China’s self-driving truck leaders say AI breakthroughs won’t accelerate rollout — here’s why

BEIJING – Although artificial intelligence updates make headlines every few weeks, these advances are not enough to get driverless vehicles on the roads faster.

That’s according to Chinese autonomous trucking companies, which say improvements in large language models from Anthropic’s Claude to China’s DeepSeek have little impact on vehicle deployment timelines.

“The best linguistics in the world [expert] It doesn’t mean he’s a good driver,” Pony.ai CEO James Peng told reporters last week. “Artificial intelligence is a very broad term. These are completely different things. Absolutely…irrelevant.”

“We all use different skills when we process language, when we play sports, when we drive,” he said.

Autonomous driving uses artificial intelligence to mimic the human driver with a combination of sensors, chips and algorithms. But the real-world training data needed is very different from the data that powers large language models, called world models, such as OpenAI’s ChatGPT.

Self-driving truck startup Inceptio is sticking to its timeline for a mid-2028 commercialization milestone, unaffected by broad advances in artificial intelligence, CEO Julian Ma told CNBC.

By the third or fourth quarter of 2028, it expects Inceptio to collect 5 billion kilometers (3.1 billion miles) of truck driving data in China; This is enough to allow fully autonomous heavy-duty trucks to be used on public roads.

Ma said that with 5 billion kilometers of driving data collected, artificial intelligence can increase this to 50 billion kilometers of experience in a world model, which is enough to allow the heavy-duty truck to drive completely on its own. He hopes the trucks could later start operating without people inside in certain parts of the country.

He stated that achieving this goal in about two years is already quite fast, and that driverless trucks will need partnerships with manufacturers and regulatory approvals, as well as technology, to become a widespread reality.

Autonomous vehicles rely heavily on data about driving on roads. Just like robotaxi companies, driverless truck operators are conducting manned tests to collect training data safely.

According to a report by ARK Invest, Inceptio has recorded the most commercial autonomous truck kilometers in the industry, surpassing its US competitors. Big Ideas 2026 report In January. At that time, the company had traveled 250 million miles; This was exponentially more than other Chinese autonomous driving companies. pony.aiIt ranked second with 4.2 million miles.

According to the report, US-based competitors Aurora, Kodiak and Gatik were in the top five with a total of 8.9 million miles.

Inceptio’s Ma said in late April that the company’s trucks had traveled 700 million kilometers (434.96 million miles) and that they were targeting 1 billion kilometers (621.4 million miles) by the end of the year. He said the company can use artificial intelligence to determine which specific scenarios to focus on to collect test data.

At the Beijing Auto Show, Pony.ai also announced an upgrade to its PonyWorld 2.0 AI model to improve its ability to collect certain data and train the model more efficiently. The company, which currently operates robotaxis in China and other countries, unveiled the fully driverless light-duty truck it developed with the battery giant. CATL.

Although China has 5-year development plans that increasingly emphasize technology goals, Ma said companies often take the lead in promoting innovation.

“We’re making this happen,” he said, before regulators saw the technology in action and were convinced enough to provide policy support.

But there’s clearly a long way to go before we see trucks and cars driving driverless across the country.

“Automobiles are actually the most challenging area for AI, and they overcome the challenge of embodied AI to some extent because it involves security,” Ma said. Embodied artificial intelligence includes humanoid robots.