Source: Xinhua.
As one of the three leading industries prioritised under national strategy, Shanghai’s artificial intelligence industry passed the 200 billion yuan mark in the first three quarters of this year, and growth above 25 percent for the full year is all but certain.
AI companies have started to report revenue matter-of-factly: SenseTime took in 1.65 billion yuan in the first half of the year, nearly doubling year on year; Xiaodu AI, set up by Baidu in Shanghai last October, earned more than 1.5 billion yuan over the same period.
That is a far cry from three or four years ago, when unicorns avoided revenue talk in favour of funding rounds and valuations, drawing scepticism about bubbles and losses and debate over how AI could reach deep industrial application. One company’s co-founder admitted at the time that AI that creates no real value would eventually be washed out. So why do Shanghai’s AI firms now dare to talk revenue, and what gives them “real value”?
Scenarios
“Real value is when the traffic police want a helmet-and-belt campaign, the housing authority wants to keep e-bikes out of lifts, and the land bureau wants to spot farmland quietly turned into fish ponds from satellite video. Demands come one after another, and they want a plan the next day,” said Xu Yan, vice president of YITU Technology, describing the urgency he has felt this year from city and district urban operation centres.
That urgency is shared by Tencent, Alibaba, SenseTime and other AI companies building Shanghai’s smart city. YITU alone has landed more than 50 concrete business scenarios in 16 fields this year, covering more than half of Shanghai’s districts.
Pragmatism became the keyword of Shanghai’s AI scene from the start of 2021. On the third working day of the year the city government signed with Horizon Robotics to build global headquarters for automotive AI chips and intelligent vehicle central computing platforms; in May it signed with Xiaoice, whose AI system carries more than 60 percent of global interaction volume, for global commercial operation and R&D headquarters. Behind these global-scale projects lies Shanghai’s dividend: the city itself is the biggest application scenario for the AI industry.
Shanghai pioneered the country’s AI scenario-opening programme at the end of 2018, publishing scenarios dynamically and soliciting solutions worldwide. Three batches have opened 58 scenarios across healthcare, education, culture, urban management, justice and finance, and this year’s World AI Conference added five more, including Pudong-led digital governance, the Lingang digital twin city and smart mobility in Jiading.
Technology
Scenario pull is one side; hard technology, laid over years, is bursting out this year.
Seven AI chips have been released in Shanghai so far this year, each a heavyweight domestically. At the WAIC exhibition, an axis of 17 chips displayed Shanghai’s generations in production, in trial and in exploration. Algorithms are equally strong: the Shanghai AI Laboratory, SenseTime and several universities recently released INTERN, a new-generation general vision system expected to handle over a thousand tasks with one model. United Imaging’s lung nodule CT detection software won one of the country’s rare Class III medical AI certificates, and Iceberg Biotech’s AC0682 investigational drug was approved by the US Food and Drug Administration, putting the world’s first AI-developed breast cancer drug into clinical trials.
Beyond chips and algorithms, intelligent hardware made in Shanghai reached scale application this year. CloudMinds’ robot industry base will start production early next year with an annual capacity of 10 million units, and robots from Broad Systems and Keenon are already sold to dozens of countries. Industry insiders sum it up: if the north is strong in algorithms and the south in smart hardware, Shanghai is strong at both, with chips strongest of all.
Ecosystem
Even as a recognised AI hub, Shanghai is planning ahead. The city’s first five-year plan for artificial intelligence is essentially drafted, and Shanghai plans to lead with local standards in key AI fields, having consulted more than ten municipal departments and issued the country’s first guidance on AI standardisation with the market regulator. “A fast-developing AI industry will inevitably leave gaps. Local standards as a first-mover tool give companies a basis to follow, enabling orderly development and safer agile governance,” said Sun Yue, deputy director of the AI development division of the municipal economy and information commission.
Shanghai is also clear-eyed about algorithms and computing power. AlphaFold, which predicts protein structures and solved a half-century biology problem, marks the kind of key algorithm China’s AI industry still lacks. To close the gap, the city issued an algorithm innovation action plan for 2021-2023, aiming to break through key algorithms in government services, autonomous driving, industry, healthcare and finance through open-source ecosystems and trading mechanisms, a rare plan devoted to algorithms alone.
Computing power is an equally big challenge: a single future street camera may run a dozen algorithms at once. SenseTime’s large AI computing centre in Lingang, expected to start trial operation within the year, will be one of Asia’s largest and can support the digital governance of four megacities of 20 million people each. And in another sign of resilience, Shanghai has won the right to host the 2024 congress of IJCAI, one of the world’s three major AI organisations. With harder technology, a larger industry and a fuller ecosystem, the shift from valuation talk to revenue is the natural result of years of accumulation.
Want to talk through what you just read?
We work end to end, from product design to launch. Tell us about your scenario and we can start there.
Shenzhen XZS Technology Co., Ltd. · Tel 0755-82494862 · 13760424861