Judy Estrin is happy. And why not? She has put behind her the hard-driving corporate life at Cisco Systems, the world's largest network-equipment group, where she spent two years as chief technology officer.

Instead, with the foundation of Packet Design, her fourth company, she is back in the sort of start-up she likes most.

"It's great to be back doing what I love to do, building something new," she said.

Her departure from one of the world's most dynamic companies might be taken as a sign of sated ambition. After all, Ms Estrin is already a very wealthy woman who could be sitting on her own tropical island.

In fact, as head of a start-up she has returned to the "trenches" in what is one of the most exhausting executive roles, according to many who have held similar positions. Moreover, Ms Estrin is aiming to make Packet Design a blueprint for a new sort of start-up, neither venture capital fund, nor internet incubator, nor research institute but something in between all three.

For Ms Estrin, creating a startup is a return to her entrepreneurial roots. And very distinguished roots they are too. In the meritocratic environment of Silicon Valley, she is the closest thing you get to royalty.

To a large extent, she earned this standing through her own hard work.

Armed with a degree in mathematics and computer science from the University of California and a master's degree in electrical engineering from Stanford University, she founded her first venture, Bridge Communications, a computer-network company, in 1981 and sold it six years later to 3Com, the large US network-equipment maker.

The following year came Network Devices, which produced computer hardware and went public in 1992.

She then started Precept Software, a multimedia-network software company in 1995, and sold it to Cisco three years later. It was said at the time that Mr John Chambers, Cisco's chief executive, bought Precept because it was the best way of getting the talented Ms Estrin on to his senior executive team.

At Cisco she headed the company's frenetic mergers-and-acquisition strategy. With 60 deals under its belt since 1993, M&A has been Cisco's chief engine of growth. Ms Estrin has done as much as anyone to turn the company into the colossus it is today.

But Ms Estrin's rise is not entirely her own doing. Mr Vint Cerf, a member of the small band of pioneers that created the basic technologies that support the Internet, will sit on Packet Design's technical advisory board. Mr Cerf worked with Ms Estrin's father and helped her with her master's degree at Stanford University. Her sister, Ms Deborah Estrin, a well-known researcher into network technology, is also on the board. As befits Valley royalty, Ms Estrin has collected a few powerful friends along the way.

One of Packet Design's investors is Mr Bill Joy, the chief scientist at Sun and father of the Java programming language. Another is Mr James Barksdale, who headed the Internet software company, Netscape Communications.

Indeed, Packet Design is almost a family affair. Ms Estrin is getting back into business with her husband, Mr Bill Carrico, in all three of her previous start-up companies. Many people would be horrified to work with their spouse, but Ms Estrin seems happy to unite business with the personal.

"That was one of the things I missed the most at Cisco," she said, "not being able to work with Bill."

As Ms Estrin negotiates a mound of cardboard boxes containing computer monitors, and surveys the half-formed space of Packet Design's office, she seems to draw energy from the unrealised potential of her latest business venture. In particular, she relishes the idea that she once again controls a company's destiny.

"At Cisco, the culture is such that senior management is through influencing the heads of the many different business units.

"I couldn't directly make things happen, which I found frustrating," she said.

But she also likes the demands of a start-up. She becomes animated as she describes what she wants to do, now that she is free from the restrictions of Cisco.

"Breadth is very important to me rather than depth, doing just one thing," she said.

"I enjoy being able to talk with our chief scientist on technical issues, then switch to meeting with our lawyers, then discuss financial issues with others."

Packet Design could test this fondness of start-ups to destruction. The business model of the company demands that she will be in almost perpetual start-up mode.

Although her friends have labelled her a "serial entrepreneur", Packet Design will make her as she puts it a "parallel entrepreneur".

All of the start-ups will share the common aim of developing important Internet technologies.

Packet Design will apply these technologies to critical problems that limit the scope of the Internet. The company will either license the technologies to other companies, or transform them into new subsidiaries. Ms Estrin said her company is not quite an "incubator", an operation to churn out start-ups, its progeny has a more direct descent than that.

Neither is it a research laboratory, like the legendary Xerox Palo Alto Research Centre (Parc), which developed innovations such as the laser printer and the graphical user interface, its research is, she said, more applied than Parc's.

If such distinctions sound grandiose, Ms Estrin is the first to play down her new venture. "I don't like to overset expectations. I've learned that it can set you up for a big disappointment."

Whatever its nature, Packet Design seems adapted to compensate for the one obvious weakness that Ms Estrin has demonstrated in her career.

According to a former longstanding colleague: "Judy is great at starting companies, but she doesn't seem to like staying around for the long haul.

"You can see that with all of her previous start-ups. After establishing a company, she gets restless and wants to move on."

The test will come if Packet Design becomes successful.

When it is no longer a start-up itself, but merely a company that makes start-ups, will it still satisfy the queen of the Valley?