What does Megvii do?

Megvii is often referred to as one of China's "four AI dragons" alongside SenseTime, Yitu and CloudWalk. Founded in 2011 in Beijing, Megvii is the oldest of the group. The three founders of the company, Yin Qi, Tang Wenbin and Yang Mu, were all computing protegees who received their training through the highly-selective experimental computer science program at Tsinghua University.

The company is the creator of the facial-recognition software Face++, the world's largest open-source computer vision platform. In 2020, Megvii launched Brain++, its own AI productivity platform, and made MegEngine, a major component of Brain++, an open-source deep learning framework. This is largely seen as one part of China's plan to develop home-grown AI technology, reducing Chinese companies' reliance on U.S. open-source frameworks.

The company has three main business segments: consumer IoT, city IoT and supply chain IoT. The city IoT business primarily entails technological solutions for local government — not just Smart City programs, but Smart Community management platforms, which involve monitoring and managing the individual residential compounds in which many Chinese urban residents live. China kicked off a nationwide Smart City program in 2012 with the goal of using new technologies made possible by IoT to aid policing, city management and the "maintenance of social stability."

Relatedly, Megvii was placed on the U.S. Department of Commerce's entity list in 2019 for its alleged role in mass surveillance of Muslim minorities in Xinjiang.

Megvii's financials

Megvii's annual revenue in 2019 reached 1.26 billion RMB ($193.6 million), quadrupling its 2017 revenue of 300 million RMB ($46.7 million). Last year, during the first three quarters, Megvii's revenue exceeded 700 million RMB ($110 million).

City IoT is Megvii's main source of revenue, accounting for 64% of its total in 2020. Megvii has amassed over 400 customers for Smart City and Smart Community management and its technologies are used in over 100 Chinese cities.

Megvii used to rely heavily on its consumer IoT business, which involves providing facial recognition-based app/phone unlocking solutions, computational photography that allows smartphone users to take high-quality HD photos and videos and B2B identity verification products. But in 2020, this area only accounted for 18% of Megvii's revenue. Megvii's enterprise clients include Alibaba, Lenovo and Huawei.

Though Megvii has seen revenue growth, its losses have also been significant, just like other Chinese AI unicorns. The company lost 6.64 billion RMB ($1 billion) in 2019 and 2.85 billion RMB ($437.9 million) in the first three quarters of 2020.

Megvii's biggest expense, by far: R&D. In 2017, Megvii spent 66.5% of its operating revenue in R&D. The share of R&D investment increased year-over-year, reaching 82.2% in 2019. In the first quarter of 2020, this ratio reached 104.2%, slightly above the industry average of 98.5%.

What's next for Megvii

The proceeds will be mainly invested in — you guessed it — more R&D. Specifically, Megvii will pour more cash into developing and upgrading its visual IoT products, which involve innovations like turning cameras into analytical tools. It will also invest more in robotics, and sensor research and design. To that end, the company plans to scale up its production of AI-driven products and to expand its IoT applications to supply chain management.

Megvii will try to further commercialize its AI and deep-learning technologies, grounded in its three main business segments. The company says it will "be more aggressive" in exploring overseas markets and leveraging its existing scalable business model to expand its business globally.

What could go wrong?

Megvii could continue to fail to turn a profit. Like most AI unicorns, Megvii will continue to plow money into R&D and market expansion, even if that means losing money. It does not expect to pay cash dividends "in the short term."

Much of Megvii's growth traces to the Chinese government's AI-driven industrial policies and state investments in building smart cities across China. But the heavy reliance on government contracts brings uncertainty. If China's economy slows, or if adjustments in industrial policies occur, or if local governments' financial situations change, investments in sectors that favor Megvii will likely dwindle.

Geopolitics pose another risk. Megvii's prospectus is silent on its involvement in building a surveillance system in Xinjiang. Although the company claims that being placed on the U.S. Department of Commerce's entity list in 2019 has not affected its business, it nonetheless cautions that if trade disputes between China and the U.S. intensify, the entire supply chain of the IT industry will suffer.

Who gets rich?

Here's what we know from the prospectus:

• Founders Yin Qi, Tang Wenbin and Yang Mu currently own 16.83% of Megvii. Together they represent 70.28% of the voting rights.
• API (Hong Kong) Investment, which is wholly owned by an Ant Financial subsidiary, owns 15.08%.
• Taobao China owns 14.33%.
• Shenzhen National Bridge Investment and Guofeng Bridge Investment together own 11.33%. These two firms are subsidiaries of China Reform Holdings, a state-owned enterprise directly supervised by a special commission under the State Council, China's Cabinet.

GGV Capital and Innovation Works, beneficially owned by Peter Liu and Kai-Fu Lee, also have stakes in Megvii, as does the Industrial and Commercial Bank of China and China Harvest, an asset manager.

What people are saying

"AI companies generally face high R&D costs and difficulty in commercializing their technologies. With the huge losses and heightened privacy regulations, it is unclear whether Megvii can continue to tell a good AI story and prove the value of the industry's technologies." — Chinese tech blog Deep Dive atom.

"The commercialization of artificial intelligence has only started to scale in the last few years. Megvii started commercializing its technologies early. With years of experience in sales, it has deeper roots in various industries and understands the needs of the industries and customers better." — Equal Ocean analyst Yuan Shuai.