Makers of chips seek a wider audience

The financial downturn has hit the semiconductor industry hard.

Orders for microchips came to an almost complete standstill at the end of 2008 as computers and mobile phones languished unsold on shop shelves.

Global semiconductor sales fell 2.8 per cent last year and are expected to fall a further 21.5 per cent to $195.6bn in 2009, according to the Semiconductor Industry Association.

Intel, the world’s biggest chipmaker, last week reported a sales revival driven by consumers buying cheaper, smaller netbooks but revenues fell in Europe and there were weaker sales recorded with business customers.

The semiconductor industry is used to severe cyclical swings. It has seen six significant cycles since the 1970s, caused by global recessions and oversupply.

The last one – in 2001 when the dotcom bubble burst – was worse than the current downturn with sales falling 45 per cent.

But this time around, strategic issues – including soaring development costs and the intensifying need to innovate – have combined with the effects of the downturn to unleash forces that are threatening to reshape the industry completely.

Crucially, the computer industry – one of the driving forces of the semiconductor sector – has been slowing, forcing companies to look elsewhere for growth.

PC shipments are expected to decline 6 per cent in 2009, according to analysts at Gartner, the research company.

But even without the recession, markets were growing saturated. In developed countries, 80 per cent of computers sold are replacements.

In the past, Moore’s Law (see Q&A box) – the idea that the chips double their performance power roughly every two years while cutting costs in half – has driven the semiconductor industry forward.

Since the invention of the silicon chip in the late-1950s, faster, cheaper chips have spurred the creation of more sophisticated computers. But as consumers become more reluctant to upgrade, Moore’s Law has become a problem.

“If you can get twice as many chips from the same amount of silicon every two years, you need to double the number of computers sold every two years, or find new markets,” said Richard Doherty, research director at Envisioneering, the technology assessment and research consultancy.

Intel, which has dominated the market for computer chips, has been eyeing the mobile and netbook markets as potential areas for growth.

It is easy to see the temptation. Netbooks were the fastest growing segment of the computer market last year and about 1bn mobile phones are sold each year, dwarfing the 250m computers sold in the same period.

Chipmakers are also looking at embedding chips into a range of devices from fridges to blood pressure monitors and television sets that could benefit from being connected to the internet.

However, a switch will not be easy as the market for mobile phone chips is highly competitive and requires far smaller, more power-efficient chips than those used in the computer industry.

Development costs are also becoming a burden as chips become smaller and smaller.

The tiny transistors that make up chips have shrunk from 130 nanometers in 2000 to 45nm, and even 32nm, today.

At 45nm, 2,000 transistors can be packed into the width of a single human hair and the cost of developing a chip using these is $20m to $50m. With 32nm technology, the costs can leap to $75m.

Three European chip companies recently merged their mobile phone operations into a single group – ST Ericsson – saying they needed scale to keep up with huge research and development costs.

Shane Rau, analyst at IDC said: “The manufacture of chips has become very capital-intensive .

“Only companies like Intel and Samsung are able to maintain their own fabrication facilities because they have scale.”

Many chip companies have closed their manufacturing units and have their chips made by specialist foundries in Asia, such as TSMC, UMC and Chartered Semiconductor.

Texas Instruments is gradually leaving manufacturing, especially at the high end, and AMD, Intel’s rival on computer chips, spun out its factories into a separate company last year.

Many companies may also be forced to look at a new technology roadmap beyond Moore’s Law.

Some may focus on different types of chips – such as radio receivers, power control or sensors – where shrinking sizes are not the key driver.

ST Microelectronics, for example, has a small but thriving part of its business making accelerometers.

Or they may need to experiment with new materials and work at nano-scale – manipulating chips at the atomic level – in order to continue. Only the very biggest companies will be able to afford the investment this requires.

George Scalise, president of the Semiconductor Industry Association, said: “We are now facing the research dilemma. Will the semiconductor industry lead the transition to the era of nanotechnology – or will we go the way of the vacuum tube manufacturers?”

None of the companies that made vacuum tubes for the first computers in the 1940s and 1950s survived the switch to silicon. It is a sobering thought for the sector.

This is the first in a FT series on the challenges facing the semiconductor industry