Global Engineering Capability Review

China case study:

AI surges as production plays catch-up

China's Engineering Index scores

Knowledge = 2nd
Labour force = 7th
Engineering industry = 4th

Infrastructure = 63rd
Digital infrastructure = 44th
Safety standards = 70th

China is aiming to become a global leader in artificial intelligence. AI is generally defined as “the ability of a machine to perform cognitive functions associated with human minds, such as perceiving, reasoning, and learning”.[53] Many AI applications are already used widely, such as virtual assistants built into smartphones and home-management systems, and facial-recognition software used by law enforcement. AI applications vary considerably, but they share a dependence on hardware (notably semiconductors) that enables logic and memory functions.[54]

Although China is ahead of the pack in AI implementation and has invested widely in applications, it does not produce many of the core technologies that power AI, such as operating systems, chips and software. This can be explained by China’s relatively late entry into the development of such technology, as a result of a long-standing focus on entry-level manufacturing. In recognition of the value of domestic production, the country is now shifting its attention to creating its own software.

AI: High on the menu

In mid-2017 the State Council (China’s chief administrative authority) issued the New Generation Artificial Intelligence Development Plan (AIDP). This document, together with a 2015 plan entitled Made in China 2025, presented the country’s AI agenda, making China one of the few countries to have released an AI strategy. The new plan frames AI as a focus of international competition and a strategic technology for future economic growth and national security. China’s state and local governments have responded by committing huge sums to investment in AI development. (While the total commitment is not publicly disclosed, at least two regional governments have each committed to investing 100bn yuan, approximately US$14.7bn). [55]

In October 2018 China’s president, Xi Jinping, held a Politburo study session on AI. Mr Xi’s comments echoed the main conclusions of both the AIDP and Made in China 2025, namely that China should “achieve world- leading levels” in AI technology and reduce its vulnerable “external [foreign] dependence for key technologies and advanced equipment”. [56] A Tsinghua University report published in 2018 found that 'China has secured a leading position in the top echelon in both technology development and market applications and is in a race of ‘two giants’ with the US.' It ranked China as the top producer of AI patents, the highest-placed country for AI venture-capital investment, and second for the number of AI companies and the size of its AI talent pool. [57]

China’s latest action plan for implementing its strategy presents four major tasks for 2018-20:

1. Identifying targets for the development of “smart products”, such as networked vehicles, intelligent service robots and video image identification systems;

2. Pursuing technological breakthroughs in “core foundations”, such as neural-network chips;

3. Nurturing the development of “intelligent manufacturing”; and

4. Building a public support system by accelerating the development of an “intelligent next-generation internet”. [58]

A Chinese guest takes food delivered by a robot in the restaurant at Alibaba's futuristic hotel "Flyzoo Hotel" in Hangzhou City, Zhejiang Province

Getting hungry: China’s demand for semiconductor chips

Trade tensions between the US and China highlight the latter’s institutional weakness in developing the core technologies and semiconductors that power AI. The experience of Chinese telecom giant ZTE is revealing. The company’s business ground to a halt when the US government prevented it from buying American-made chips in early 2018. The US later offered a concession, but at no small cost to ZTE, which was compelled to restructure its management team and pay a US$1bn fine. A computer scientist at Tsinghua University put ZTE’s woes down to 'inadequate ‘core technology’ and weak innovation'. [59]

Chips are at the heart of this issue. China accounts for more than 50% of global demand for chips but produces only 8% of those that it uses, according to a Chinese newspaper, Yicai. In 2016 the country imported US$227 billion worth of semiconductors – almost double the amount spent on crude oil shipments. [60] China’s weakness in developing chips stems primarily from its late start in semiconductor manufacturing. This is compounded by the low profit margins for chip businesses and the country’s poor previous investment in R&D, especially compared with Japan and South Korea.

Crucially, the labour force is also lacking in terms of both skills and scale. In 2017 fewer than 300,000 people worked in the domestic chipmaking industry in China, and reaching the government’s goal of increasing the industry’s size fivefold by 2030 will require an estimated 400,000 additional employees. This gap is linked to limited educational resources and career prospects, as well as poor salaries. [61] Although Chinese engineers tend to be very well trained, developing the core technologies for AI requires a different set of industry skills and deep research knowledge. Additionally, many of China’s engineers who are educated abroad tend to seek employment in the West, and large Chinese companies can lose their best talent to foreign competitors.

Finally, the emergence of several vast companies in China over the past decade, such as Huawei, Alibaba and Tencent, has signalled a change in the way in which industrial R&D is carried out. As an expert noted, companies would previously contract universities to develop new equipment. Today, universities are less well funded and big firms have the resources to do much of their own R&D, resulting in lower rates of collaboration between the two realms.

The emergence in China of several vast companies, such as Huawei, Alibaba and Tencent, over the past decade has signalled a change in the way that industrial R&D is carried out

Chips are at the heart of the challenge. China accounts for more than 50% of global demand for chips but produces only 8% of the chips that it uses

R&D: The secret sauce for China’s semiconductor sector

China is seeking to reduce its dependence on semiconductor imports by strengthening its domestic semiconductor sector. The country has launched a fund focused on semiconductor development (the China National Integrated Circuit Industry Investment Fund), which in mid-2019 raised around US$29bn in its second round of financing.

The ongoing US-China trade war has certainly made greater self-reliance in this sector more of a priority. [62] Other measures to encourage domestic production include tax breaks for up to five years for local chipmakers. [63]

Investment in R&D will need to be coupled with the development of a talented labour force. Even if China invests substantially in its semiconductor sector, it will be forced to compete with established US giants such as Intel, which spends more than US$10 billion annually on R&D. For the investment to succeed, China will need to cultivate and retain a talent pool to lead innovations in AI. China can also strengthen career services for graduates, and improve benefits for engineers doing research in electronic engineering and computer science, in order to reduce the number of graduates leaving to work abroad. [64]

Letting the chips fall

China is implementing AI at a rapid pace but lacks the domestic capability to develop the core technologies it requires. This is a major concern for the country, especially in light of the trade war with the US, and its resolution could lead to various economic and security benefits. To try to wean industry off its reliance on imports, China has begun various investments and provided incentives to encourage domestic production. However, the success of this approach may be limited without the parallel adoption of a more interdisciplinary learning model (in place of the current insular approach). The sector would also benefit from more collaboration between universities and industry, improved funding for science professors, and measures to improve English language proficiency. Given that the payoff for R&D investment in AI can take time to materialise, this will be a long-term project.

The sector would benefit from more collaboration between universities and industry, improved funding for professors in core science topics


Footnotes