful of research projects in a few high impact, Internet-oriented areas to yield big results, produce new products, and deliver revenue to the bottom line instead of funding R&D on many small projects.

Under the new approach, announced Thursday, 150 small research projects will be consolidated around 20 to 30 "big bets" that HP will pursue and try to make pay off, said Prith Banerjee, HP's lanky director of research now seven months into the job.

The announcement had the air of a work in progress as Banerjee said he couldn't disclose what internal research projects would be favored and which choked off. That information was "too commercially sensitive," he said. An internal review committee of both technology and business managers was tackling the issue, he said.

"The plan we have announced sharpens the focus on high-impact research," he told 40 members of the press gathered at HP's Palo Alto, Calif., headquarters.

HP is starting to form partnerships with major research universities, venture capital firms, and government agencies to get feedback on what are the most promising research avenues. "Not all the smart people are working for HP Labs. By forming partnerships, we will know what the trends will be," Banerjee asserted. HP now has a working partnership with Foundation Capital venture capital. "More are coming," Banerjee promised.

HP CEO Mark Hurd made a cameo appearance at the HP Labs new direction unveiling. The new direction "is expected to have a big impact on HP. We will be making innovation that is meaningful," he said.

Both Banerjee and Shane Robison, HP's executive VP and chief technology officer, outlined areas where future research will be focused.

They project that use of computing devices must seek to become seamless, so that users are transferring information and working within a familiar interface as they move beyond their offices. Information will continue to explode, so capturing and storing information will be less important than analyzing and delivering the right information at the right time.

Content will need to flow from device to device and from digital format to analog and back again, "as a seamless experience," said Banerjee.

"Cloud computing is taking off," said Robison. Instead of depending on faster and faster personal devices, users will start to depend on the network cloud to deliver personalized services, based on where they are and what they're doing, he said. "Software as a service is just getting started," he predicted. In the future, "everything is delivered as a service."

Robison illustrated the point by saying a family won't necessarily buy a cookbook online. They will design one -- perhaps starting with a 1906 classic that includes a recipe for "baked squirrel" -- and build out a book that captures family favorites since then. The book would be published and delivered to the household through editing and interactions over the Internet.

HP is going to try to find new ways to tap the insights and intuitions of people outside its halls by calling for ideas or requesting a design from a group that may not be known to HP but is rated highly by other users of its services. "Crowd sourcing," where professional services are solicited on the Internet, is "going mainstream," Robison predicted.

Robison said HP will continue to acquire startup companies that represent cutting-edge ideas and services as well as coming up with them in its labs. "We'll do both," he said, pointing to a string of recent HP acquisitions, such as Snapfish, the online digital photo printing service. But it wants to come up with more basic technologies like Precision Architecture/RISC that lead to a long line of servers, and ink-jet technology that lead to long lines of printers.

Banerjee said the amount HP will spend will be about the same as what it's spent in the past. The goal isn't to spend more, he said. The goal is to get better results and get them to market.