The video streaming service of Chinese online search giant Baidu is being taken public in the United States. Known as iQiyi, the video streaming service intends to raise a minimum of $1.5 billion in the IPO which was filed earlier this week. The ticker symbol of the service will be IQ.
According to Karen Chan, an analyst at Jefferies, iQiyi has a subscriber base of around 50 million and a market share of 28% in China. iQiyi generates revenues not only from subscriptions but also from selling advertising.
“Online advertising grew 44% in 2017 to $1.3 billion (47% of total revenue), largely on par with industry average. This is largely driven by 16.3% year-over-year increase in average brand ad revenue per advertiser and in-feed ad service launched in 2016,” wrote Chan in a client note.
Not yet profitable
At the moment about a fifth of Baidu’s revenue is generated by iQiyi. Compared to its U.S. equivalent, Google, that’s a little higher. Though Google doesn’t release separate financials for YouTube estimates from brokerage firm William Blair indicate that in 2018 revenue from the video sharing site will be $18.4 billion which translates to around 14% of Alphabet’s revenues.
According to William Blair iQiyi is yet to become profitable while YouTube recorded its first annual profit four years ago. iQiyi has also witnessed cash burn as a result of acquiring licensed content. Last year iQiyi generated revenues of $2.54 billion and this was a year-over-year increase of 54%.
The on-demand video content of iQiyi is like that of Netflix and offers movie and television shows. Like its American equivalent Netflix, iQiyi also employs artificial intelligence tools with a view to making programming recommendations to its subscribers. After the IPO Baidu will still own the majority stake in iQiyi. Currently Baidu’s stake in the video streaming service is 69.6%.
Baidu’s decision to publicly list iQiyi comes in the wake of the online search giant’s researchers an AI tool based on deep learning which can mimic voices after listening for under a minute. According to the Baidu’s Leo Zou it would previously have required numerous examples before a model could learn but this has now been reduced to a fraction of that.
Per Zou some of the uses of voice cloning would be assisting patients who had lost their voices. The technology could also be applied in speech-to-speech language translation as well as in the development of digital content such as video games and audio books.