China seeks its own brains to steer 30M autonomous cars
(Bloomberg) -- China’s aspiration to deploy 30 million autonomous vehicles within a decade is seeding a fledgling chip industry, with startups like Horizon Robotics Inc. emerging to build the brains behind those wheels.
The Beijing-based company is taking aim at Nvidia Corp. and Mobileye NV just as the autonomous-driving business takes off and uncertainty looms over international trade. Annual revenue from the chips used in driverless vehicles globally should more than double to $5 billion by 2021, according to Gartner Inc.
Horizon Robotics is an example of China’s resolve to move up the manufacturing value chain by focusing less on commodity smartphones and TVs, and more on sophisticated semiconductors and artificial intelligence that can help cars drive themselves or spaceships land on the moon. That industrial policy is meant to help China reduce its 1.75 trillion yuan ($276.4 billion) in annual chip imports, a value dwarfing its oil imports.
“China has to spare no efforts to pick up and develop its own chip technology to improve our own sense of security, especially when the U.S. government is making us fearful about any protectionism against China,” Wei Shaojun, director of the Beijing-based Institute of Microelectronics at Tsinghua University, said at a forum in Shanghai.
China’s push for self-reliance is a priority for President Xi Jinping’s administration, which set up a 200 billion-yuan fund for investments in homegrown chipmakers, Bloomberg News reported.
Overseers of the world’s largest car market -- and electric-vehicle market -- want to put Chinese-developed chips under those hoods and behind those dashboards. The government expects to have a manufacturing industry for parts such as sensors and embedded chips with a production value exceeding 100 billion yuan by 2020.
“Everyone is on the same starting line, thus creating an opportunity for China,” said Ding Wenwu, president of the state-backed China Integrated Circuit Industry Investment Fund Co.
Horizon Robotics was founded in 2015 by Yu Kai, the former head of Baidu Inc.’s artificial-intelligence business called the Institute of Deep Learning. Its investors include Intel Capital Corp., Shanghai-based Harvest Fund Management Co., Russian billionaire Yuri Milner and state-owned China Jianyin Investment Ltd.
The company completed a $100 million funding round in December.
Horizon’s smaller-than-a-postage stamp circuit board is called “Journey 1.0,” which was showcased inside a General Motors Co. GMC Yukon XL at the Consumer Electronics Show in Las Vegas in January. The processor can detect as many as 200 targets -- including pedestrians, vehicles and lane markings -- in real time and help the driverless car avoid collisions.
“Chinese companies have been evolving from the positions of made-in-China to created-in-China,” said Yu, 42. “And now we are heading to invented-in-China.”
The company is cooperating with Volkswagen AG’s Audi; Chongqing Changan Automobile Co., the local partner of Ford Motor Co.; and parts supplier Robert Bosch GmbH to get inside future autonomous vehicles. A self-driving Changan car completed a 1,200-mile road trip in China in 2016.
In an industry where the number of LEDs in a brake light is scrutinized for impact on gas mileage, processing data from laser, radar and camera sensors without draining an electric battery poses an enormous challenge for chipmakers.
“What we are seeking is a horse that can run fast and long but eats as little as possible,” Yu said. “If the embedded vehicle chip consumes too much power, it will lead to a burning car rather than a running car.”
Audi selected Horizon Robotics for a China-specific project and will evaluate the outcome before taking any next steps, Johanna Barth, a Beijing-based spokeswoman for Audi, said without elaborating.
Audi still uses Nvidia and Mobileye sensors in cars sold outside China, she said.
Nvidia, based in Santa Clara, California, has recorded revenue growth of more than 20 percent for the last seven quarters, partly because of increasing demand for autonomous vehicles. It teamed with Baidu to develop an autonomous-vehicle platform, with plans to install that in cars made by Chery Automobile Co.
“We are working with Chinese automakers and the Chinese ecosystem, as well as global automakers that sell vehicles in China,” said Danny Shapiro, senior director of automotive at Nvidia.
One of those customers is Tusimple Inc., a Beijing-based startup developing self-driving trucks. Tusimple, set to start commercial operation of truck fleets in Arizona and Shanghai next year, is backed by investors including Nvidia.
“We are satisfied with Nvidia’s chip, which has the strongest computing power for self-driving solutions,” said Hou Xiaodi, Tusimple’s chief technology officer.
Jerusalem-based Mobileye, bought by Intel last year for $15 billion, provides image sensing and processing technology for driver-assistance features. It wants to expand in China through its fourth-generation EyeQ chip and a mapping system called Road Experience Management, said Danielle Mann, a spokeswoman for Intel.
Mobileye will bring the mapping system to China via partnerships with SAIC Motor Corp. and Beijing-based digital map provider NavInfo Co.
Yet, China has the money and commitment to make itself a force in semiconductor development, just as it’s doing now with electric vehicles.
China’s industrial priorities include becoming a leader in advanced manufacturing businesses, including integrated circuits, 5G mobile communications, aircraft engines and new-energy vehicles, Premier Li Keqiang told the National People’s Congress on March 5.
The government envisions spending about $150 billion in the next decade.
“All major governments around the world are deploying all kinds of resources in policies, talents and capital to upgrade their industries,” Tsinghua University’s Wei said. “And China is no exception.”
--With assistance from Ian King, Gao Yuan and Ville Heiskanen