More than a dozen companies-including International Business Machines Corp. and American Telephone and Telegraph Co. -- are expected to join with government and university researchers to develop new ways to transfer information from one computer to another at speeds hundreds of times faster than now possible.

The joint project was announced yesterday by officials of the National Science Foundation (NSF). The project eventually is expected to attract a far larger number of participants than is usual for such collaborative research efforts.

"We're all excited and challenged by this great effort," said Robert Lucky, an executive director at AT&T Bell Laboratories. "To us, it's like they're saying, let's go to the moon in 10 years."

Researchers for years have been exploring the possibility of building an extremely high-speed nationwide data network, but yesterday's announcement will give it more money, more visibility and a greater degree of coordination.

The NSF chose a small Reston firm, the Corporation for National Research Initiatives (CNRI), to oversee the work done by the companies, three supercomputer centers, several national laboratories and at least eight universities.

The government's financial commitment to the three-year project -- $15.8 million from the NSF and the Pentagon's Defense Advanced Research Projects Agency -- will be used to fund university and national laboratory research, while the participating companies are expected to make financial commitments reaching several times that amount, the NSF said.

Firm agreements with the participants are expected to be signed within several weeks, according to CNRI President Robert Kahn.

Driving the project is a vision of a "wired" nation -- one in which any business, school or home could receive information or video images from distant locations over networks transmitting more than 1 billion bits of information each second.

While networks are already in use by researchers, a host of technological breakthroughs must be made before they will operate several hundred times faster than they now do.

For example, new technologies must be developed to switch the information traveling along the networks so it arrives at the right place. Also, computers must be altered so that they can receive so much data at one time without garbling it.

Researchers for years have been exploring many of these technologies, but until now have not had a laboratory -- in this case a network itself -- on which they could test how the different approaches would fare and interact.

"Everybody has been working on it alone," said Alan Baratz, manager of communications and concurrent systems at IBM's research laboratories. "There's no way this can become a reality until we come together."

Officials involved with the project emphasized that, in terms of putting an operating network in place, the three years of research would barely scratch the surface.

A report issued last year by the White House Office of Science and Technology Policy said it would require federal spending of $400 million over five years to develop a high-speed network connecting research centers by the year 2000.

Staff writer John Burgess contributed to this report.