Despite HP's declared commitment to keeping its investments in research and development intact despite the 14,500 job cuts, the company has decided to cut four research projects from its HP Labs organisation.

The projects include an effort to develop a next-generation operating system for the Internet, work on high-density portable storage devices and research on technologies for health, said Dave Berman, an HP spokesman. Details of the project cancellations were first reported by the San Jose Mercury News.

The company has decided to focus on research projects that have the best chances of paying off for HP in both the short and long term, Berman said. For example, HP Labs' work on developing a new "crossbar latch" transistor design using carbon nanowires will continue, as will other projects focused on quantum computing, he said.

HP is also changing the way its research efforts are organised. All research groups will now report to the head of HP Labs, Dick Lampman, Berman said. Previously, some of the work at HP Labs reported directly to individual research centres, such as the Cambridge Research Laboratory in Massachusetts and the Advanced Software Research programme, he said.

Some of the projects within those laboratories will transfer to other departments within HP Labs, along with some of the researchers working on those projects, Berman said. But the heads of individual research centres, such as the Cambridge Research Laboratory, are leaving HP, he said.

HP spent about US$3.5 billion on R&D in its 2004 fiscal year. Advanced research projects have traditionally been very important to the company, following in the footsteps of founders William Hewlett and Dave Packard. But 2004 research spending was down from the prior year, as HP struggled to generate profits across its businesses.

HP CEO Mark Hurd promised to keep spending money on research projects when he outlined his restructuring plan for the company on Tuesday, saying that HP's work on advanced research sets its apart from its rivals in the industry.