The dilemma for AI startups in China

In China, AI industrial chain usually forms at 3 layers:

 

In essence, almost all the AI related startups fall into these 3 categories, focusing either on infrastructure building, pure technology development or wide commercial application. Today we see far less China startups in the infrastructure R&D, whereas most of them cluster around the commercial applications.

For China AI startups, four core factors should decide their ultimate fate.

  • Market: 

The size has still been growing. By 2020, China AI market is forecasting to be USD 10.6 billion with a compound annual growth rate at 44.5%.

  • Capital:

The investment market is blowing tailwind and the trend looks positive, albeit with a sluggish sign.

 

  • Talent:

So far the statistics suggest that China AI talent has been much lagging behind the US. That is why it is said the Chinese government has been spending big money in order to lure away Chinese background AI experts from the US.

  • Technology:

Now leaves the discussion of the technological supremacy of AI startups. Three key metrics are usually brought into the evaluation table. However, Algorithm is the most talked about aspect when startups try to boast about their competitive advantage. According to my interview with local AI specialists, Algorithm is not necessarily a winning formula. Though a superior algorithm generates better accuracy, 90% accuracy is not drastically distinguished with 80% or 70% accuracy as for every 10% accuracy improvement, you would still need human intervention to remove errors which often does not result in substantial labor cost reduction.

Finally, many AI startups in China face one dilemma. As mentioned before, most of the local startups prefer to profiting from huge potentials of commercial application in diverse industries. One fundamental question they often ask themselves is:

Should our AI products serve corporate clients or regular consumers?

Take the robot as an example. The To C market China has witnessed the mushrooming of robots roaming around the hotel, restaurant, banks, library, shopping mall etc (see below demo from a Chinese startup).

 

 

However, many consumers consider the whole experience is lackluster or sometimes unnecessary. Even in Japan, it is said one robot hotel decided to lay off half the robots after they created more work for humans (see link)

So some startups are more inclined to put robots into B2B manufacturing process. A company like WestWell-Lab is using robots to assist in the port area navigation or mining operation process. The CTO told at this stage they have no intention to move the robot application into the consumer side as most of the user experiences remain to be unsatisfactory. He is fully convinced to target at heavy industrial applications.

 

In many ways, the answer should arrive with different Pros and Cons, depending on startups market positioning, pricing model, entry strategy and sales capability.

 

By: Cecilia Wu