Two years ago, Xiaodi Hou was ousted from his position as CEO of TuSimple, an autonomous trucking startup he co-founded that has since shut down U.S. operations and delisted from the stock exchange — and, in an odd twist, the company has pivoted to AI animation and gaming in China.
Now Hou is back with a new Houston-based self-driving truck startup called Bot Auto that recently made its public debut armed with $20 million in pre-Series A funds. In a way, Hou is getting the AV band back together; the majority of his 40-person team consists of former TuSimple engineers who had been laid off in several rounds of cuts that hit the U.S. team.
While the mission and much of the team are the same, Hou claims the underlying technical architecture is different. The founder says Bot Auto has entered the second wave of autonomous vehicle technology, where the last couple of years of AI advancements have made it easier and cheaper to accelerate AV development.
“We basically developed a machine learning infrastructure with eight people in four months,” Hou told TechCrunch. “In my last life, that probably would have taken more than 100 people and a year to develop.”
AV 2.0 era
Hou isn’t the only founder talking about the AV 2.0 era, which is seeing renewed investor interest despite the numerous self-driving startups that have fallen.
TechCrunch recently reported on autonomous delivery startup Nuro’s shift toward a more streamlined AI architecture. Autonomous trucking startup Waabi, which raised $200 million this year, has also touted a more advanced AI model. And Wayve, which raised $1.05 billion this year, is developing a self-learning system.
One of the AI advancements that has helped AV companies in this second wave, including Bot Auto, is transformers — the same architecture behind large language models like OpenAI’s ChatGPT. Hou said transformers allow Bot Auto’s gigantic neural network to take in data from different sensors — like cameras, radar, and lidar — to create a “world model” of the vehicle’s surroundings and simultaneously recognize objects and traffic signals and plan pathways.
The neural network, which Hou refers to as a “foundation-to-all” model, stops short of decision-making. That’s the kind of data-in-decision-out flow that happens in an end-to-end model, which Hou says results in the black box effect. Instead, the decision-making is done via a “principle-based system,” which Hou says leads to “explainable AI.”
Bot Auto can also speed up the process of data collection and processing through pre-training, according to Hou, which involves training neural networks on large datasets before they’re fine-tuned for specific tasks or environments.
“Pre-training is the key to fast-forward training and jumpstarting deep learning endeavors without needing to accumulate data for years and label it all by hand,” Hou said. He noted that boosts in supervised self-learning have allowed Bot Auto to pre-train models from unlabeled data without human feedback, which speeds up the process significantly.
Hou said Bot Auto also doesn’t have to think about how to orchestrate GPUs or do high-performance data storage because he can tap open source tools that weren’t available when he headed TuSimple.
For example, Bot Auto uses Kubernetes as its cloud operating system to automate almost all critical data-related tasks like large-scale simulation and mining road test data, according to Hou.
Focusing on operations
Taking the brunt work out of building the AV infrastructure means Bot Auto can focus more on the core business of bringing AVs to market, according to Hou.
“Of course we are always trying to stay abreast of the best technology, but we are not here to show off the technology,” Hou said. “We are here to facilitate operations, to make it happen. That’s why the operations are in Houston.”
Texas has become the leader in autonomous trucking, and not just because the state’s sunny skies make it easier to test there. Texas is home to some of the most lucrative freight routes in the country, including the I-45 corridor between Houston and Dallas. Both Waabi and Kodiak Robotics are testing trucks on these routes daily.
And now so is Bot Auto.
Hou said his startup has begun testing autonomous trucks on public roads in Texas with a human safety operator in the front seat and has already secured two letters of intent from customers in the logistics and shipping industry. Bot Auto is also on the right track to complete a hub-to-hub demo in 2025, Hou said.
Despite the scalability of its foundational models, Bot Auto is being careful when it comes to scaling up operations on the ground.
“We are going to contain ourselves in a very cost-effective way,” Hou said, noting that investors in 2024 are only investing in real technology that can be sustainable. “We have a very small operating fleet. We focus on a small number of lanes, and then only after we know how to save money and operate efficiently, then we’ll start to scale up.”
Not without its challenges
Despite its wins, Bot Auto will likely face some headwinds.
Investors may be brightening to the idea of a new era of AV, but they haven’t forgotten the trials of the past couple years, which includes the shutdown of Embark Trucks and Waymo’s autonomous freight business.
And, of course, there’s the failure of TuSimple, the worst of which happened after Hou had been ousted as its leader.
TuSimple’s board had voted to oust Hou after an internal investigation revealed some overlap between the company and Chinese startup Hydron, which Hou’s co-founder Mo Chen founded to produce autonomous-ready hydrogen trucks in 2021 around the time TuSimple went public in the U.S.
Such a public firing, and one made spicier by geopolitical tensions, might also give investors pause, although Hou has maintained he was let go without cause — a claim that sources at the company have backed up to TechCrunch in the past. He also said that experience taught him not to try to straddle the lines of unfriendly countries, which is one of the reasons why Bot Auto is a fully U.S.-based company with no ties to China whatsoever, according to Hou.
When asked why Hou didn’t launch Bot Auto in China instead, where regulations are more favorable for AV companies, he said the higher cost of labor for truck drivers in the U.S. increases the potential value for autonomous driving.
“It was a business decision,” he said.
Then there’s the threat of legal action from TuSimple itself. After being fired, Hou maintained his board seat at TuSimple but ended up resigning in March 2023 after the board accused him of trying to poach staff for his new venture. Hou denied this at the time and told TechCrunch he only went public with his decision to start a new company after he left the board. He said the TuSimple alum who make up his staff today came to him out of loyalty and a desire to keep working on AVs after being laid off.
Whether those accusations will resurface is unclear. Although Cheng Lu, TuSimple’s CEO, told TechCrunch the company would likely sue Hou for alleged IP and trade secrets theft.
Hou said he’s not worried.
“Everyone sues everyone here,” he said. “And I think after telling you about the technologies we’re building, most of the things we’re doing weren’t even happening before 2021.”