After deferring a potential Hong Kong IPO estimated to raise USD 750 million, SenseTime, the world’s most valuable AI unicorn backed by giants like Alibaba, SoftBank, and Qualcomm, has turned to the primary market to raise between USD 1 billion and USD 1.5 billion instead, LatePost and Nikkei Asian Review reported.
SenseTime was not the only company that retreated from a public offering at the last minute.
Its computer vision peer Megvii, who reportedly initiated the IPO preparations back in 2018, saw its Hong Kong Stock Exchange application expire this February after it appeared on the US government’s entity list last year.
Not long after the IPO was scrapped, Lenovo Venture Capital as well as Sinovation Ventures, one of the most active funds in artificial intelligence, redeemed their shares and exited Megvii’s shareholder lineup as public records on local business database Qichacha showed.
Megvii later rebutted the then rampant coverage on the exits, claiming in a statement “there are no exits of stakeholders at the group level” and all its strategic backers “maintained sound tie-ups” with the company.
These companies’ advanced computer vision technologies have demonstrated strength and promise, especially during the COVID-19 global health crisis, in areas including facial recognition, medical imaging diagnosis among others. However, they are still facing difficulties in fundraising.
A combination of factors ranging from capital intensive R&D process, low profitability, high valuation, to the US-imposed sanctions that tempered investor sentiment, all contributed to Chinese AI startups’ current conundrum.
The four Chinese computer vision tigers
Nicknamed as “the four Chinese computer vision tigers”, SenseTime and Megevii, together with CloudWalk and Yitu Technology are widely considered as China’s four leading startups in the field of computer vision.
Though majoring in different fields and coming from different backgrounds, the four share a couple of things in common. First, the high valuation. CloudWalk, the world’s largest facial recognition device maker in 2018 collected an RMB 1.8 billion (USD 254 million) investment from a group of provincial and municipal funds this May to be valued at USD 5.1 billion, according to data service provider ITjuzi.
SenseTime’s new fundraising, if secured as planned within 2020, would raise its valuation to around USD 10 billion, according to 36Kr, further consolidate its position as one of the country’s most valuable AI startups.
Second, they are all among the companies blacklisted by the US for American companies to do business with. Though the four companies conduct business mostly in China, their presence on the “entity list” has no doubt pose challenges to potential international expansion, plus stifling the fundraising possibility from foreign capital markets.
Notably, except for SenseTime, all the other three “tigers” have participated in the construction of China’s surveillance camera and facial recognition network. Though Megvii has shifted its business focus to smart city and internet of things, according to its 2019 prospectus.
High valuation, low profitability
Moreover, the four leading computer vision enterprises are known as “fundraising behemoth”. Founded between 2011 and 2015, they have each collected on average 8 rounds of investments so far, with well-known investors including Sequoia Capital, SoftBank, Temasek, and Alibaba.
On the flip side, the frequency of fundraising in this sector illustrates how much these AI firms depend on investors’ funding while reaching profitability remains a challenge.
Part of the reason is AI research and development cost a lot, especially when it comes to hiring talents.
According to Megvii’s prospectus, the company’s net loss has been widening over the years, while its gross profit margin in the first half of 2019 was 64.4%. The company has significant R&D expenses, including the recruitment of high-end AI talents.
Other reasons for the troubling profitability can be found in their business model, computer vision firms’ clients are usually governments and large enterprises with long payment cycles, leading to increased pressure on the company’s cash flow.
In 2017, 2018, and the first half of 2019, Megvii’s receivables were USD 22 million, USD 150 million, and USD 220 million, respectively.
As one of the earliest computer vision startups, SenseTime has a relatively more mature business model compared to others, 36Kr reported, it claimed to have earned a profit in 2017. In 2019 the company earned an expected revenue of USD 750 million, an increase of 200% from the previous year, Reuters reported, but it still had a negative cash-flow.
2019 capital freeze
China had its AI fever in the early 2010s. Based on iYiou’s research, the number of newly founded AI companies increased sharply in 2012 and spiked again in 2016. The total amount of investments in the sector also peaked in 2018, totaling USD 18.4 billion.
SenseTime, for example, bagged a total of USD 2.2 billion from as many as three rounds of investments in 2018, from investors like SoftBank China, Qualcomm Ventures, Temasek, and Alibaba. While Yitu also collected around USD 300 million from two rounds of fundraisings in that year.
However, since 2019, the venture capital landscape has cooled down and investors have become more cautious, questioning the profitability of AI firms.
Last year, China’ AI start-ups attracted less money compared to the US, they raised slightly more than USD 2.9 billion, compared to about USD 4.7 billion the year before, while US AI start-ups raised USD 17 billion last year, compared to roughly USD 13.3 billion the year before.
In China, more than half of the fundraising deals in the AI sector so far are Series A and B rounds, as per ITjuzi’s analysis. Investors are increasingly focused on a path to profitability, while charismatic founders pitching unrealistic idealism are less valued in a more conservative investment climate.
Update: CST 13:22, July 22, 2020, added Megvii’s statement on the matter of investors’ exits.