China, Semiconductors and some implications
China’s growing presence in technology is hard to ignore. But to date the direct impact on semiconductors has been minor. That is going to change in a big way.
China’s impact on technology is impossible to ignore. Lenovo buying IBM’s PC division in 2005 was perhaps the first symbol of China’s technology leadership: the exit of the PC’s inventor was symbolic and now Lenovo leads that space. That was followed last year by them buying the server business too. In communications infrastructure the success of Huawei and ZTE is hard to ignore; some people wrongly think they are “cheap followers” anyone knowledgeable knows that is far from the case. In mobile phones brand and design are as important as technical prowess and the emergence of Xiaomi, Huawei and OnePlus is impressive: smartphones designed in China now have over 50% of the global market, compared with negligible presence five years ago. Last year Lenovo made another iconic acquisition, buying Motorola from Google for almost $3 billion.
But, while there are impressive Chinese chip companies (Spreadtrum, HiSilicon, Loongson, Phytium etc) and foundries (SMIC, XMC) they are not yet making a major global impact.
China currently consumes about half the world’s US $350 billion in chips, making semiconductors China’s biggest import — surpassing even oil. Yet, domestic fabs account for just 2.5% of worldwide IC revenue, while Chinese fabless firms account for just 8% ($80.5 billion) of the global market.
But this is changing fast and will accelerate: the Chinese government is determined to see success in this space and is pushing hard, just as it pushed towards Chinese success in space technology and in telecoms.
That 8% of fabless market is already significantly more than Japan (only 1%). Chinese companies now account for 9 of the top 50 fabless companies, up from zero a decade ago – and will rise much more soon.
Similarly, on the foundry side Chinese companies are developing. For reasons including costs and scale—and, in some cases, export controls—these have lagged three or four years behind the industry leaders. But the performance gap is shrinking. As global players such as Samsung, TSMC and TI set up shop in China, leading local foundries such as Shanghai Huali Microelectronics Corporation, SMIC, and XMC are poised to benefit from the development of a true technology cluster. Indeed, SMIC is perhaps the only global foundry reporting sustained growth at present, and domestic customers make up a large portion of that.
Meanwhile the slowing of economic growth and market correction in summer 2015 has emphasized the urgency of this: the government wants high-growth industries which can contribute to GDP growth.
Figure 1 China’s integrated circuit consumption and production comparison. Source: PwC.
There have been many similar government initiatives in EU and US over the years. But the new Chinese plan is different in two important respects. First, foreigners often underestimate the significance of the Five Year plans: these are not symbolic or notional, but are serious projects through which the shape of the whole economy is directly shaped.
Second is the sheer scale of ambition. The “National Guideline for the Development and Promotion of the IC Industry” says that the China IC industry revenue should reach RMB350 billion in 2015, and maintain a staggering CAGR of 20% through 2020. In other words, 2020 revenues will be US$143 billion, 3.5 times that of 2013’s US$40.5 billion.
To enable that total Chinese government (central plus local) funding will reach US $100 billion over the next five years (according to SEMI), while McKinsey estimates US $170 billion over the same time period.
Figure 2: China’s Semiconductor Industry Investment Fund: Source: E E Times
The Chinese have had previous schemes. But this is far bigger and in a significant change, this time investment decisions will be made by professional VCs (albeit following government strategy) and not by bureaucrats or local politicians.
This approach reflects a wider strategy: under President Xi, a clear emphasis is being placed on creating private/public partnerships. The aim is to stop corruption, make large-scale investments smarter, and less subject to local government decisions. That was a problem with previous schemes that were often sub-scale or led to local favourites.
Those investors will be spending the government’s money – and they will have a lot of it: the Chinese government, through its control of the banking industry, is able to commit huge amounts of funding to the industry. No other government can do anything similar at this scale.
To avoid the fragmentation and sub-scale activities of the past, the focus is explicitly on creating national champions: a small set of leaders in each critical segment of the semiconductor market (design, manufacturing, tools, and assembly and test) and a few provinces in which there is the potential to develop industry clusters at scale.
One example is the creation of Tsinghua Unigroup which bought Spreadtrum and RDA. (incidentally, that also illustrates how China is also leveraging money from abroad: for example, Intel’s $1.5 billion investment).
Another example of investment with scale and emphasizing that semiconductor eco-system is more than just chips, see JCET acquiring STATS ChipPack, creating the global #3 in device packaging. Indeed, 30% of the fund is allocated to “package & test”.
So, what will happen?
I can see four implications:
China’s semiconductor foundry capacity will expand massively, impacting costs globally. In an industry dominated by fixed costs this increase in supply is going to have a huge impact on global prices and on fabs around the world. That is even more true given the low cost of capital and power in China (the cost of labor is lower too but that is less relevant). It is hard enough to be a fab anyway, and this increase in capacity/reduction in cost is going to make those difficult economics even more challenging.
This coincides with an industry shift: fewer chip designs are moving to the latest process nodes. For many applications 28nm is an optimum, and there are very few that justify the expensive nodes of 20nm and below. In other words, leading edge in process technology is less relevant commercially.
As a result, the Chinese companies are addressing the biggest and most price sensitive parts of the market: as well as increasing volume there is a convergence between market sweet spot and low-cost supply.
Leading edge (Finfet and sub 20nm) may be less impacted, but the price pressure in the largest and most cost-sensitive segments of the market will have repercussions on utilization and revenue for all suppliers.
Second, it will be much tougher for chip firms too. That increase in supply and lower costs, is good for global economy and good for some firms – but maybe less good for those suddenly facing aggressive new competition.
Already, more than 50% of PCs, and 40% of embedded systems are designed in China. China could soon control 50%+ of hardware designs globally, and domestic suppliers have obvious advantages.
So as costs fall, competition intensifies: with credible local suppliers it will be tougher to sell into China, and global companies will face increased competition around the world.
That said, the third trend: there will be major strategic opportunities for Western firms – especially smaller players.
The Chinese government’s previous policies have not offered many opportunities for foreign companies. However, leaders in China’s semiconductor sector realize that the country needs to partner to improve the local talent base and supply chain.
This is a global industry and semi-conductor players understand they need to tap into global knowledge – particularly the innovation and invention that arises from clusters of excellence around the world, driven often by SMEs and Universities.
Chinese firms are no exception.
Smaller players may fare better in this than global giants: they have less to lose—and everything to gain.
While global design centres and collaborations are essential, they will not be enough. As such, it will be difficult for Chinese players to build a complete and competitive semiconductor value chain without capitalizing on foreign assets, and acquisitions will become more common (Huawei buying Neul is an example).
America has often regarded strategic industries as off limits for acquisitions, so this may well mean opportunities for European firms.
(The situation for Taiwanese companies is, of course, more complicated).
Finally, this will impact all sectors of the industry. Some argue China will only impact low-complexity systems: that is naïve.
Processors are globalized (thanks to ARM, Imagination or the MIPS V open core architecture) and companies like Spreadtrum, Phytium and HiSilicon are already developing very high-end products. Chinese industrial dominance make IoT an obviously crucial market, with a lot of innovation, while Xiaomi, Huawei and the like will drive progress in telecoms and cellphone chips. DRAM might be the only exception – but plans to buy Micron (if allowed by US) would remedy that gap.
To summarize: these big picture trends (increasing supply, lower costs, increasing competition, need for collaboration across all segments) combine with more familiar, more parochial ones. For example, the growing cost and complexity of designing complex SoCs. Rising mask costs, rising software costs, integration becoming more difficult: it is becoming harder and harder to understand what is going on inside complex SoCs. This leading to a crisis in development and debugging for companies, leading to delays and problems in the field, with growing recall and warranty costs. China will be no exception in having to face this.
Another industry wide issue is security. This is especially relevant for China: rightly or wrongly, there is a strong concern about security over Chinese product. Dealing with these concerns pro-actively is essential. Huawei in the UK has an interesting precedent: the Cyber Security Evaluation Centre in Banbury is funded by Huawei as an partnership with the UK Government to mitigate any risks to UK national security from supplying critical national infrastructure. Other companies could do likewise.
But when those products are buried in silicon, with limited visibility inside that becomes a bigger concern. How will things be audited or trusted? Industry wide initiatives on security are important.
Western companies need to recognize that China is not only the largest semiconductor market, not just a manufacturing base with a cost advantage but is becoming a true industry leader.
As such, even small electronic companies need to establish relations with the many stakeholders in China – government, customers, suppliers, and even competitors – and to seek opportunities through mutual understanding and engagement.
Conversely, Chinese firms need to recognize that money alone is not enough: technology relies on deep expertise, innovation, co-operation, eco-system, which may require cooperating with SMEs, universities.
To make these connections firms in both directions need to be looking to world class clusters for learning, partnership and acquisition.
Organisations, such as CW (Cambridge Wireless) in the UK to which I belong, along with global industry bodies like the GSA, have an essential match-making role between the emerging Chinese companies and Western innovators.
China’s investment in chips will massively impact the global semiconductor industry.
Many of those changes (increased supply, cost-pressure, innovation) will accelerate industry-wide trends, both supportive and challenging. Trade is the ultimate “win:win” and the success of Chinese companies will only benefit the global economy. However, both Chinese and Western firms must plan on how to cooperate and engage to best deliver the benefits: to themselves, their investors, their countries and to the wider society around the world.
Figure 3: Location of Semiconductor Operations in China
Acknowledgement: grateful thanks to Bob Driver of Cambridge Wireless and Paul Denlinger in China
Rupert is a CEO of UltraSoC, developing IP to help debug, optimize and secure complex SoCs. He is a veteran of the global semiconductor industry, and has held senior roles in both start-ups and prominent trans-national companies. Before joining UltraSoC he was VP of Strategic Marketing at Mindspeed following that company’s acquisition of Picochip (now part of Intel), where he was VP of Marketing and helped pioneer the concept of small cell basestations. His CV also includes spells at first:telecom, Arthur D Little and Analog Devices, where he played a key role in the development and mass-market adoption of digital subscriber line (DSL). Rupert is a Fellow of the IET.
 For example, in the UK the NMI’s IoT Security Forum