But what will the semiconductor industry look like when the upturn comes, and how will its the downturn have affected the wider electronics industry?

The semiconductor industry is well known for its cycles of boom and bust. The compensation for the chaos these cycles cause has been an average compound annual growth rate of 17%.

But that 17% looks highly optimistic for the near future. The chip business is still set to turn over slightly more than $143bn this year. That's a long way down from the $204bn the industry booked in 2000, at the height of the dot.com bubble. So what happened?

The industry is well used to profitability cycles as supply and demand go in and out of step and pricing rises or falls to match. But, usually, unit demand has continued to grow even when excess capacity has driven prices down. In this latest cycle, real demand fell at the same time that overcapacity drove down average selling prices as inventory piled up.

Why did real demand fall? Last year, global demand for PCs fell for the first time in the 20-year history of the product as upgrade cycles lengthened. Cellphone sales flattened as regional markets reached saturation. And the combination of spending on Year 2000 compliance and over-investment in infrastructure for a dot.com future that is yet to arrive damped further IT and communications spending.

Malcolm Penn, founder of semiconductor industry analyst Future Horizons, said: "This downturn was worse because the economy, demand and fab capacity all went in the wrong direction at once."

Jim Tully, a research director at semiconductor industry analyst Gartner Dataquest, says this downturn has been characterised by declining demand going hand-in-hand with the crash of equities in the technology business and with the wider problems of big business.

"A normal downturn is pricing- based but with unit demand just going up and up," he said.

This time demand has fallen because consumers and businesses feel poorer due to the crash in equities. So the point when an upswing could happen is more difficult to predict because it is based on macro-economic factors such as public confidence rather than simple supply and demand economics. But will this downturn result in a restructuring of the industry?

"Right now, I don't think the downturn will result in any fundamental change in the industry," said Tully. "But we won't know until the downturn has passed."

He points out that some players are knocked out of the market at every downturn, others are weakened and the competitive landscape changes. The downturn does seem to have accelerated some trends, such as the sharing of R&D costs for new processes and the shift to outsourced manufacturing.

Tully said: "The whole industry is moving to this method of working, where advanced processes have got to be shared, because the work on 300mm wafer processing is too expensive for individual vendors.

"It's more painful than usual, but I don't know that the downturn has driven much of this thinking because these issues of collaboration would have had to happen even without a downturn."

Similarly, the shift to outsourced manufacturing has been a continuing one as individual semiconductor companies have recognised that very few of them can sustain the product volumes necessary to make the massive investments in manufacturing pay back.

Penn views the situation differently. He says the current downturn will lead to structural changes in the industry. He gives the example of Agere Systems, a semiconductor company whose technology stems from what was AT&T's Bell Labs, where the transistor was invented.

This year, Agere is selling most of its fabs and moving to outsourced manufacturing, "having to get out of what was the company's long-term family jewels," as Penn puts it.

"Companies everywhere are in survival mode, even Intel," he added. "This is a recession like previous recessions; and when recessions happen, people don't make money. In a deep recession, they don't make money in spades.

"This time the severity has been such that, this year, everyone is losing money."

Penn believes the depth of this recession will force the issue of process-sharing and a shift to outsourced manufacturing for all but the biggest players.

"The foundations for structural change are now in place, but these things have a pretty long hysteresis," he said. "Some people may just drop out of the business. It is going to show when the market takes off: the walking wounded will be left in the gutter."

He adds that companies that cut back on R&D now are storing up problems for the future, since future profitability springs from current R&D. The ones which are weak on R&D are the ones which will be vulnerable because that's tomorrow's profit."

Joe D'Elia, of market analyst iSuppli, agrees that structural change is coming to the semiconductor industry as this downturn forces people to choose their core competencies.

With fabs costing $2bn to $3bn for 300mm wafer processing at 0.13µm process rules, the foundries are expected to gain more customers as chip companies of turnovers of up to $5bn realise they cannot afford fabs.

Because the foundries must run their fabs all day, every day, to be economic, they will standardise their processes. So chip companies will have to differentiate their products on design and systems expertise.

Perhaps the biggest change D'Elia sees is happening in Japan: "The most profound thing we've seen is coming out of Japan, where those companies which believed in true vertical integration are now outsourcing, reducing costs and collaborating."

But he believes the Japanese industry is in such turmoil that "even some of these deals are beginning to fall apart". Despite this, the Semiconductor Industry Association is now forecasting that chip sales this year will grow 3.1% over 2001, to $143bn. It also forecasts that 2003's sales will be 23.2% up on 2002, and 2004's 20.9% up on 2003's, to $215bn.

But will it do the industry any good? For the moment, Penn says that although unit growth demand is rising sharply, average selling prices are declining equally sharply, driven by overcapacity: "What we have now is profitless prosperity, where we're working twice as hard and getting half as much money."

A profitable recovery will need real end user demand to grow enough to absorb the excess capacity and drive up average selling prices: "It wouldn't take a lot of change in confidence to give both of those factors a real kick up the backside."