Get the DealBook newsletter to make sense of major business and policy headlines — and the power-brokers who shape them.
China’s facial recognition start-ups are attracting big investments at a time when Silicon Valley’s giants are facing a backlash over their use of the technology.
In the past week, Chinese facial recognition companies, according to a pair of reports, were close to raising as much as $1.6 billion. Those investments would build on billions of dollars that investors have already put into the companies.
The investments reflect facial recognition’s embrace by Beijing, which has created a better environment for companies to test the technology than in the West. And that could help China deploy advanced systems faster and more broadly than America.
SoftBank’s $100 billion Vision Fund was seeking a $1 billion stake in SenseTime, an artificial intelligence company in Beijing, Bloomberg reported last week, citing anonymous sources. SenseTime makes facial recognition software used in applications like monitoring CCTV footage and checking a consumer’s likeness to validate digital payments.
On Monday, Bloomberg reported that, according to anonymous sources, the Chinese A.I. company Megvii would close a $600 million funding round “within weeks.” Alibaba and Hong Kong-based Boyu Capital are expected to contribute. Megvii is the developer of Face++, the facial recognition system already used in China to grant or deny people access to buildings, or check if the driver of a Didi ride-hailing car is legitimate, among other things.
SenseTime has already raised over $1.6 billion in funding since it was founded in 2014, according to data from S & P Global Market Intelligence, and as recently as May was valued at about $4.5 billion. Investors included Alibaba and Bank of China as well as the American firms Qualcomm, Silver Lake, Tiger Global Management and Fidelity International.
Megvii’s previous funding round raised $460 million from Chinese companies like Foxconn and Ant Financial, as well as several domestic investment funds. It also received money from the Moscow-based Russian Direct Investment Fund and South Korea’s SK Group.
The facial recognition systems being built by SenseTime and Megvii use artificial intelligence to discern individuals from one another in images or video, by studying libraries of existing content. It is one of the more established machine learning techniques, which means that it has found some practical application in consumer products — such as Apple’s Face ID, which allows a user to unlock his or her iPhone by looking at it, or Facebook’s tools, which can recognize users in uploaded images.
Such software has proved controversial because of the privacy issues that it raises. Law enforcement’s use of it to identify criminals or track citizens, in particular, has drawn pushback.
But even if its citizens have such concerns, China’s government wants to use facial recognition to create a vast national surveillance system. And it has created an environment to foster its development.
The country’s push to become a technological powerhouse has created a wave of promising A.I. start-ups. Its desire to track its citizens has also meant that many of them have focused specifically on image recognition. And there are close ties between many of these start-ups and the government, which allows companies to apply their software to huge state data sets. The chief executive of SenseTime, which has contracts with the Police Department in Chongqing and China Mobile, told The Financial Times in January that company had “processed 500 million identities for facial recognition. U.S. companies can’t test on so many customers.”
Similar arrangements have existed in the United States. But American technology companies have recently run into resistance when working with government agencies.
Amazon endured heavy lobbying from shareholders, the American Civil Liberties Union and staff about the use of its Rekognition software by police to identify people. Microsoft, too, has suffered blowback over a project to help develop A.I. for Immigration and Customs Enforcement. And Google ultimately decided not to renew a contract with the Pentagon that called for using image recognition to improve the performance of drone operations.
In the wake of those criticisms, Microsoft has actually called on Congress to regulate the use of facial recognition technology.
The nature of A.I. means that software systems can be improved by crunching through larger and larger data sets that can be provided by the government — from things like CCTV footage, driving license images, and more. China’s start-ups, then, are enjoying the benefits of improving their technology inside projects that America’s tech giants might be lobbying their own way out of.