As a result, there has been a pronounced swing back towards a centralised computing model in the past few years, accelerated by the growth of the internet.

The internet has its origins in the 1970s and work undertaken by Vinton Cerf and otters to design systems that would enable research and academic institu­tions working on military pro­jects to co-operate.

This led to the development of the Ethernet standard and TCP/ IP, the basic internet protocol. It also led Bob Metcalfe to promul­gate "Metcalfe's Law" which states the value of a network is proportional to the square of the number of nodes attached to it.

But arguably, it was not until the mid-1990s and the commer­cialisation of the Internet that the true value of internetworking became apparent. The growth of the internet and the world wide web in particular since then has been astonishing.

With the help of tools like web browsers, the internet was trans­formed in just four years from an arcane system linking mostly academic institutions into a global transport system with 50m users. Today, that figure has swollen to about 160m and esti­mates for the electronic com­merce that it enables are pushed up almost weekly.

According to the latest Gold-man Sachs internet report, the business-to-business e-commerce market alone will grow to £l,500bn in 2004, up from $114bn this year and virtually nothing two years ago.

Two inter-related technologies have been driving these changes:

semiconductors and network communications.

For more than 25 years, semi­conductor development has broadly followed the dictum of "Moore's Law" laid down by Gor­don Moore, co-founder of Intel.

This states that the capacity of semiconductor chips will double every 18 months, or expressed a different way, that the price of computing power will halve every 18 months.

Moore's Law is expected to hold true for at least another decade but around 20l2 scientists believe semiconductor designers will run into some physical (atomic) roadblocks as they continue to shrink the size of the components and lines etched onto of silicon chips.

At that stage, some computer scientists believe it will be necessary to look for alternatives to silicon-based computing. Research into new materials and computer architectures is mostly focusing on the potential of quantum computing.

Meanwhile, the deadline keeps being pushed back by improvements to existing processes. At the same time, there have been big leaps in communications technologies and, in particular, fibre optics and IP-based systems.

Today, one strand of Qwest's US network can carry all North America's telecoms traffic and in a few years, the same strand of glass fibre will be able to carry all the world's network traffic.

"We are going to have so much bandwidth, we are not going-to know what to do with it," says John Patrick vice president of internet technology at IBM.

"I am very optimistic about the future."

He believes this telecoms capacity will enable the creation of a wide range of internet-based new services including digital video and distributed storage and medical systems.

But he cautions: "The evolution of the internet is based upon technical things, but in the end it is not about technology itself, it is about what the technology can enable."