Research now under way will produce supercooled computers that could far outstrip the fastest machines now available, while also being physically smaller and using less power. They would be built using transistors so tiny that they would be turned on and off by the passage of one electron - the smallest unit of electrical charge.
Present computers use transistors, which are operated by the passage of thousands of electrons. But this generates heat, and also imposes limits on miniaturisation of computer chips.
Modern-day chips are already reaching a lower size limit, because as they shrink they must cope with "quantum effects" in which electrons can effectively pass through solid material. This phenomenon, known as "quantum tunnelling", makes it impossible for a transistor to be below a certain size and still operate at room temperature.
But the new work, being led by Professor Haroon Ahmed, at Cambridge University's microelectronics centre, together with the Hitachi Cambridge Laboratory, aims to use quantum effects to their advantage.
The new computers, cooled by liquid helium, will operate at -269 Centigrade, just four degrees above absolute zero. At that temperature, a single electron can be "trapped" on an island of atoms - effectively storing a single "bit" of information. Once that is achieved, the researchers will aim to build "warmer" versions that operate at -196 Centigrade. This is a crucial temperature as such a system could be cooled by liquid nitrogen, which costs about the same by volume as milk.
A system using that technology would be highly attractive to engineering, scientific and financial users, said Dr Yutaka Kuwahara, Hitachi's general manager of research and development.
"We are moving towards the era of the `terabit' chip - able to store 1,000 billion bits on one chip," he said. Present desktop PCs have chips that store about 8 million bits at a time.
"After about 2010, conventional memory structures will have too much difficulty keeping up with the demands of technology," said Dr Kuwahara.
Among the possible benefits would be more accurate weather forecasts. Forecasting systems treat large areas as collections of interacting squares, in which the weather changes in each square affect the others. The finer this "mesh" of squares is, the better the forecast.
Financial analysts would be able to sift through huge amounts of data arriving simultaneously and pick out the vital clues for market success.
"There's no way that computers can continue to progress as they are doing. The physics breaks down," said Dr Kuwahara. But the single-electron devices will overcome that. "We are moving towards precise control of a small number of electrons."
Laboratories all over the world have been working on single electronics and trying to achieve such a breakthrough. Hitachi's effort was made not only with Cambridge University but in conjunction with laboratories from the UK, France, Germany and Greece. The project is funded by ESPRIT, the European Programme for Collaborative Research, and a working version should be available by 2001.
The cost of building such single-electron transistors could be enormous. But Hitachi expects that chip manufacturers will club together in order to realise the benefits of the radical approach.
"We haven't estimated the cost," said Dr Kuwahara. "It will probably be an important factor. But manufacturers will share facilities. It will be too expensive to not cooperate."