[lbo-talk] The rise of China chips, Part 2: Analog

uvj at vsnl.com uvj at vsnl.com
Sun Apr 11 04:39:56 PDT 2004


Business Standard

Monday, April 05, 2004

MIHALCA ON CHINA

The rise of China chips, Part 2: Analog

Greater China is entering a new realm of semiconductors, says Matei Mihalca

Not long ago, I argued that the Asian semiconductor landscape is changing. But the change is not only China's emergence as a new force.

The product focus of the industry in Greater China is also beginning to shift, from a traditional concentration in logic and memory chips to a new area: analog. This is important because it shows Asia moving up the value chain.

Perhaps the easiest way to understand the difference between these types of chips is this: memory chips remember things, logic chips do things. These two products still represent the mainstay of the industry in Asia and indeed the world.

Analog semiconductors represent only about a sixth of the total market. As opposed to logic or memory, they capture and relay inputs from reality. Amplifiers are one example. While logic and memory are digital, working with 0's and 1's, analog chips reflect the uninterrupted continuums of the physical world.

Perhaps the most famous description of analog calls it 'the shell of the electronics world'. As the digital world expands, the shell may get thinner and thinner, but it will cover ever-greater territory.

Analog semiconductors have a higher design, rather than manufacturing, component. Perhaps for this reason, industry disintegration has not yet occurred in analog as it has in logic.

In logic, chip design and manufacturing have become increasingly distinct, with more and more companies doing either one or the other.

This trend has greatly benefited Asia: Taiwan is a global centre for both logic chip design and contract manufacturing. Of course, so-called integrated players still remain: companies such as Intel, Texas Instruments, and Motorola, but disintegration appears to be a powerful industry trend.

The analog semiconductor industry is either at an earlier stage of development than logic, or perhaps the two sectors are fundamentally different in nature and unlikely to develop in a similar direction.

This is a big question. There are some preliminary signs that analog is following the logic model: analog design specialists are emerging, especially in Taiwan, while analog manufacturing is beginning to migrate to China. But it may be too early to tell whether this is a fundamental shift.

As in logic, integrated companies, mostly in the West and Japan, still design, make, package, and test most analog chips - the value chain is still mostly in-house.

The reason why analog design and production may be coming to Greater China is because, as in the case of logic, many customers are here. In addition to computers, analog semiconductors are used in many consumer products (such as household appliances and automobiles) for which China is either already an important source of supply or a new source of demand, which in turn is driving the localisation of supply.

China buys far more chips than it makes, but in analog the balance is even more extreme: out of some $ 6 billion in analog chip demand, Chinese domestic producers account for less than $100 million.

These chips are manufactured in China at companies such as Advanced Semiconductor Manufacturing Corp. (ASMC) and SIM-BCD, both based in Shanghai.

These firms don't need multi-billion dollar facilities to fabricate advanced 8 or 12-inch silicon wafers; rather they employ much cheaper, older technology to produce four, five or six-inch wafers. China's position in global manufacturing shouldn't rest in capital-intensive industries. Analog better fits the country's industrial profile.

For this reason, ASMC and SIM-BCD may turn out to be more successful than Chinese contract manufacturers like Semiconductor Manufacturing International Corp. (SMIC), which focus on logic. A foretaste of this came last month when SMIC had a poorly-received IPO in New York and Hong Kong.

Supported by advantageous bank loans, tax benefits, and trade protectionism, SMIC may win some market share from rivals in Taiwan and Singapore, but its operations will continuously require high levels of investment. Financial returns at analog manufacturers are more favourable to investors.

Meanwhile, in Taiwan, a customer base is emerging for these Chinese analog semiconductor manufacturers. These customers are a new breed of analog design houses. Taiwan has long had a semiconductor design industry, made up of hundreds of fabless companies, so called because they do not own fabrication facilities, or fabs.

The largest of these, Mediatek, is the number five semiconductor design house in the world by revenues ($ 1.2 billion). What has changed in recent years is the emergence of analog design houses. Of these, Richtek, Anpec, and System General are the best-known names, all recently listed. They design mainly power management semiconductors.

This is a promising segment: anyone who uses a mobile phone understands the importance of power. This analog group is growing at much faster rates than logic design, albeit from a smaller base (Taiwan has only 1per cent market share in analog).

India can benefit from the rise in analog contract manufacturing in China, just as Taiwan is doing, through the emergence of analog design houses. It is always helpful to have semiconductor design, manufacturing, and downstream customers together in a 'cluster', and Taiwan's success as a technology centre owes much to such proximity.

India lacks semiconductor manufacturing, but analog (as opposed to logic) production is cheaper to put into place if demand is sufficient. Even in the absence of such capacity, the split between Taiwan-based design and China-based manufacturing shows that local manufacturing is not required.

An absence of customers would be more difficult to overcome: the creation of an Indian electronics industry on par with that in Taiwan or China would be a tall order, although it may happen in time as the Indian economy moves to a larger industrial sector. We can expect labour-intensive manufacturing to shift from China to India in the next few decades.

Analog chips are also used in consumer goods and automobiles where India has an increasingly important presence today - arguably, a preview of things to come. There is, therefore, some demand for analog semiconductors in India and this demand is likely to grow further.

But it is in terms of semiconductor design talent, not demand, that India stands out. It's estimated there are several thousands of semiconductor engineers in India who have very large-scale integration (VLSI) skills - that is, chip designers.

This is in the context of a worldwide design industry that employs around 200,000 people. These numbers suggest India has 1-2 per cent of the global chip design talent pool. So far, the efforts of these engineers have focused on logic-related areas: semiconductor intellectual property (SIP), design services, and software (protocols, real-time operating systems, etc). Results have been mixed, as both standards-based SIP and design services are competitive and fundamentally not very attractive businesses.

But India's analog capability is strong, certainly stronger than Taiwan's, although in general design - logic - these strengths are reversed. (Taiwan has three to four times more chip designers than India.) The strength of Indian companies, and of Indian-backed companies in the United States, in networking is an example of this capability: transporting a signal is an analog endeavour.

The Indian Institutes of Technology have traditionally emphasised analog design. IIT Madras, for example, is an outstanding centre for analog. It was India's analog experience that reportedly first drew Texas Instruments to begin outsourcing to India in the 1980s.

Fundamentally, it makes sense for Indian companies to become involved in analog semiconductors because more of the value-added resides in design, rather than in manufacturing as in the case of logic. Indeed, many specialist analog chip companies (Maxim, Linear) are similar to software companies in terms of profitability.

The lack of local manufacturing capacity and demand may have prevented the development of analog chip design in India, just as in Greater China. This is now changing. We may be at the beginning of a new era of analog semiconductors in Asia.

matei_mihalca at hotmail.com

Business Standard Ltd.



More information about the lbo-talk mailing list