Is China’s AI Bubble About to Burst?

To grow simply for the sake of growth is the ideology of a cancer cell. Today AI unicorns might want to follow the same ideology for the ultimate market dominance. Yet 2019 probably has remained a difficult year for Chinese AI fundraising, the vital fuel for the engine of startup growth, due to mounting trade tension and political complexity between the two biggest economies in the world. Many local private VC firms decided to tighten the belt for the coming months and Chinese AI startups already felt the pinch. Since the second quarter of 2019, it is said the AI funding activities have declined 45.5% and the total funding value amounted to less than 40% of the same period last year. Some AI startups told they are really having a tough time seeking capital. Their previous round of overvaluation has turned into a hot burning pan, scaring away new investors to take over and handle the heat upon facing a rather uncertain prospect. Few experts proclaimed the China AI bubble might start to show the sign of bursting.

To add insult to the injury, it is said 90% of the local AI startups are actually not making any money. So the lack of fresh cash should be a hard blow for keeping their business running properly in the near term.

The year 2012 China witnessed the very beginning of AI investment enthusiasm.  Major AI unicorns in China are established more or less around this time. After 7 years, not many AI startups can be considered leapfrog in technology advancement. For instance, in the overcrowded space of visual and voice recognition applications,  technical accuracy turns into a pointless comparison benchmark: startup A’s 90% accuracy should distinguish no further than startup B’s 80% accuracy. The competition all boils down to either “pricing” or “connection”.

Perhaps the only species that still can be bankrolled by powerful investors are those first-tier startups, led by so-called big four AI unicorns in the area of visual and facial recognition.

Down below the first rung on the ladder, the rest of the players have to live by the curse of “winner takes it all”. In fact, it is estimated that the big four AI unicorns alone have swallowed at least 1/5 of the total venture funding available. However, by scratching off their glittering veneer, critics pinpointed the long term profitability of the “big four” might be evaluated as precarious as well. In China, visual and facial recognition is mostly applied to security and surveillance projects which often found itself signing contracts with two types of clients: public sector or state-owned enterprises. For instance, about 70% of the revenues for Megvii come from “IoT smart city” projects. This implies the cycle of their account receivable is going to be very long, usually waiting a lengthy time to receive the due payment because of bureaucracy; so any mishap, such as the departure or the downfall of their key connection in the organization, would incur the high risk of putting the on-going project and pending payment into jeopardy.

Ironically some AI unicorns resort to the strategy of becoming a VC. Below is the example of Sensetime’s investment portfolio.

While their left hand is burning money on a daily basis, their right hand is also plowing part of venture money to feed other baby startups. Their pretext of “the current unicorn is incubating the next Gen AI startup or building next AI ecosystem” might be interpreted as a savvy way to inject extra substance into their roadshow story and support their gasping billion-dollar valuation.

To add one more wrinkle, recently US put 8 Chinese tech companies on the trade blacklist, including Sensetime and Megvii, preventing them from buying American hardware technology such as sensors, GPU, chips, etc on which they heavily rely for AI algorithm training. It does not mean the dead end for the Chinese AI unicorns, but it does mean they have to gain the capability of self-reliance or work on non-US alternative supplies, which is very likely to eat up their existing resources and capitals. Of course, the central government will step in and lavish abundant cash to foster China’s AI supremacy, but again the benefits usually tend to fall on the heads of few anointed unicorns, not on those second, third-tier startups. At the end of the day, it is a zero-sum game.

By: Cecilia Wu