Deal focus: Deeproute founder jumps at second chance

Roadstar.ai, one of China’s autonomous driving pioneers, went into liquidation in 2019, two years after its establishment and amid internal acrimony. But the story didn’t end there. One of the company’s co-founders held course and launched Deeproute.ai, also in 2019.

It recently entered the autonomous driving big league with a $300 million Series B led by Alibaba Group. Jeneration Capital also participated, while RZ Capital, Yunqi Partners, and Glory Ventures re-upped.

Back in 2017, Roadstar was the quintessential VC rockstar, quickly raising $10 million in seed funding from Yunqi, Glory, Shenzhen Green Pine Capital, and Ventech Capital China. This was followed by a $128 million Series A in 2018, which supplanted Pony.ai’s Series A as the largest-ever funding round in China’s autonomous driving industry.

What came next was shocking. Two of Roadstar’s co-founders, CEO Xianqiao Tong and CTO Liang Heng, dismissed the third, Chief Scientist Guang Zhou (pictured), in January 2019. They accused him of various misbehaviors, from hiding code to falsifying production data. Zhou denied the allegations, sued Tong and Heng for defamation, and the court found in his favor.

Roadstar’s investors declared a day later that the dismissal was invalid because they had not approved it. They called on the founders to resolve their differences. But the writing was on the wall. The investors stuck with Zhou when he moved on to Deeproute.

“I only focus on one thing – trying to do a really good job in autonomous driving. I don’t think too much about other issues. My thinking is very simple. Maybe because it’s simple, I can focus and keep moving forward," says Zhou.

Grace Liu, co-president of Fosun RZ Capital, tells AVCJ that she looked at Roadstar when it was founded, but didn’t pursue an investment due to discomfort with the shareholding structure. Deeproute posed no such problems, given Zhou is the sole founder and wields the most voting power. RZ Capital led a $50 million pre-Series A round in August 2019. Existing investors re-upped.

Focus is a virtue

Deeproute has caught up with the competition in autonomous driving – and won investor support in the process – because of its focused strategy, Zhou suggests. Two of China’s top-three robotaxi players, Pony and WeRide, have expanded into trucks, vans, and minibusses. Deeproute, the third, has only one product that transferred smoothly into the light truck or cargo van space.

“I don’t think there will be a winner-takes-all scenario in the autonomous driving industry. But as a start-up, you must concentrate your resources on your core products and establish a competitive advantage, not only at the algorithm level, but also at the application level. That tests the team’s problem-solving and execution ability,” says Liu.

Deeproute moved into light trucks because there is a 90% technology overlap with robotaxis. For example, the wheelbases are similar – 3.36 meters for a light truck and 2.48 meters for a robotaxi.

“We will have one team and we have built one technology system, which can be used in two scenarios," says Zhou. “Heavy trucks are very different from passenger cars in terms of size and control. The usage scenario is also different: heavy trucks run between cities on high-speed roads, while light trucks operate in cities. Light trucks are a subset of robotaxis.”

Deeproute’s light truck product line is likely to become the first to achieve profitability. Regulation is liberalizing faster than in the passenger vehicle space because the risk levels are lower when carrying cargo as opposed to people. Driverless light trucks might be first allowed to run at night, which means they can enter cities during light-traffic periods. Alibaba’s involvement as an investor – and ultimately a major customer – only magnifies this potential.

“Alibaba has the deepest roots in logistics and transportation among China’s internet giants. Whether it’s e-commerce delivery or intra-city physical store deliveries, Alibaba plays an important role, and we can help them,” says Zhou.

He doesn’t envisage robotaxis turning profitable until security drivers – who sit in the driver’s seat even though the car is fully autonomous in certain environments, also known as level-four autonomy (L4) – are removed. This won’t happen until 2024 or 2025. Profitability also depends on scale: a robotaxi platform must have 10,000 vehicles to break even, enough to serve a tier-one city.

RZ Capital‘s Liu is more optimistic, drawing comparisons between the consumer internet and autonomous driving business models. The former endured an extended period of losses before turning profitable, and the latter will do the same. She believes the tipping point will be when vehicles becoming fully remote-controlled, with one person overseeing multiple vehicles.

“The difference between autonomous driving companies and the internet platforms is that internet platforms must build to national level before achieving large-scale commercialization. Autonomous driving companies can deliver unmanned driving in a specific area of a certain city, realize profits in that area, and establish a model or a benchmark for the entire industry," Liu explains.

She adds that the technology gap between the US and China on autonomous driving is not large, and China can implement comprehensive industry-level policies to drive growth.

Points of differentiation

Deeproute has about 80 robotaxis with an aggregate traveling distance of about 2 million kilometers. The company plans to have 150 by year-end, of which about 100 will be self-operated. The team will also expand from 400 to 600, including safety drivers.

Zhou claims that Deeproute’s capital expenditure efficiency is 2-3x higher than its competitors. The company has spent $50 million to date, rising to $70 million if the investors’ initial commitment to Roadstar is included. The likes of Pony and WeRide have raised hundreds of millions of dollars. One reason for the discrepancy is Deeproute opting to run tests in central business districts (CBDs).

“We are the only robotaxi provider testing in CBDs in first-tier cities. Each mile provides higher value than in the suburbs because we are dealing with more scenarios in these high traffic areas. Waymo ran more than 10 million km in Phoenix, but it might get less data than Cruise running 2 million km in San Francisco,” Zhou explains.

Another differentiator is Deeproute’s commercialization-oriented design. The company’s equipment is small and energy-efficient – to the point that it doesn’t take up any trunk space – and low power consumption means longer recharge mileage. An electric vehicle with a 320 km range can travel about 300 km using Deeproute, compared to 200 km with other autonomous driving solutions.

The who-can-commercialize-L4-first debate globally is positioned as Tesla versus Waymo, or shadow versus non-shadow.

Tesla’s shadow mode involves installing autonomous driving systems in vehicles driven by humans. They take no action, simply registering when they would have taken action. This facilitates data accumulation and allows the company to upgrade software on a continuous basis. The counter-argument is that Tesla’s vehicles are not equipped with LiDAR, so the data isn’t as relevant.

Zhou makes another observation: the biggest difference between L2 and L4 is not equipment – L2 vehicles can use LiDAR if it is cheap enough – but design architecture.

“L2 can compromise safety for comfort, while L4 cannot compromise on safety. This means the design concept can be totally different. When an L2 vehicle can’t identify an object, it continues to drive because it’s not responsible for the accident [under L2 the driver is still responsible]. If an L4 vehicle feels uncertain, it must stop,” says Zhou.

This explains the series of accidents involving Tesla’s Autopilot system, as well as more recent accidents to befall China-based Nio.

• Tweet  
• Facebook  
• LinkedIn  
• Google plus  
• Save this article  
• Send to  

• Topics
• Greater China
• Expansion
• Technology
• automobiles
• autonomous driving
• Alibaba Group
• Yunqi Partners
• Glory Ventures
• Jeneration Capital
• Growth capital
• Fosun RZ Capital
• China