With about $120 billion in funding at its disposal, China is looking to buy key technologies it doesn’t already have.
Over the years, China has unveiled several initiatives to advance its domestic semiconductor industry. China has made some progress at each turn, although every plan has fallen short of expectations. But now, the nation is embarking on several new and bold initiatives that could alter the IC landscape.
China’s new initiatives address at least three key challenges for its IC industry:
1. China continues to lag behind its foreign rivals in IC technology and the nation’s chipmakers hope to close the gap.
2. China consumes nearly 60% of the world’s chips today, but its chipmakers produce less than 10% of the world’s ICs, according to analysts.
3. China must import about 90% of its chips from foreign suppliers, which in turn, has created a staggering $150 billion trade deficit in ICs alone, according to Gartner.
Hoping to solve these problems, China recently launched two major initiatives. Unveiled in June of 2014, the first program is called the “National Guideline for Development of the IC Industry.” In simple terms, the plan is designed to accelerate China’s efforts in several areas, such as 14nm finFETs, advanced packaging, MEMS and memory.
As part of the plan, China created a $19.3 billion fund, which will be used to invest in its domestic IC firms. And over the next decade, local municipalities and private equity firms could spend a whopping $100 billion across China’s IC sector.
Then, in May of 2015, China launched another initiative, dubbed “Made in China 2025.” The goal is to upgrade and increase the domestic content of components in 10 key areas—information technology; robotics; aerospace; shipping; railways; energy systems and vehicles; power equipment; materials; medicine; and agricultural machinery.
In both initiatives, the country hopes to achieve a simple goal. “China wants to be self-sustaining,” said Samuel Wang, an analyst with Gartner. “They want to make everything, including all machinery, instruments and components.”
But if China can’t develop a given technology in a realistic time frame, the government has another plan in place. It will simply acquire a company to obtain the technology.
“China wants to support homegrown semiconductor manufacturers, including the foundries,” Wang said. “Mergers and acquisitions are also part of their strategy. They are looking for missing pieces of the puzzle. What they don’t have they want to acquire. And whatever they already have, they consider the foreign company as being a competitor.”
Indeed, China has embarked on an aggressive acquisition strategy to accelerate its efforts in select markets. In just one of many examples, a group from China recently acquired Singapore’s STATS ChipPAC, a move that put China in the upper echelon in the outsourced assembly and test (Outsourced Semiconductor Assembly and Test) market.
Not all of China’s acquisition plans have panned out, however. For example, China wants a domestic memory maker. So in July, Tsinghua Unigroup, China’s largest chip design company, launched an unsolicited bid to buy Micron Technology for $23 billion. That deal failed to transpire amid national security concerns.
In any case, the questions are clear. Have China and its chipmakers made any progress since launching its various initiatives over 18 months ago? Will China accomplish its goals and what are the challenges?
It’s too complex to look at every industry in China. But to help answer some of these questions, Semiconductor Engineering has taken a look at China’s ongoing efforts to develop three key technologies: 14nm finFETs; memory; and advanced packaging.
Chasing after finFETs
As part of its ambitious plans, China hopes to close the gap in process technology. Semiconductor Manufacturing International Corp. (SMIC), China’s largest foundry vendor, has been tapped by the government to lead the charge.
Today, SMIC is ramping up its 28nm planar technology, but the ultimate goal is to develop 14nm finFET technology by 2020 or sooner.
SMIC faces some challenges. FinFETs are difficult to develop. Moreover, it takes technology, know-how, and, of course, money.
So will SMIC succeed? “SMIC should be able to develop 14nm finFETs before the 2020 deadline,” Gartner’s Wang said. “We are still looking at another five years away. By that time, the industry knowledge will be available to develop the technology.”
SMIC will not only get help from Imec, but also the IC equipment industry. “When the equipment guys sell equipment, they more or less provide the recipe to the fab guys to guide them. Today, this is considered confidential information. But after three years they may be able to release such information,” he said.
Still, SMIC might not be that far behind. Today, the 16nm/14nm finFET rollout is taking longer than expected amid a slew of challenges. Plus, 16nm/14nm finFETs will be a long-running node.
In addition, the 10nm finFET market is also in flux. “If the industry pushes out 10nm, 14nm can be extended,” Wang said. “So, maybe the 14nm market will continue to be there by 2020.”
China has other options. To speed up its efforts in advanced logic, China could make an acquisition. Multiple reports have surfaced that China has talked to GlobalFoundries about an acquisition. A spokesman for GlobalFoundries declined to comment.
“I’m not sure who is making the first move, but now that GlobalFoundries owns the IBM Microelectronics assets, the sale of GlobalFoundries to China would run into major U.S. opposition,” said Joanne Itow, an analyst with Semico Research.
Besides finFETs, China could go down another path. GlobalFoundries, for one, recently held a forum in China to push its 22nm fully depleted silicon-on-insulator (FD-SOI) technology. FD-SOI has some advantages for China’s fabless chipmakers. “It’s easier to design with it,” said Gary Patton, chief technology officer at GlobalFoundries.
China is still evaluating FD-SOI, according to Gartner’s Wang, who added that China is also looking at developing other chip technologies, such as FPGAs, power semiconductors and SSD controllers. “Software programmability is big for FPGAs. So I don’t see that China can easily catch up in FPGAs,” Wang said. “In power management ICs or power MOSFETs, China can catch up.”
China imports the vast majority of its memory chips from foreign suppliers, but there is some memory production in that nation. For years, SK Hynix has been producing DRAMs in China.
In 2014, Samsung began ramping up its new 300mm fab in Xi’an, China, which is expected to produce 3D NAND. “(The) Xi’an fab, as we have mentioned, is mainly dedicated to V-NAND,” said Ji Ho Baek, vice president of memory marketing at Samsung, in a recent conference call. “The current ramp-up is being conducted gradually and according to plan. So, in addition to developing new products and enhancing our process capability, we also use the Xi’an fab to respond to the demand for enterprise and high-end datacenter SSDs.”
Then, in October of 2015, Intel announced that its China fab will be converted from the production of 65nm chipsets to 3D NAND and 3D XPoint over the next few years. It expects to begin selling 3D NAND products from this fab in the second half of 2016. Intel plans to invest $5.5 billion in the Dalian-based fab.
In addition, China also boasts one memory foundry vendor. That company, XMC, produces NOR on a foundry basis for Spansion, now part of Cypress. Spansion and XMC are also working on 3D NAND.
It’s also no secret that the Chinese government wants a China-based memory supplier, especially DRAM. Given that it’s too late to start a new memory company, China must acquire the technology.
For example, a Chinese consortium recently entered into a definitive agreement to acquire ISSI, a U.S.-based, niche-oriented memory supplier.
Meanwhile, Tsinghua hasn’t thrown in the towel after making an unsuccessful bid to buy Micron. In fact, Tsinghua and Micron are still in talks about forming a business deal, including a joint fab venture in China, according to multiple sources.
To help with the talks, Tsinghua recently hired Taiwan chip veteran Charles Kau, sources said. Kau recently resigned as chief executive of Nanya, a Taiwan DRAM maker. However, Kau is still chairman of Inotera, a joint DRAM venture in Taiwan between Nanya and Micron.
It won’t be easy to lure Micron to China, though. “Memory manufacturers will look to expand their footprint in China, if they’re able to get preferential incentives and/or funding for fab investments and operations. But this is not at the expense of transferring core technologies or R&D,” said Greg Wong, an analyst with Forward Insights.
Others agreed. “China is trying to get hold of DRAM production for its internal use. But with the preposterous bid for Micron this July and the possible delay in acquiring ISSI, China may have to position itself a bit more friendly to attract more memory makers,” said Alan Niebel, president of Web-Feet Research.
Advances in packaging
For years, China has been a major hub for IC packaging, where OSATs offer competitive pricing. Several OSATs have manufacturing sites in China. In addition, Intel, TI and other multinationals also have IC-packaging manufacturing plants in China.
Meanwhile, China’s domestic OSATs have been smaller players that focused on the Chinese market. Then, in 2014, Jiangsu Changjiang Electronics Technology (JCET), a Chinese OSAT, shook up the landscape by announcing a deal to acquire STATS ChipPAC for $780 million. The deal was completed in August of 2015.
With the acquisition of STATS ChipPAC, JCET jumped from sixth to fourth place in the worldwide OSAT rankings with combined sales of $2.6 billion in 2014, according to Gartner. JCET still trails Taiwan’s ASE ($5.2 billion), U.S.-based Amkor ($4 billion) and Taiwan’s SPIL ($2.7 billion), according to Gartner.
The deal also propels China into the advanced packaging market. Generally, JCET provides low-to-middle pin-count packages. STATS ChipPAC focuses on advanced packaging, such as 2.5D/3D, flip-chip and wafer-level packaging.
JCET is strong in the China market, while STATS’ customers are mainly outside of China. “From a portfolio point of view, there is a good marriage,” said Scott Sikorski, vice president of product technology marketing at STATS ChipPAC, now a subsidiary of JCET. “This also positions us much more strongly to compete with the other tier-one OSATs.”
Analysts agree. “Over the long run, this will probably benefit both STATS and JCET,” said Jan Vardaman, president of TechSearch International, a market research firm.
Still, there are challenges for OSATs in China. “Competing on price alone may not be sufficient (in the China market),” Vardaman said. “In addition, a complex process typically requires trained operators and engineers. A lower operator turnover rate is required. But the turnover rate is very high in China. Retraining an operator every three to four months is detrimental. Yield improvement and more efficient operations are required.”
Meanwhile, the JCET-STATS deal makes sense for other reasons. “The merger will also give us a much bigger access to the China market,” STATS ChipPAC’s Sikorski said.
In fact, the packaging business is heating up in China. For one thing, China’s fabless chipmakers are rapidly migrating towards more advanced designs, which, in turn, require new packaging types.
“As these Chinese brands try to compete on the international stage, they can’t do it with low pin count packaging only,” Sikorski said. “The Chinese customer base requires increasingly more advanced packaging.”