Forging ahead

In Only the Paranoid Survive, Intel founder Andy Grove argues that the key to success in the commercial world is being able to spot and respond to inflection points which signify major structural shifts.

The title of his book is a rhetorical allusion to the competitive landscape of the semiconductor industry, where Intel (computers) and more recently Qualcomm (mobile phones) have dominated the world of application processor design – and shunned complacency so as to stay ahead (in Intel’s case since the early 1970s, with The Economist pointing out the US giant still has a market share of “about 80%” for the PC processors known as CPUs).

But their grip is now being called into question by Chinese ambitions to foster homegrown rivals. In Grove’s reasoning, this could turn out to be the moment when the fundamentals change forever as the Chinese move forward from making products containing the all-important chips to manufacturing the chips themselves.

In the first of a two-part series WiC looks at China’s progress in manufacturing semiconductor chips for the two products that have dominated the chip industry over the past two decades: computers and mobile phones.

In the second, we will examine its position in products likely to dominate the coming decade such as self-driving cars and deep-learning machines.

To build China’s chip capacity, government guidelines drawn up in 2014 envisaged $150 billion of investment through state subsidies and a raft of newly-created private equity funds under the provincial and municipal governments.

The plans were refined in 2015 with specific targets to boost domestic semiconductor output to 40% of the global total by 2020 and 70% by 2025 (the Chinese consume about 60% of the world’s semiconductors, but make just 10% of them).

Until three or four years ago, most of the expansion in homegrown production centred on chip assembly and packaging. A key problem was that the country’s leading producers Semiconductor Manufacturing International Corp (SMIC) and Hua Hong Semi weren’t challenging the industry leaders in technical terms, manufacturing chips using ‘trailing-edge’ technology. But that situation is starting to change. Analysts say SMIC is less than two years behind one of Taiwan’s leading producers, United Microelectronics Corp (UMC), although it is still about five years back on global leader Taiwan Semiconductor Manufacturing Corp (TSMC).

Domestic companies are also the drivers of new foundry investment in China rather than the foreign firms. SEMI, an industry association, believes they will account for over two-thirds of foundry investment by 2019.

One of Barack Obama’s final acts as US president was commissioning a report on the magnitude of the Chinese challenge. Earlier this year, the President’s Council of Advisors on Science and Technology published its conclusion: China is already undercutting the American position. That view was reinforced this week with the publication of a report by the European Union Chamber of Commerce in China that complains that the Chinese are closing the technology gap by unfair means (primarily through subsidies and guided investment funds). The study sees this as a state-driven strategy, describing China’s ‘Made in 2025’ programme as a “large scale import substitution plan aimed at nationalising key industries, or at least severely curtailing the position of foreign businesses in them, both as suppliers of key components and finished goods”.

Added to the list of complaints by the foreign foundries is that they are being forced into technology transfer in exchange for market access but then finding that Chinese customers are being compelled to source chips from domestic suppliers.

Part of the American response: to defend its know-how on national security grounds. Washington’s Committee on Foreign Investment (CFIUS) has blocked Chinese companies in a number of their attempts to acquire US technology. In 2016, this resulted in failed bids from the omnipresent Tsinghua Unigroup, whose Unisplendour offshoot wanted to buy a 15% stake in hard disk manufacturer Western Digital for $3.78 billion.

In another case, CFIUS effectively blocked Fujian Grand Chip Investment Fund’s €670 million acquisition of Aixtron, which deploys gallium nitride technology for semiconductors utilised in the defence industry.

Fear of regulatory intervention also prompted Fairchild Semiconductor to reject a bid from China Resources Microelectronics and Hua Capital Management (an integrated circuit fund backed by Beijing’s local government) in favour of a lower offer from an American competitor.

Finding it harder to acquire foreign technology, China has resorted to hiring more of the industry’s formative talent, particularly from Taiwan. This recruitment policy (sometimes of retirees, admittedly) has stung the Taiwanese, leading to talk about the hollowing out of its own chip sector. Chief among the hires are TSMC’s former COO Jiang Shangyi who has gone to SMIC and Charles Kau (known locally as the godfather of the DRAM industry) who has joined Tsinghua Unigroup.

Tsinghua Unigroup is leading the campaign to boost Chinese output. Having failed to buy DRAM manufacturer Micron in 2015, it has been focusing more on homegrown M&A, ploughing $70 billion into the construction of foundries in Wuhan, Chengdu and Nanjing. In Wuhan it is building a $24 billion plant via its offshoot Changjiang Storage (also known as Yangtze River Storage), formed last year through a merger of Unigroup’s chip-making operations with XMC (Xinxin Semiconductor Manufacturing). Much of the capital is coming from state-backed investment funds.

After the success in boosting the potential volume of their chip production, the Chinese still need to develop the capacity to make the highest-performing products.

Mastering 3D NAND – the most powerful flash memory technology – is one of the key tasks ahead in meeting demand from makers of mobile devices and computer servers.

Unigroup’s investment trajectory shows that the Chinese chip makers are moving away from trailing edge technology towards leading edge design. At February’s Mobile World Congress in Barcelona, it unveiled a processor called the SC9861G-A, the third SoC (system on chip) from its design unit RDA Spreadtrum. Developed for mid-range smartphones, the integrated circuit has a 14nm 8-core 64-bit SoC with a CAT7 LTE modem (able to download up to 300 megabits per second and upload 100Mbps). The chips will go into mass production in the second quarter of this year. But they rely heavily on Intel technology and they are being contract manufactured at an Intel foundry in China on Spreadtrum’s behalf.

(Intel’s 2014 purchase of a 20% stake in RDA Spreadtrum was a highly strategic move to make sure it remains a case of ‘Intel inside’ where China is concerned).

Chinese companies may be taking the lead in building new foundries, but the global leaders are trying to stay in the game. California-based Global Foundries recently unveiled a $10 billion investment at a facility in Chengdu, for instance. As the Economic Observer reports, the plant will take advantage of local subsidies and stay cost-competitive by locating close to smartphone manufacturers, which use its chips.

A few of the mobile phone brands are chipmaking themselves, including Xiaomi which has moved up the value chain with its first self-designed processor called Pengpai 1. The progress makes Xiaomi the second Chinese manufacturer with chip-making capacity, alongside Huawei. Lei Jun, Xiaomi’s founder, tells CBN that developing in-house technology is the only way to avoid being knocked out in the fiercely competitive mobile phone market. “If you want to become a great company you need autonomy in the core technology,” he says.

But as Digitimes notes, while Huawei has also developed its own SoC, it uses it more as a “bargaining chip” to get better prices from its main suppliers, Qualcomm and Mediatek. Xiaomi may try the same.

The R&D costs to make semiconductors at ever-smaller nodes are prohibitively expensive, rendering it unaffordable for all but the largest players. Telecom equipment maker ZTE failed to commission application processors in-house, for instance, and Lei Jun says that it cost Rmb1 billion for Xiaomi to make the leap. But as well as giving their developers more bargaining power, the pioneers could provide a leg-up for the sector as semiconductor firms push for the next round of breakthroughs needed to deliver the advances in processing power required for self-driving cars and the artificial intelligence of deep-learning machines.

---

© ChinTell Ltd. All rights reserved.

Sponsored by HSBC.

The Week in China website and the weekly magazine publications are owned and maintained by ChinTell Limited, Hong Kong. Neither HSBC nor any member of the HSBC group of companies ("HSBC") endorses the contents and/or is involved in selecting, creating or editing the contents of the Week in China website or the Week in China magazine. The views expressed in these publications are solely the views of ChinTell Limited and do not necessarily reflect the views or investment ideas of HSBC. No responsibility will therefore be assumed by HSBC for the contents of these publications or for the errors or omissions therein.