SenseTime’s Hong Kong IPO could be a boon for China’s controversial ‘A.I. dragons’

August 30, 2021, 11:47 AM UTC

SenseTime Group, China’s most valuable artificial intelligence startup, has filed for an initial public offering in Hong Kong, potentially clearing the way for similar offerings by other Chinese A.I. ventures whose previous efforts to go public have been hindered by U.S. allegations of their ties to state security agencies.

The Hong Kong–based company, which makes A.I. technology for interpreting images, will reportedly seek to raise up to $2 billion. That would make SenseTime’s the richest offering by a Chinese technology startup since Beijing cracked down on the sector last month, and would give the venture a market valuation of as much as $12 billion.

SenseTime’s backers include Qualcomm Ventures, SoftBank Vision Fund, Silver Lake Partners, Temasek Holdings, and Alibaba Group Holding. The company has retained Chinese investment banks China International Capital Corp. and Haitong International, as well as HSBC, to help it pursue the listing, according to documents filed with the Hong Kong stock exchange Friday.

SenseTime is the largest of four Chinese “A.I. dragons” known for their capacity to scan and recognize faces from vast data sets at lightning speed. Analysts expect the other three dragons—Megvii Technology, Yitu Technology, and CloudWalk Technology—to float shares in the coming months, collectively raising about $2.7 billion.

Yitu is expected to seek permission to list in Hong Kong, while Megvii and CloudWalk have filed to go public on Shanghai’s Nasdaq-like STAR Market.

The pandemic has boosted sales for the four dragons as lockdowns increased demand for facial recognition software in China. Provincial and local governments across the country rely on cameras made by the companies to read temperatures, confirm that people are wearing masks, and verify identities even when faces are covered.

“The COVID-19 pandemic is expected, in the long run, to accelerate the digital transformation of enterprises and city management, indicating more opportunities for the A.I. industry,” SenseTime said in its filing with the exchange.

But the dragons aren’t yet profitable, and their efforts to raise the additional capital they need to keep growing have been caught in political crosscurrents at home and abroad.

Caught in the crosscurrents

In China, regulators last month began issuing a raft of new rules aimed at strengthening government oversight of technology firms, particularly those whose business involves collecting and analyzing large volumes of consumer data. In the aftermath of the crackdown, scores of Chinese startups have called off planned listings in the U.S. and Hong Kong.

China is preparing to implement two key laws relating to tech firms’ handling of data. A Data Security Law will be enforced starting this Wednesday. A Personal Information Protection Law will take effect on Nov. 1.

In its filing Friday, SenseTime acknowledged that it is subject to “complex and evolving” laws and regulations regarding privacy, data protection, and cybersecurity in China and “cannot predict the impact of the draft measures.”

The A.l. companies remain vulnerable to moves in Washington as well. SenseTime, Megvii, and Yitu were among 28 Chinese tech companies placed on a trade blacklist by the Trump administration in October 2019 amid U.S. allegations that the companies supported state surveillance of Muslim Uyghurs in the northwestern Chinese province of Xinjiang. The action barred the firms from buying components from American companies without Washington’s approval. The U.S. added CloudWalk to the blacklist in May 2020.

U.S. claims that the A.I. companies had committed human rights abuses in Xinjiang derailed Megvii’s 2019 attempt to float shares in Hong Kong. The company submitted a 600-page document to convince the Hong Kong exchange’s listing committee that it had only minimal ties to Xinjiang, with only 1% of its 2018 revenue and none of its revenue in the first half of 2019 generated by projects in the province.

The listing committee remained unconvinced. And the timing of Megvii’s appeal—at the height of Hong Kong’s pro-democracy demonstrations—didn’t help. Activists bombarded the Hong Kong exchange with petitions objecting to Megvii’s application. Eventually, Megvii dropped its bid to list in Hong Kong and refiled with Shanghai’s STAR Market.

SenseTime’s Friday filing acknowledged that if it remains on the U.S. blacklist on a “prolonged basis, we may not be able to compete effectively in certain business lines, and our business…could be materially and adversely affected.”


Analysts are optimistic about the four firms’ prospects for winning permission to list this time around. SenseTime, along with smaller A.I. hopefuls including 4Paradigm Technology and Qingdao Ainnovation, is likely to see its applications to list in Hong Kong approved because the exchange there is now more eager than ever to “attract Chinese homegrown businesses,” and is “locked in a tight race with the mainland’s STAR and ChiNext technology exchanges,” argues Bruce Pang, head of macro strategy and research at China Renaissance Securities.

Thomas Gatley, senior analyst at Beijing-based Gavekal Research, expects SenseTime to win approval for a Hong Kong IPO as long as members of the exchange’s listing committee are convinced the venture has the blessing of regulators in Beijing.

Others say the public outcry that scuttled Megvii’s 2019 IPO bid is virtually unthinkable in Hong Kong since Beijing pushed through strict new national security legislation last June. Since the passage of the law, Hong Kong authorities have charged dozens of pro-democracy activists, jailed several, and shut down a popular pro-democracy tabloid. “It’s highly unlikely that we’ll see the same sort of citizen petitions against SenseTime [or other Chinese A.I. companies] given the current atmosphere,” says one Hong Kong–based corporate lawyer. “People are fearful of what will happen if they protest against the wrong thing,” he notes.

At the same time, China’s A.I. companies have worked to minimize their ties to Xinjiang. SenseTime in 2019 divested from its joint-venture businesses in the region by selling stakes in Leon Technology, which provided the local government with security and surveillance systems, and SenseNets, which sells facial recognition and crowd analysis technology.

In a South China Morning Post ad this month, SenseTime reiterated its commitment to “ethical A.I.,” noting that the technology’s rapid evolution prompts “the question of how it can be harnessed for the greater good…and not fall into…abuse or misuse…SenseTime does not…shy away from these challenges.” It released its “Code of Ethics for A.I. Sustainable Development” in June, in collaboration with the United Nations.

SenseTime’s revenue grew 14% in 2020 to nearly $526 million; revenue in the first three months of this year surged 92% to $247.4 million. The U.S. blacklist has not prevented SenseTime from expanding in other markets like Japan and Southeast Asia. By this August, global daily users of SenseTime’s AR (augmented reality) technology grew to 100 million—with Thailand and Malaysia as especially popular markets.

Given the size of China’s market “they can continue ignoring the U.S. and look to other markets, especially along the Belt and Road,” says Adam Segal, director of the digital and cyberspace program at the Council on Foreign Relations.

Financial security

Experts say the big challenge facing China’s A.I. startups will be convincing regulators and investors that they can eventually turn a profit, not just increase sales. Founded in 2014 by computer scientist Xu Li, who previously worked at Lenovo and Microsoft, and Tang Xiaoou, a professor of information engineering, the two set out to commercialize computer vision and A.I. tech. Together, Xu and Tang developed DeepID, a facial recognition tech that has (according to SenseTime), a 98.52% accuracy rate—surpassing its Facebook counterpart.

Research and development remains an enormous expense for the four dragons. Yitu, for instance, bled $1 billion from 2017 to the first half of 2020. SenseTime’s operating loss jumped 13% to $278.3 million last year, and reached $324.7 million for the first six months of this year.

China’s A.I. firms haven’t yet built up a “sustainable, long-term business model” given their heavy dependence on government clients, says Nina Xiang, author of Red AI: Victories and Warnings From China’s Rise in Artificial Intelligence.

Yitu offers A.I. tech for developing smart cities that is applicable to the health care and financial industries. It counts the Shanghai government as a key client. The venture’s top five clients are state-linked and contributed 62% of its 2020 revenue. SenseTime says it has up to 1,200 clients from both the public and private sectors. But the company is best known for its facial recognition tech, which is used across Chinese cities and subway systems for public surveillance and security.

In the end, it may be government support that keeps China’s A.I. giants going. When the COVID-19 pandemic hit, business boomed for SenseTime. Its ties with Chinese authorities increased, as local governments dialed up their use of SenseTime technologies to track citizen movement, temperature checks, and mask-wearing in order to curb infection outbreaks beginning last year.

It helps that the companies are perceived as vital to China’s bid to become a global A.I. leader by 2030. “The state can’t afford to bail on its A.I. champions now,” says Kendra Schaefer, head of tech policy at consulting firm Trivium China.

Subscribe to Eastworld for insight on what’s dominating business in Asia, delivered free to your inbox.