Forget the internet and social networks – the hotspot for start-ups and investors today is the future of transportation. Driving this shift is a series of cutting-edge technologies such as image recognition, advanced mapping, vehicle-to-infrastructure communications and even machine-learning operating systems.
Arguably the main beneficiary of this trend – at least in terms of industrial and investment interest – is autonomous driving. Once the stuff of science fiction, the idea of cars that drive themselves is on the cusp of reality and is reeling in capital from established auto players, tech and internet titans, as well as venture capitalists. According to think tank The Brookings Institution’s calculation, based on public sources only, more than $80 billion has been invested into self-driving cars from August 2014 to June 2017. Venture capital firm Comet Labs in May mapped start-ups working on the different pieces of autonomy and found 263 firms involved.
To be sure, most of these investment and entrepreneurial activities happened in the US, where experiments on automating cars have been conducted since at least the 1920s, and several states have now passed laws permitting autonomous vehicles. It is hotter than ever though – after years toiling away on a secret car project, Apple’s chief executive Tim Cook in June finally unveiled the its plans: “We’re focusing on autonomous systems”.
What's less well known is that much of this progress has been built on the efforts of the brightest minds from a country far away: China. Now, a growing number of scientists are quitting their US posts – in Silicon Valley or the nation's universities – and returning to start their own ventures in China. Investors have been quick to take note.
The talent war
Ask any investor in artificial intelligence projects and they will say talent is a major – if not the most important – asset they value. This is even more true when investing in autonomous driving start-ups.
Unlike, say, the e-commerce sector – which deploy data analysis and algorithms to match shoppers with items they’d like – self-driving cars need to have sensors and robotic algorithm that can capture and interpret the world at perception levels which are identical to those of human beings, and respond within a fraction of second.
That's why autonomous driving is the sector where AI technologies are most likely to prosper. In the words of Apple’s Cook in his Bloomberg TV interview, autonomous technology is “the mother of all AI projects”.
It follows that the expertise needed to develop and manage this kind of technology is scarce – and highly valued. On October 22, The New York Times reported that typical AI specialists – including PhDs fresh out of school and those with less education and just a few years of experience – could expect $300,000 to $500,000 a year in salary and stocks in Silicon Valley.
But even at that salary, tech powerhouses including Google, Uber and Tesla are losing top-notch self-driving engineers who start their own businesses or look to make fortunes by being in at the beginning of a start-up.
The picture is similar in China, where the development of driverless cars lags behind the US.
Among the less than a dozen Chinese start-ups working on autonomous driving technologies targeting the domestic market, a FinanceAsia study of publically-available information only showed almost every founding team has at least some veterans of established tech companies, such as China’s search engine operator Baidu.
The returnees
Take newcomer Pony.ai as an example. The fresh start-up, based in Silicon Valley and Beijing, was founded in December last year by Lou Tiancheng, a megastar programmer nicknamed “ACRush” or “Bishop Lou” by fellow coders.
Lou worked on the self-driving car team at Google parent Alphabet and Baidu's autonomous driving unit. He studied at Tsinghua University under Professor Andrew Yao – the only Chinese recipient of computer science's answer to the Nobel Prize, the Turing Award – and has since brought Yao in as chief advisor to the fledgling start-up. CEO James Peng was Baidu’s chief architect for autonomous driving before going into business with Lou.
Younger still at about seven months is Jingchi.ai, founded in China by Wang Jing, former chairman of Baidu’s technical strategy committee and general manager of autonomous driving unit, and Tony Han, the unit’s former chief scientist.
Roadstar.ai’s co-founder Zhou Guang was software architect on fusion 2D/3D sensor data for autonomous driving technologies at Baidu in the US. Its chief technical officer, Heng Liang, was tech lead on the sensing team of Baidu’s autonomous driving project, and earlier worked on Tesla’s Autopilot project – tycoon Elon Musk’s search for full driving autonomy by 2019. CEO Tong Xianqiao was a founding member on the mapping technologies unit at Apple’s driverless car initiative “Project Titan”.
At this early stage, with start-ups and investors still rooting through their commercialisation options, it's the strength of the founding teams that investors are betting on.
“I believe it’ll be Jingchi.ai that can make autonomous driving happen in China in the end, because I believe in Wang Jing and his team,” JP Gan, partner at Qiming Venture Partners, told FinanceAsia. Jingchi.ai in September announced a $52 million pre-A investment round from lead investor Qiming, strategic investor Nvidia GPU Ventures and a consortium including China Growth Capital. It has set a $100 million fundraising target for its A round, according to Chinese tech website Leiphone.com.
“Pony.ai has the best autonomous driving engineering team in China, which helps it to obtain the leading position in new business model exploration,” said Young Guo, a partner at IDG Capital. Zhou Kui, partner at Sequoia Capital China, added “the team has an established reputation on tech expertise and demonstrated highly-efficient execution capability, and we are excited to become a partner of Pony.ai in exploring such a promising industry”.
Sequoia and IDG in March invested an undisclosed amount in Pony.ai at a valuation of just short of $100 million. According to COO Harry Hu, the firm plans joint-venture partnership with tier-one auto suppliers, as well as becoming a transportation network operator in a push to become an “Uber for driverless cars” in the future.
Market, policies and giants
To a large extent, China is the right place to be for these ambitious AI scientists. It has a huge market and the right policy support. But can these newcomers defeat the likes of Baidu, which has been diving in this industry since at least 2014?
According to Boston Consulting Group’s projection, China will become the world’s biggest market for autonomous driving by 2035, reaching 5.2 million units of partial autonomous vehicles, and 3.4 million fully autonomous ones.
It has a more urgent need to revamp its transportation system than any other big nations given the complexity of its road network. The World Health Organisation last year estimated that more than 700 people were killed in road accidents across China every day. And nowhere has the inefficient nature of the road network been on display than in 2015's infamous “China National Highway 110 traffic jam”, where thousands of vehicles were stuck on 60 miles of road in a span of 10 days.
The State Council’s mega scheme “Made in China 2025”recognises that using fully autonomous smart vehicles could remove traffic jams, improve road efficiency, cut incident rates by 80% and basically eliminate road death, among other advantages.
Now some key features of China’s traffic situation that differ from the US – including an often mixed flow of pedestrians and vehicles, uneven roads, and traffic congestion – mean these scientists can’t simply copy mature autonomous driving technologies in Silicon Valley. And that's before you consider data and IP sensitivity.
At the top level, Beijing has laid out its “Smart City” plan, coherent to “Made in China 2025 initiative”, to develop driver-less car technologies, prompting some municipalities – such as Shanghai, Shenzhen, Chongqing, Anhui Province’s Anqing and Guangdong Province’s Huizhou – to roll out “pilot towns” where start-ups can conduct road testing.
One obstacle cited by many AI professionals is the absence of laws or bylaws regarding the use of application of self-driving technologies in China. But investors are confident the cost-saving principle, held by both the business world and politicians, will fence off regulatory uncertainties in the end.
The biggest concern for start-ups – as cited by one investor who has put money into a Chinese autonomous driving hopeful – is their ability to compete against giants. That includes Baidu, which is leading the self-driving efforts in China in terms of capital input, $50 billion car-sharing unicorn Didi Chuxing which just opened an AI lab in the US for autonomous driving development, and big auto makers.
Even less relevant names are joining the dance. Internet giant Tencent in August announced an alliance with a clutch of auto players including BAIC Group to ramp up AI technology development for autonomous driving, in the hope of filling the streets with self-driving vehicles by 2020. Alibaba has also teamed up with Shanghai-based SAIC Motor to develop internet-connected cars.
“When there’s a key breakthrough, be it on technology or government policy, big players with a deep pocket will always be in an advantageous position to lure talents and acquire market share,” the investor told FinanceAsia.
Na Xiaochuan, VP of strategy at Roadstar.ai, disagrees. “There are a lot of legacy issues with established industry giants in terms of technology infrastructure. They can only create something that’s adaptable to their existing infrastructure, something they’ve spent years to build up and can’t give up. That means they are unlikely to pioneer fully autonomous driving technologies, which need agility.”
Take Apple for example. It is building an the autonomous car operating system (OS) based on the iPhone/Apple Watch – which in theory tracks your location and health – as the original portable telematic device.
“The expertise in robotics and AI technologies required for vehicle autonomy pulls from a small pool of people in the world,” said Na.
That explains why money is following those people ... and it's going to keep on coming.
