Autonomous trucking company TuSimple used its second-quarter earnings call to address an April crash during which one of the company’s autonomous trucks suddenly veered across the I-10 highway in Tuscon, slamming into a concrete barricade.
The crash first came to light via a YouTube video that showed footage of the crash along with a letter from the Federal Motor Carrier Safety Administration (FMCSA), dated May 26, alerting TuSimple to a “safety compliance investigation.” The accident was later reported on by The Wall Street Journal.
“An error occurred when a test driver and safety engineer tried to reenter autonomous driving mode before the system computer was primed to do so, and the truck swerved, making contact with the highway barrier,” said Xiaodi Hou, TuSimple co-founder and CEO, during Tuesday’s earnings call. “No one was hurt. And the only evidence of the accident are a few scrapes and some minor damages on our truck.”
Hou noted that in the past seven years, TuSimple had accumulated 8.1 million miles of road testing with “precisely one incident.” When the incident happened on April 6, TuSimple grounded the entire fleet and began an independent investigation, said Hou. After determining the cause of the error, the company then upgraded all of its systems with an overhaul of its human machine interface to make sure the same problem would never happen again, the executive continued.
That internal report, which was reviewed by WSJ, revealed that the truck abruptly swerved left due to an outdated command, which was 2.5 minutes old and should have been erased from the system but wasn’t.
Researchers at Carnegie Mellon University told WSJ that common safeguards, had they been in place, would have prevented the crash. For example, the truck shouldn’t be responding to a command that’s even a couple hundredths of a second old, let alone more than two minutes old. The system also shouldn't be able to turn so sharply while traveling at 65 miles per hour, nor should a safety driver be able to engage a self-driving system that’s not properly functioning.
The National Highway Traffic Safety Administration has since joined the FMCSA’s investigation into the TuSimple highway crash.
Hou said that the two agencies have yet to find any anomalies in their investigation or give TuSimple any safety recommendations, but the investigation is not yet complete.
During the earnings call, TuSimple repeated its plans to commercialize driver-out operations, in which no human safety operator is present in the vehicle. The company first completed a driver-out demonstration along an 80-mile stretch in Arizona in December, and has completed several more runs since.
TuSimple said the crash wouldn’t affect its plans to begin driver-out operations for Union Pacific Railroad, but it's unclear if the company is even on schedule for that at present. TuSimple was meant to launch fully autonomous freight hauling for Union Pacific in the spring of this year and scale to commercial viability by the end of 2023, but Hou said the company has encountered a complete road closure in front of the distribution center at its destination point, which has delayed the run by “a couple of weeks.” He also reiterated that the company’s deadline for driver-out in Texas is set for 2023, but did not specify if those will be initial test runs or fully commercial operations. TuSimple did not respond in time for requests for clarification.
TuSimple Q2 financials
TuSimple’s total revenue was $2.6 million in the second quarter, which is up 73% year-over-year and 13% sequentially. Wall Street analysts expected TuSimple’s revenues to come in at $4.06 million; moreover, they expected the company to beat those estimates.
The company attributed its growth, such as it was, to increased utilization of existing assets and year-over-year price increases.
TuSimple’s net loss came in at $108.6 million, versus $116.5 million in the same quarter of last year. The company appears to have slimmed down on total operating expenses, which came in at $107.5 million this quarter versus $119.4 million last year. However, R&D spending was up 13% year-over-year at $85.5 million. TuSimple said the largest portion of R&D expense was $60.8 million related to hiring, including a stock-based compensation expense of $22.4 million. That said, sales, general and administrative spend was significantly lower than last year.
To prepare for driver-out operations and to expand its autonomous freight network, TuSimple invested a total of $3.8 million in purchases of property and equipment. The company ended the quarter with $1.16 billion in cash.
Updated full-year guidance
TuSimple’s updated guidance on 2022 revenue remained unchanged at $9 million to $11 million. Generally, the company intends to spend less, and therefore lose less money this year. TuSimple’s adjusted EBITDA loss for the year is now expected to be between $360 million and $380 million, versus previous guidance of $400 million to $420 million.
In addition, TuSimple will spend less on stock-based compensation -- attributable to a hiring slowdown -- as well as purchases of property and equipment. The company is hoping to end the year with $950 million in cash versus previous guidance of $900 million.
Hou touched on some key leadership changes that had been announced in June, including chief financial officer Patrick Dillon leaving the company, to be temporarily replaced by Eric Tapia, TuSimple’s global controller and principal accounting officer.
In addition, Dr. Ersin Yumer, previously head of TuSimple’s autonomous freight network, was promoted to EVP of operations, and Dr. Lei Wang was promoted to EVP of technology. Both were promoted to support TuSimple’s driver-out operations.
It’s worth noting that TuSimple would not address a question about the company’s tentative plans to sell off its China operations, something that was touched on during the first-quarter earnings call.
At the time, TuSimple told TechCrunch that the company’s stock price today doesn’t reflect the value of the China autonomous freight business, so it would be a good idea to split off APAC operations. A perusal through the company’s 10-Q revealed TuSimple is more likely looking to sell its China operations because it’s too expensive to keep it going, given the National Security Agreement the company agreed to as part of a review by the Committee on Foreign Investment in the United States.
Tapia, TuSimple’s interim CFO, also shared that the company is in the process of upgrading most of its older trucks to its newest AV hardware technology, a process which will continue through 2023 and will involve adding upgraded sensors to the vehicles.
“While we plan to introduce some new trucks into the fleet, our ability to add a significant number of trucks is difficult, given the challenges in purchasing new or even slightly used trucks,” said Tapia. “Lastly, we plan to continue to invest in adding terminals to the [autonomous freight network], primarily around the Texas triangle. Our intention is to do this in a capital-light manner, partnering when possible.”
In 2020, TuSimple partnered with Navistar to build fully autonomous trucks, and has previously set a deadline to begin manufacturing by 2024 and deliver to certain customers, like DHL, by 2025. Hou and Tapia dodged one analyst’s repeated attempts to get clarity on this timeline.