By 2025, China is to be the world’s leader in some AI technologies and their applications, while making major breakthroughs in fundamental AI theory. Applications will include smart manufacturing, smart healthcare, smart cities, smart agriculture, and national defence, with the value of core AI industries reaching 400 billion RMB (US$57.41 billion), while AI-related industries reach 5 trillion RMB (US$717.69 billion).
While the technical jargon of artificial intelligence (AI) may often seem opaque – “convolutional neural networks”, “machine learning”, “inferencing” – the applications can be extremely practical. Driver safety is a prime example, where complex camera- and radar-derived datasets need to be processed and finally “inferenced”, in the jargon, in milliseconds if not microseconds. Add several more years of R&D, and it could be AI that will make self-driving cars a reality.
China’s leaders have not been slow in making some inferences of their own, and in July 2017, the State Council officially released the Plan for the Development of New Generation Artificial Intelligence. The goals are certainly ambitious. By 2020 (this year!), China is to have the world’s most advanced level of AI technology. AI becomes an important driver of economic growth, with core AI industries reaching 150 billion RMB (US$23.53 billion), while AI-related industries are to reach 1 trillion RMB (US$143.54 billion).
By 2030, China is to be the world leader at all levels in AI theory, technologies and applications.
Core AI industries are to exceed 1 trillion RMB (US$143.54 billion), while AI-related industries are to exceed 10 trillion RMB (US$1.43 trillion).
Huawei – China’s AI world-beater
While these top-down aspirations might be dismissed as unrealistic, by some in the industry, in reality the Shenzhen-based communications giant Huawei is now the world leader in AI “training”.
Adding to its AI processor portfolio, which already included the company’s Ascend devices, in September 2019 Huawei released Atlas 900, the world’s fastest AI training cluster. Delivering the processing power of several thousand Ascend units, the Atlas 900 takes only 59.8 seconds to train ResNet-50, the gold standard for benchmarking AI training performance. This beats the previous world record by 10 seconds.
Atlas 900 is a powerhouse of AI computing and will bring new possibilities to different fields of scientific research and business innovation – anything from astronomy, weather forecasting, and autonomous driving, to oil exploration.
“The future of computing is a massive market worth more than two trillion US dollars,” said Ken Hu, Huawei’s Deputy Chairman. “We’ll keep investing with a strategy that focuses on four key areas. We will push the boundaries of architecture, invest in processors for all scenarios, keep clear business boundaries, and build an open ecosystem.”
AI by the BAT
China’s established triumvirate of internet leaders, Baidu, Alibaba and Tencent, collectively known as the BAT, have also been busy venturing into AI technology.
Online search specialist Baidu, the “Google of China”, is involved in three major AI-related ventures – Apollo, an ambitious open-source project in autonomous driving, global in scope; DuerOS, a voice-enabled digital assistant; and Baidu ABC, a ‘cloud’ platform for businesses.
Having built some AI credibility, in 2018 Baidu announced a partnership with Apple, Amazon, and Google. This four-cornered arrangement intends to establish best practices for AI, and study its social impact.
In September 2019, Alibaba announced an AI accelerator chip, the Hanguang 800. Built in a 12nm process, the Hanguang 800 contains 17 billion transistors and is capable of processing 78,563 images per second (IPS) and 500 IPS/W (images-per-second-per-Watt), when benchmarked in ResNet-50. On the strength of these results, Alibaba CTO Jeff Zhang claimed that the Hanguang 800 is 15 times more powerful than Nvidia’s T4 GPU and 46 times more powerful than the Nvidia P4 GPU.
Tencent is well known as the intelligence behind its hugely popular and versatile WeChat social media platform and for being the world’s largest video game publisher. Tencent launched an AI Lab in China in 2016 and then in 2017 opened an AI R&D centre in the US, in Bellevue, just across Lake Washington from Seattle, to work on speech recognition and natural language processing (NLP). Presumably this is the brainy tech that now enables users of WeChat to communicate while using different languages.