Has the center of gravity for system on chip (SoC) innovation shifted to China? If you're planning to start a fabless semiconductor company, should you pack your bags, leave Silicon Valley, and head for Shenzhen or Shanghai? Not so fast, according to panelists at the recent Design Automation Conference (DAC 2014).
The Chinese fabless semiconductor market is growing rapidly, there are over 600 fabless companies, and the Chinese government is (reportedly) gearing up for a major investment in semiconductors. However, many of those 600 companies are small and/or unprofitable, there's a cultural bias against mergers and acquisitions, and only a small portion of the ICs consumed in China are actually manufactured there.
The panel was organized by Thomas Wong, engineering director at Cadence. It was moderated by Junko Yoshida, editor in chief at EE Times. Panelists were as follows, shown from left to right in the photo below.
To begin the panel, Prof. Wei was asked to provide an overview of the Chinese semiconductor industry today.
Prof. Wei—Fast growth, but many small companies
As of last year, there were 632 fabless semiconductor companies in China, according to Prof. Wei. They generated around 84 billion RMB (Yuan renminbi, about $0.16 USD) revenues in 2013, which is around $14.2B USD. "If everything goes well, this year we will record around $20B US revenues for [Chinese] fabless companies," Prof. Wei said. "That is very surprising for me, because after 14 years of growth, we still have a growth rate around 39%."
The largest product category for Chinese fabless providers is communications. Spreadtrum and HiSilicon are large Chinese fabless companies that serve this area. The second largest area is computers and peripherals. The third largest is smart cards. Finally, power ICs claimed just 8.4B RMB last year, but saw 161% growth.
Last year China imported about $200B USD in semiconductors, while domestically made ICs represented only $40B USD. Moreover, Prof. Wei said, "we have too many small [fabless] companies." He noted that 124 fabless companies have revenues over 100M RMB, 134 are in the 50M-100M RMB range, and 196 have revenues below 10M RMB. 223 fabless companies lost money last year. Many of the fabless companies have fewer than 50 people.
Prof. Wei said he believes that most fabless companies are still under control of their founders, and they don't want to merge or sell. "There are some cultural impacts," he said. "If a company is acquired, that means there is some kind of failure. They don't like it." If you look at the size of fabless companies in China and abroad, Prof. Wei said, "mainland China still has a long ways to go."
Jin Zhang—Government takes the initiative
Yoshida asked Zhang to comment on rumors that the Chinese government plans to spend up to $10B USD per year on the domestic chip industry over the next five to ten years. "Various numbers are thrown about, but the policy has not been announced," Zhang said. "New policies will come out very soon, and there are big numbers involved. There is a big gap between consumption and production. Local domestic [semiconductor] production is going very slowly, and that's a big strategic issue for the Chinese government."
Sharing a historical perspective, Zhang noted that prior to 2000, all semiconductor companies were state owned. In 2000, a new policy—so-called Rule 18—was put in place to begin the privatization of the semiconductor industry. It encouraged direct investments and provided favorable tax treatment. The government was hoping that by 2010, domestic production would meet domestic needs, with some left over for export. While the Chinese semiconductor business grew dramatically, the hoped-for domestic production did not happen.
When Rule 18 expired on 2010, stronger policies were put in place. The new Rule 4 provides more incentives for private and government investment, and encourages mergers and acquisitions. "We'll see a lot of money come in," Zhang predicted. "It will be a great opportunity for a lot of people."
(Note: Zhang provides considerably more detail in a recent EE Times article that she authored.)
Limin He—Comparing the US to China
Limin He described differences in the investment environment between the US and China. In the US, he noted, a lot of venture capital money was going into semiconductors 10 years ago—but that has slowed greatly due to the high cost of chip design at advanced process nodes ($50M for a 20nm chip). Chinese companies, on the other hand, are usually working at much more mature nodes such as 180nm or 160nm. R&D costs are less, and it's easier to maintain a small team of 50 people.
He said that the center of gravity for chip design has not shifted to China. Despite a few well-known Chinese companies like HiSilicon and Spreadtrum, the top ten fabless companies are all in the US, Taiwan, or Japan. These companies are spending billions of dollars to invest in new development, He said.
The US has strong engineering talent, He said. However, China has 1.4B people and a huge market opportunity. China is the number one market for PCs, tablets, cell phones, and more. China also has a lot of talented young engineers. "If I was going to do a startup," He said, "I would do a combination of both—a small team in the US and a larger Chinese complement."
China isn't cheap
In the Q&A period, Prof. Wei noted that China is not a cheap place to start a business. "The average salary for a Chinese engineer is 13% higher than engineers from Taiwan, and efficiency is 20% lower," he said. A related problem is that not enough engineering students are going into semiconductor design. The Chinese semiconductor industry employs around 72,000 people, but only about 4,000 or 5,000 engineering graduates per year are going into microelectronics.
"We need people, that's true. We need a lot of people," he said. "We've heard the government will encourage university collaboration with industry, but when that will start, I don't know."
Collaboration between US and Chinese semiconductor companies is also badly needed, Prof. Wei said. "The most advanced technology is in the US, and the most experienced talent is in the US," he said. "But Chinese companies are closer to the end customers and they understand the domestic demands."
Related blog posts
- Mixed-Signal Gets Clear Message in China
- Virtuoso AMS Wins the China ACE Best EDA Product Award