Pony.ai, the self-driving startup that operates in both the U.S. and China, has lost at least two executives from its trucking team following the combination of the company’s passenger car and self-driving truck R&D teams.
The development, initially detailed by two sources familiar with the internal restructure that happened quietly in September, was later confirmed by Pony. The Toyota-backed startup said the restructuring was made because combining the teams would lead to more efficient outcomes.
However, the move has called into question Pony’s plans for commercially developing autonomous trucking technology, specifically in the United States.
Frank (Zhenhao) Pan, one of the executives who recently resigned, was the chief technology officer for Pony's trucking business. The team he led included senior engineers of the autonomous truck driving team in the U.S., which has dissolved in the recent restructure. At least five members on that team, which previously had 20 staff, have left the company, a person familiar with the departures told TechCrunch. The rest were transferred to other business lines within Pony, although a number of them have also expressed dissatisfaction about the reorg, the person said.
Pan and another colleague, Youhan Sun, who used to lead planning and control for Pony’s trucking business in the U.S., left to start another self-driving truck company together, Qingtian Smart Trucking Technology. According to Tianyancha, a Chinese database that compiles business records, Zhenhao owns 82% of Qingtian and Youhan owns 18%.
Sun Haowen, former head of planning and control for Pony’s autonomous driving in China, also left following the merger, according to China’s LatePost. He has gone on to work at Qiangua Technology, a company that specializes in Level 4 autonomous trucking technology.
Pony.ai's thwarted IPO
Until August this year, Pony was gearing up to go public in New York with a $12 billion valuation through a SPAC merger. The company put those plans on hold as it struggled to gain assurances from Beijing that it wouldn’t become the next target of a crackdown against Chinese technology companies going public in the U.S., Reuters reported. Pony says it’s an American company with headquarters in Silicon Valley. However, much of its operations, including all of its self-driving truck testing, take place in China, a country that exerts considerable regulatory power over autonomous vehicle companies.
Pony does have some operations in the U.S., notably in California, where it tests passenger cars equipped with autonomous vehicle technology. The company has been testing its AVs on public roads in California since 2017. It was even providing a drivered passenger service in Irvine, California until service was suspended due to the pandemic. The service should return in 2022, the company says.
Pony’s SPAC deal would have given the startup an additional $1.2 billion to play with, but since the listing was reportedly blocked, Pony has had to think a little differently about its finances and the allocation of resources, according to LatePost. Pony declined to comment on its IPO, but a lack of resources could be the reason the two R&D teams were combined.
A short history of PonyTron
The company established a separate trucking division in 2020, which it branded PonyTron at the end of March. It was actively hiring for the division, Pony confirmed. The team had grown to about 100 people by September, according to LatePost. While PonyTron was growing fast, it still lagged in terms of total workforce compared to others in the self-driving trucks industry. TuSimple’s R&D team is about 1,000 people, for instance.
Pony, while demonstrating potential, was still lagging in terms of resources. Now recruitment has slowed nearly to a halt and road tests for trucks based in the U.S. have been suspended, TechCrunch's source confirmed.
Pony denied that it had intentionally slowed its hiring in trucking, saying it has enough staff and will continue to fill roles as needed. The company also said it has the same approximate headcount working on trucking and auto as before, with plans to continue to hire in trucking to support future growth.
Divide and conquer?
The current trucking staff on the newly combined R&D team are mainly junior engineers, since most of the senior ones left, according to a person with knowledge of the matter. The source said at least some of the senior engineers believed strongly that smart trucking and passenger trucks should be separate, not combined. Many experts in the industry agree it makes more sense to train different models for different use cases. Passenger cars usually go to market as robotaxis, which are often trained to drive in urban environments. Trucks, which carry freight between cities, are often trained to drive on highways.
Some companies, like Aurora and Waymo, have been known to follow a similar approach to Pony and combine forces, saying experiences learned by one driver can benefit others. Others tend to focus on one or the other because they say it's a faster and more direct route to market, which will allow for a better business and possibly room in the future to move into other markets. For example, Kodiak Robotics, TuSimple and Einride all work to solve the problem of highway driving, whereas companies like Motional, Cruise and Argo AI focus specifically on passenger cars and smaller delivery vehicles that are trained for urban autonomy.
"It all comes down to management because the technology behind them is similar,” a former Pony employee told TechCrunch. “From a purely technological point of view, passenger cars and trucks could be worked on together. But when a company reaches such a scale, I think there are considerable obstacles if the two are combined."
“If you mix the technology, who is responsible for the product at the end of the day?" said the person who asked not to be named.
Pony said both its robotaxi and trucking operations have always included urban and highway use cases and that the company remains equally focused on global robotaxi and trucking, adding that the reorganization has not changed its go-to market in China, at least.
“To increase our engineering efficiency -- and recognizing the many overlaps between virtual truck driver and virtual auto driver, in both coding and testing, especially in China -- we optimized our R&D processes,” a Pony spokesperson told TechCrunch. “We have the same approximate headcount working on trucking and auto as before, but the engineering process is now more fluid and integrated.”
The spokesperson went on to say Pony is continuing to invest in its trucking business, “and we have made significant progress on that front, which we hope to announce shortly.”
Pony is also still focused on scaling robotaxis into a commercial service beyond Irvine and into other parts of California. In May this year, Pony obtained a permit from the California Department of Motor Vehicles to test fully driverless AVs on public roads, and the company says it’s been testing both driverless and drivered AVs on public roads in the state. Meanwhile in China, Pony just got the OK to perform driverless testing on public roads in Beijing, something it's been doing in Guangzhou, as well, as of this year.
Update: The story was corrected to clarify that some members of Pony's U.S. trucking team are staying at the company.