News APP

NewsApp (Free)

Read news as it happens
Download NewsApp

Available on  gplay

Rediff.com  » Business » The rise of China chips

The rise of China chips

By Matei Mihalca
January 27, 2004 13:11 IST
Get Rediff News in your Inbox:

This year, Semiconductor Manufacturing International Corp of Shanghai plans an initial public offering in New York and Hong Kong.

Although an intellectual property lawsuit from rival Taiwan Semiconductor Manufacturing Corp may impact the plans, SMIC stands a good chance of raising an expected $1 billion.

Other China chip companies planning overseas public debuts this year or next include Advanced Semiconductor (ASMIC), Grace Semiconductor (GSMC), Central Semiconductor (CSMC), Huahong-NEC, and Shanghai Belling.

This flurry of activity makes it a good time to ask a few questions: Is China emerging as a new force in semiconductors, as in many other manufacturing industries?

Does China need semiconductor manufacturing? And what would such success mean for the country's larger emergence as a world economic power?

The first thing to note is that China's niche, like Taiwan's, is so far the contract manufacturing of chips, known as 'foundry'.

TSMC started the worldwide foundry industry in 1987 and it hasn't looked back since: TSMC now holds about half the market for contract chips globally and its stock market capitalisation is $40 billion. That makes it the third semiconductor company in the world by market value, after Intel and Texas Instruments.

The idea behind foundry is to serve two types of customers: semiconductor design houses, which don't have the money or interest to operate fabrication facilities ('fabs'), focusing instead on design and marketing, and integrated players, which are finding it difficult to invest in ever more expensive facilities.

As a result, they are outsourcing the manufacturing of some (and, in time, perhaps most or all) of their products to dedicated manufacturers.

TSMC's bet was on specialisation: by doing only one thing, it would do it better and cheaper. Aggregating demand from the marketplace, TSMC makes more efficient use of very expensive assets: a new fab today can cost billions of dollars. The Taiwan government lent a helping hand with favourable taxation and other incentives.

This is the recipe China's government is now replicating, and taking further: concessionary loans, low-rent land, other assistance. And labour cost, although only a small part of total cost in semiconductor manufacturing, is even lower in China.

The capital markets, whose support is necessary to fund fab after fab, are weak in China, but foreigners seem happy to lend a hand. SMIC, GSMC, and CSMC have all made ample use of venture capital funding.

In light of the above, it would seem the conditions for success are in place. But not so fast. China's fundamental strength is productive, low-cost labour used in volume manufacturing.

But new fabs use no people at all, and foundry isn't volume manufacturing: it's a service -- the provision of custom products to many customers. Chips are small and light, hence easy to transport. Customers routinely fly finished chips across the world for assembly and test, before delivery elsewhere.

But this is not quite a borderless world, and China's entry into semiconductor manufacturing has coincided with Beijing's levying a 17 per cent tax on imported chips. Take away the protectionist tax, as many in the US lobby, and the rationale for a Chinese domestic industry weakens.

What customers would this industry serve? China does consume a great many chips: $24 billion's worth, going into all kinds of hardware assembled there. But Chinese companies supply very little, around $1billion.

That's less than 1 per cent of the $166 billion worldwide semiconductor market. Can local Chinese designers and manufacturers supplant foreign suppliers? Not likely. Most of the chips used in personal computers or handsets still come from the US and Europe, and their manufacturing isn't about to migrate to China where intellectual property rights are weak.

The rise of Taiwan foundry relied heavily on a few advanced products that aren't likely to move to China: graphics chips, for example, a market dominated by foundry-friendly independent companies like NVIDIA.

Such customers are likely to stay with TSMC and United Microelectronics Corp (UMC) in Taiwan or go to China with them, when they put in place suitable facilities there.

Taiwan's own semiconductor design industry is a candidate to send orders across the strait, given linguistic and cultural ties, but it, too, has close historic links with local foundries, and its volumes are still relatively small.

If there isn't evidence of strong demand, why is China putting supply in place? Because it can. China's banking system is bursting with liquidity and the country aspires to international prestige.

This is a nation that recently put a man in space so that it can say it did so. But perhaps there is a case for regarding the semiconductor industry as strategic.

Semiconductor manufacturing is a little bit like the airline industry: equally capital intensive, highly cyclical, not particularly good to shareholders, but important because it makes other things work.

The financial profile of semiconductor manufacturing is such that some years foundries make lots of money, other years they don't.

Each year they need to invest billions of dollars into new equipment capable of making ever more advanced chips. Shareholders are regularly diluted as new capital is raised.

Return on equity is low, while the cost of capital is high, not least because of the earnings volatility derived from high operating leverage.

High financial leverage is always a danger, and China's new foundries need to resist the allure of cheap local credit.

But to focus on the ongoing financial dynamics of semiconductor manufacturing would be to miss the point -- the future.

One is in foundry because one dreams of a future of monopolistic dominance, when rivals are defeated, and when high profits are finally there for the taking. Sadly, this future is perpetually postponed by the entry of new industry players.

After TSMC's initial success, United Microelectronics Corp., earlier an integrated player in Taiwan, shifted to foundry. Then Singapore gave the world Chartered. Now China has a handful of foundries. And South-east Asia retains ambitions, as well.

The 'last man standing' scenario I sketched above is one reason behind China's semiconductor industry, but there is another.

Semiconductor manufacturing is strategic for China not only because it brings technology and provides employment, but because, alongside foundry, an indigenous semiconductor design industry may appear.

If foreigners fail to provide demand for China's new fabs, perhaps locals will. It's easy to say that China doesn't have the research and development capability to compete internationally.

But maybe it won't need to. Many Chinese semiconductor design houses, though still young and small, focus on products indigenous to China -- like chips for the Chinese wireless local loop standard known as xiaolingtong.

If China adopts its own 3G, DVD, or HDTV standards, this would help create a niche that's insulated from the reach of the dominant global players.

Underpinning the rise of Taiwan's world-class semiconductor industry was a functional, deep capital market. If China gets that right, much could follow as entrepreneurial energies are unleashed and hundreds of design houses mushroom across the land, possibly under the umbrella of protectionist policies.

To return to our questions: Does China need semiconductor manufacturing? Probably not. If China's advantage is, and likely to remain, in labour-intensive industries, then the combination of low-cost labour and heavy capital investment is a misallocation of resources.

Will China develop a semiconductor industry? Yes. What will this success mean for China's emergence as a world economic power? In time, it may mean that China moves from semiconductor manufacturing to design.

Since chips are the brains of appliances, the flow of semiconductor design and manufacturing to China would raise difficult questions for the global community.

Powered by

Get Rediff News in your Inbox:
Matei Mihalca
 

Moneywiz Live!