China's Momenta adds $500m to Series C round

Waymo and Tesla illustrate the main paths to fully autonomous transportation. Waymo focuses on the ultimate solution, building a robo-taxi fleet and accumulating 20 million miles on public roads by the end of 2019. Tesla’s approach is more gradual, using information collected from its electric vehicles (EVs) to train a partially autonomous system as a stepping stone to full autonomy.

Most China-based autonomous driving start-ups are taking the Waymo path based on level-four (L4) autonomy, where the car is fully autonomous in certain environments but still needs a driver in the seat. Momenta is different. In addition to focusing on L4, it provides L2 solutions – much like the Tesla offering, with the driver able to override – to mass-market carmakers.

“We walk on two legs,” explains Hardy Zhang, the company’s CFO. “The L2 and L4 solutions share data and a technology framework: mass-production data can be fed into the L4 algorithm; the L4 solution strengthens the security and reliability of L2 scenarios.”

With the timeline for generating profit from robo-taxi services unclear, the two legs strategy offers a swifter route to commercialization. And investors have once again demonstrated their faith in the Momenta approach by committing to a Series C extension of $500 million. Following the close of a first tranche of equal size in March, the entire round amounts to $1 billion.

The round was led by several strategic investors – including SAIC Motor, General Motors (GM), Toyota, and Robert Bosch – and Temasek Holdings and Yunfeng Capital. The likes of Mercedes-Benz AG, IDG Capital, GGV Capital, Shunwei Capital, Tencent Holdings, and Cathay Capital also took part.

Many of Momenta’s investors are also customers. SAIC, GM, Toyota, Mercedes-Benz, and Bosch are among the 10-plus global car manufacturers and component suppliers to have signed contracts with the company. GM’s participation underlines Momenta's technical capabilities, Zhang says, noting that the company also backs Cruise, one of the leading autonomous driving players in the US.

The L2 solution is scheduled to enter mass production next year, featuring in SAIC’s premium EV model under the Zhiji brand. On the L4 side, AVCJ understands that Momenta will launch tests through local ride-hailing platforms, also in collaboration with SAIC.

Momenta refers to the L4 product as "flywheel L4." The idea is that by creating an automated closed-loop comprising mass-production data and algorithms, the flywheel will turn ever faster, propelling the company towards large-scale implementation. However, this model only functions when the L2 and L4 solutions run off unified systems.

"The software interface must be consistent; the underlying platform architecture of the data must be consistent. Only in this way, the two sides – L2 and L4 – can benefit from each other," says Zhang.

In terms of hardware, LiDAR, or light detection and ranging, is optional for L2. But most of the sensors and cameras are the same for L2 and L4. Momenta advises original equipment manufacturers (OEMs) to install these devices in the same places within all vehicles, making it easier to introduce updates.

“When OEMs decide to upgrade from L2 to L4, they will prioritize our solution, because it has the least business interruption, which also means faster technology landing,” Zhang explains.

While the algorithm framework is the same for L2 and L4, the latter has a deeper understanding of different scenarios. For example, the L2 solution can be trained to identify different styles and shapes of traffic light. L4 goes a step further, addressing scenarios where traffic lights are suddenly cut off, colors are no longer visible, or large vehicles are obscuring the view.

There is a logic to focusing on L2, which can be implemented commercially, as well as L4, which isn't there yet. However, industry participants must be patient in moving from R&D to mass production. Zhang emphasizes the challenges involved in collaborating with OEMs, integrating different hardware and software platforms, and understanding the needs of engineers.

"We started to build this ability from an early stage. Our competitors may take a year or even two years to build the same capabilities," he maintains.

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

• Topics
• Greater China
• Technology
• Expansion
• China
• autonomous driving
• automobiles
• deep-tech
• Growth capital
• Temasek Holdings
• Yunfeng Capital
• GGV Capital
• IDG capital
• Shunwei Capital Partners
• Tencent
• Cathay Capital Private Equity