Right now, to be absolutely "fabless" is to be top of the pile in the semiconductor industry. Fabless companies, semiconductor design houses which have no manufacturing operations but either contract out their designs to silicon foundries or licence their intellectual property to other chipmakers, are outperforming their vertically integrated contemporaries. Some think these latter organisation are coming close to their sell-by date.

The US Fabless Semiconductor Association, the sector lobby, is already predicting that by 2010, half of all integrated circuit revenues will come from pure-play fabless companies and the fabless operations of vertically integrated manufacturers and systems houses.

The evidence supports its contention. Last year was without doubt the worst in the history of an industry which has seem some dramatic swings of fortune over time: about 80 per cent of chip companies failed to make their financial targets in the first nine months of the year, and only some frantic massaging of expectations reduced the carnage in the final quarter.

Fabless companies, however, fared better than their vertically integrated contemporaries on every measure. Revenues for quoted fabless companies fell 21 per cent to $12.7bn in 2001 compared with the year before while integrated manufacturers suffered a 27 per cent decline. Despite the 2001 downturn, quoted fabless companies have performed 60 per cent better than the entire semiconductor industry over the past five years.

And despite the venture capital drought which followed the collapse of the technology bubble last year, some 111 North American fabless companies were able to raise $2bn in private equity.

The emergence of the fabless company as a significant force in the industry is at the moment chiefly a US phenomenon. But by lowering the technical and financial barriers to industrial participation it offers the possibility that European and Asian groups will be able to build a significant presence in the market that has hitherto been denied them.

Indeed, Hans Rohrer, newly appointed chief executive of Acuid, a semiconductor group based in St Petersburg, Russia, and Edinburgh, Scotland, argues that the fabless industry in Europe could grow to rival its US competition within a decade.

Mr Rohrer's views carry the weight of somebody who has been intimately associated with the other aspect of the polarisation of the semiconductor industry, the emergence of silicon foundries concentrating wholly on the manufacture of chips.

Chipmakers such as Texas Instruments or National Semiconductors were largely vertically integrated, combining both design and manufacturing, until the beginning of the 1990s. Intel, the world's largest microprocessor manufacturer, remains an example of this model. These integrated groups were able to take advantage of economies of scale in both research and development and manufacture. They were also able to maintain data integrity across the whole design and manufacturing process: no mean achievement at a time when computer-aided design tools and manufacturing methods were both relatively unsophisticated and proprietary.

The rise of fabless companies is generally attributed to the growth of the market for semiconductor devices in the communications and multimedia industries. The large integrated manufacturers failed to recognise the potential for these devices, letting in a host of smaller, quicker competitors.

ARM, the UK group whose processor designs have been licenced by many leading mobile phone handset manufacturers, is typical of this group.

The integrated houses, of course, had been able to act as gatekeepers, allowing only their choice of customer to take advantage of their production capacity. But in the late 1980s, Morris Chang, a Taiwanese semiconductor expert, persuaded the Taiwanese government to fund the establishment of the first pure-play silicon foundry. The result, TSMC, is now the world's largest independent chip manufacturing group. Before his move to Acuid, Mr Rohrer was European head of TSMC.

The rise of the silicon foundry has put manufacturing technologies, once the sole property of the integrated companies, into the hands of start-ups outside the hothouse of North America. They already have the design expertise: Eastern Europe, in particular, has a wealth of skilled mathematicians and electronics engineers. If they fail to thrive, it will be a failure of imagination rather than of infrastructure. Intellectual property (IP): how chip companies exploit IP rather than physical assets.