China lags significantly behind global leaders. The same is true in newly emerging fields like self-driving cars.
One major obstacle to Chinese self-driving cars is conservative regulation. Chinese career officials cannot tolerate accidents like the fatal Uber crash in Arizona in March 2018 which would surely destroy their political prospects. Therefore, self-driving regulations are highly risk averse in China.
Since December 2018, American self-driving company Waymo has been offering robotaxi services to riders in Phoenix, Arizona; the cars are autonomous but currently retain a human driver for safety reasons, as do almost all other autonomous cars. Today, no Chinese companies are allowed to do this.
An attempt in November 2018 by Chinese self-driving startup WeRide.ai to launch such a service was halted the next day by local regulators citing safety concerns. Another Chinese startup, Pony.ai, is internally testing robotaxi services outside of Guangzhou, waiting for official approval.
Chinese companies also face much higher and more costly requirements than their American counterparts when obtaining a self-driving testing license. In Beijing, for example, autonomous driving vehicles are required to have completed over 5,000 km of driving at approved enclosed testing grounds; companies need to book such grounds far in advance and demand is disproportionate to supply.
In California, by contrast, vehicles are not required to have driven a certain number of miles in enclosed testing grounds, and cars are issued permits as long as they demonstrate the required skills for road testing
More importantly, Chinese self-driving car companies have only driven a fraction of the test miles that American companies have, and road tests are the key to improving self-driving systems.
In the decade to 2018, Waymo’s self-driving cars had driven more than 10 million miles on public roads. By contrast, Beijing city released data showing that eight companies have together conducted road tests of 154,000 kilometers (95,000 miles) in 2018.
Another indicator for measuring the quality of self-driving systems is how often human intervention is required. Though many argue the frequency of human intervention depends on road and traffic conditions, this data can still serve as a rough but useful reference for how reliable self-driving systems are.
Waymo’s autonomous vehicles experienced one intervention by a human driver for every 5,596 miles clocked by its self-driving systems in 2017, according to the most recent figures available. In comparison, Baidu’s cars required one intervention for every 41 miles driven, according to California’s Department of Motor Vehicles.
Beijing is the capital of self-driving cars in China as the earliest city to allow testing, allocating the most roads (by distance) for testing and issuing the highest number of licenses. The city counts 56 self-driving vehicles being testing on roads, which is about half of all registered test vehicles in the country.
In comparison, Waymo alone is believed to operate at least 600 vehicles in its self-driving fleet. Having the ability to operate a large fleet is necessary for self-driving cars’ commercialization.
“I feel that the past three years have been a low point for the Chinese autonomous driving sector,” one industry veteran told me. “While Waymo and others were conducting massive amounts of road tests, Chinese companies were busy staging news conferences.”
“They were announcing partnerships and demos in which cars were moving slower than humans… I feel it is a problem of culture… They are not focused on the technology.”
Chinese companies also lag behind in autonomous driving patents. In Nikkei’s 2018 ranking of patent-based competitiveness, no Chinese companies were ranked in the Top 50. Waymo was in the lead, followed by Toyota Motor, GM, Ford and other primarily North American, European and Japanese companies.
On a more fundamental level, the U.S. is unshakably the international innovation hub: it remains where global talent gathers to develop frontier-pushing technology. While half of California’s permits are issued to non-U.S. companies, 90% of China’s permits are issued to domestic players. The remaining 10% are issued to multinationals who want to gain access to the future Chinese transportation market.
Most of the founders of Chinese self-driving companies worked in companies like Google and Apple, or have been educated in the U.S. Their business model echoes that of Waymo. To some extent, the “copy-to-China” model is still the mainstay of the Chinese tech scene.
Similar vulnerabilities exist in China’s tech sectors from robotics to semiconductors. China’s tech capabilities have been greatly hyped by entrepreneurs, public companies, Chinese media and government officials. Such rhetoric has resulted in misinformed policy-making in Washington D.C., while Beijing realizes that its tactic to bolster national pride has backfired.
U.S. fears are partly driven by misinformation, including media hype about Chinese tech and AI prowess – its position as a global tech leader and pioneer is not yet under any immediate threat from China.
China’s tech development is still in an early stage. It’s only been two decades since the days when Chinese entrepreneurs copied everything in Silicon Valley during the dot-com era. The gap between China and the U.S. in tech prowess, in America’s favor, is astronomical. China’s success has been massive application and business model innovations.
These are the things easy to learn. The secret sauce of innovation – education, a free and open academic and business environment – will take a long time to master, and will require China to let the market sit in the driving seat, with less government intervention and fewer political campaigns about industrial development.
China should recognize its weaknesses. Beijing needs to hold to its heart Deng Xiaoping’s advice always to keep a low profile. When it comes to technology, China should remain a modest student – and the U.S. should better assess its rival.