In 2014, more than half the world’s semiconductor sales were in China, but the country accounts for less than a third of production. That is set to change.
In the 15 years since China as a nation started to talk about becoming a leading producer of semiconductors, the cost of establishing a competitive manufacturing base in the sector has more than quadrupled.
From 2000, the government published a series of Five-Year Plans meant to encourage local companies to invest in chipmaking but competitors from the US, Taiwan and South Korea far outpaced their efforts. But in the past year government and state-backed companies have claimed they have billions of dollars available to buy a bigger share of production in an attempt to satisfy a higher proportion of the country’s enormous demand for chips by its local equipment makers.
Figures collated by PwC show that in 2014 China consumed almost 57 per cent of the world’s semiconductors by value. Local fabs, many of them foreign-owned, supply less than a third of that market.
A $50bn war chest
State-owned investment firm JAC Capital paid close to $2bn for NXP Semiconductor’s RF amplifier business in a deal that would help fund the Dutch company’s merger with Freescale. Tsinghua Unigroup’s chairman Zhao Weiguo told Reuters in November the company’s war chest of close to $50bn, to be spent over five years, would be enough to make the company third-placed in global chipmaking. But, even if the firm was already number three in chip production, it would easily spend half that amount simply keeping pace with the competition.
“With the technology moving to 10nm and beyond over the next five years, investment in fabs, which is a annual commitment not a one-time investment, will require at least $5bn per year,” says Bill McClean, president of US-based analyst firm IC Insights.
TSMC, which as a foundry in neighbouring Taiwan produces chips for fabless companies such as Qualcomm, and is effectively the world’s third largest chip producer, spends around $10bn a year on capital expenditure. That level of investment has allowed it to become four times as large as its nearest direct competitor, GlobalFoundries.
In the summer, Weiguo reportedly pledged half of the company’s acquisition budget on Micron Technology, the Idaho-based memory supplier. The deal looked realistic because Micron’s share price dropped sharply on the back of a collapse in DRAM prices in what remains a strongly cyclical market.
Amid concerns that the US might block the deal to avoid a Chinese company owning a supplier of devices to the military, a formal offer did not emerge. But the emergence of billion-dollar funds for purchases indicates a desire by China’s government to wrest greater control over what it now sees as an increasingly strategic industry.
“It’s moved from being just a plan to knocking on doors saying ‘we would like to buy you’. It’s a step forward and they now seem to have a chequebook,” says Malcolm Penn, chairman and CEO of semiconductor analyst Future Horizons. “But you do get a sense that there doesn’t seem to be a real strategic implementation plan behind it.”
Handel Jones, president of International Business Strategies, said at Semicon Europa 2015: “There is a lot of activity: some of it is well planned but a lot is not well focused.”
The question is whether China can realistically buy its way into production while the leading suppliers are themselves investing aggressively. It is currently several generations behind in process technology. Of the eight 300mm fabs it has either in production or under construction, just three can run processes more advanced than 32nm. Intel, TSMC and others are developing 10nm processes.
Two of the advanced fabs in China are owned by South Korean companies SK Hynix and Samsung. China-based SMIC, with annual revenues a tenth of TSMC’s, is developing a 14nm finFET technology but “by the time 14nm at SMIC is up, it will not be competitive,” Jones says.
The country has had greater success with chip design, turning to SMIC, TSMC and others for fabrication. The fabless model is one that US-based Qualcomm has used to become the world’s third largest supplier of semiconductors worldwide. According to PwC, Chinese companies accounted for close to one fifth of worldwide sales by fabless suppliers.
“I believe China would be much better off investing in design capability in which even one billion dollars would be a huge investment with a great deal of potential for significant return,” says McClean.
However, Penn argues the case for chipmaking. “The chip is a sizeable proportion of the product. And if you do control chip production it’s a tremendous benefit.”
The memory market is one where foundries have failed to make an impact – the margins are too low to support the business. Penn also points to smartphone processors as a sector where the fabless model may prove to be uncompetitive in the long term.
A flattening chip market in 2016 is likely to lead to further large-scale mergers and acquisitions following those of Freescale by NXP and Broadcom by Singapore-based Avago. The $100bn in government funding that is apparently now on the table would allow Chinese companies to buy a sizeable share, if not the number-three position that Tsinghua’s Weiguo covets. But the deals available would make it hard for any company to buy its way into the top ten with a coherent collection of businesses, warn McClean and others.
Jones argues it would be better for China to focus on joint ventures with foreign suppliers, continuing a trend that began in the early 2000s. There are strong reasons for foreign investment, he says: “It will become increasingly difficult to sell into China if you don’t have some form of manufacturing ecosystem in China.”
Whether it pursues wholesale acquisitions or pledges the cash for joint ventures, China now seems ready to finally deliver on the chipmaking ambitions it has harboured for years.