Autonomous Trucking 2025: What Actually Happened

Eighty thousand pounds. Seventy miles per hour. Zero humans. On May 1, 2025, that combination became reality. An Aurora-powered Peterbilt merged onto I-45 outside Dallas, hauling FedEx freight to Houston with no one at the wheel. For the first time ever, a fully autonomous Class 8 truck moved commercial freight on public roads.

Twenty-six days later, The Wall Street Journal revealed that TuSimple, the company that promised to get there first, had secretly transferred its core technology to a Beijing-connected manufacturer after signing a national security agreement prohibiting exactly that.

Welcome to 2025 — the year autonomous trucking arrived, and the year that proved technology was the easy part.

The Milestone

Aurora’s achievement is real. By Q3, they’d logged more than 100,000 autonomous miles with zero safety incidents. Two corridors, Dallas-Houston and Fort Worth-El Paso, now are operational. Night driving works. The technology does what it’s supposed to do.

But complications emerged fast. Two weeks after the launch, OEM partner PACCAR asked Aurora to move their human observer from the sleeper cab to the driver’s seat. A short-seller report dropped, claiming PACCAR executives said the tech “wasn’t there yet.” Aurora’s stock fell 25%. Their chief product officer left for GM.

The financials tell the real story: $1 million in Q3 revenue against $222 million in losses. Aurora has $1.6 billion in cash, enough runway into late 2027. But the path from five trucks to 500? That’s where every spreadsheet gets uncomfortable.

The Competition

Aurora isn’t alone. Kodiak went public in September and is running fully driverless trucks in the Permian Basin, hauling frac sand for Atlas Energy, 24/7, on private oilfield roads. Real revenue is $16.4 million trailing 12 months. The catch? Private roads aren’t public highways. Their on-highway operations still use safety drivers.

Gatik signed a $50 million deal with Loblaw for middle-mile delivery in Canada. Torc, a Daimler subsidiary, is targeting 2027 for its commercial launch with factory-integrated Freightliner Cascadias. Waabi is betting that simulation can replace real-world testing.

Not everyone made it. Embark shut down. Waymo Via wound down. TuSimple imploded spectacularly. The field has consolidated. More consolidation is coming.

The Technology

2025 saw genuine breakthroughs, not just incremental progress.

• NVIDIA DRIVE Thor became the industry standard. Aurora, Kodiak, Gatik, Torc, Waabi, and Plus all adopted the 1,000 TOPS platform. The platform wars are over.
• 4D LiDAR went commercial. Aeva and Innoviz sensors now measure distance and velocity simultaneously, telling trucks not just where objects are but how fast they’re moving.
• LiDAR costs collapsed from $75,000 per unit in 2019 to under $1,000 today. That changes the economics of everything.
• End-to-end neural networks replaced hand-coded rules. The industry calls it “AV 2.0,” the same foundation model approach behind ChatGPT, now applied to 80,000-lbs. trucks.
• V2X communications expanded. Trucks now can receive data about hazards around blind corners, upcoming signals, and emergency vehicles beyond sensor range.
• Cybersecurity architectures matured. With GPS spoofing incidents hitting 1,000 daily worldwide, multi-layered sensor fusion and Zero Trust frameworks became standard requirements in shipper contracts.

The China Problem

Here’s a number that should keep American executives awake: 620,000 miles per day. That’s what Inceptio Technology’s trucks log on Chinese roads daily. Aurora’s total accumulated miles through Q3? About 100,000. China logs six times that every single day.

Chinese companies operate 2,000+ trucks equipped with autonomous systems. They’ve pioneered the 1+4 model, one human-driven lead truck guiding four driverless followers. They control two-thirds of global LiDAR production. McKinsey projects China’s autonomous trucking market at $327 billion by 2035 and the U.S. market at $178 billion.

The TuSimple scandal casts a long shadow. But the technology already crossed the Pacific. China is scaling it faster than anyone in the West.

The Labor Fight

On June 9, the Teamsters secured contract language at Albertsons requiring a human operator, not just a monitor, in any autonomous truck. That’s the template they want everywhere.

California’s AB 33, which would mandate human operators in commercial AVs, passed the Assembly before stalling in the Senate. A January poll found 79% of California voters support human operator requirements. The Teamsters are pushing similar legislation in Nevada, Texas, Delaware, and Massachusetts.

The trucking industry argues that there’s an 80,000-driver shortage. Labor argues that 94% of truck accidents involve human error, but the humans are usually other drivers, not professional truckers. This fight is just beginning.

The California Question

California has banned autonomous trucks over 10,000 lbs. since 2019. That may change. The DMV released proposed regulations on April 25 and revised rules on December 3, with the comment period closing December 18. If adopted, testing permits could be available by Q2 2026.

This matters enormously. The Ports of LA and Long Beach handle 40% of U.S. containerized imports. Without California, there’s no coast-to-coast autonomous network. Every autonomous trucking company needs California. Whether California needs them is another question.

The Winter Barrier

Here’s what nobody talks about: every commercial autonomous trucking operation runs in the Sun Belt, Texas, Arizona, or the Southwest. There’s a reason for that.

Snow blinds LiDAR. Ice breaks braking models. Fog defeats cameras. Lane markings disappear under six inches of snow. While the technology solutions exist — sensor heating, enhanced cleaning, better algorithms —none have been validated at commercial scale.

The Chicago-to-New York lane is one of the highest-volume freight corridors in North America, a $200+ billion market. It’s effectively off-limits for driverless trucks. The I-70 corridor through the Midwest has a winter pilot planned, but that’s testing, not commercial operations.

For fleets with northern exposure, the practical reality is hybrid operations for years to come —autonomous where weather permits, conventional where it doesn’t.

What to Watch in 2026

Here are some things to look for in 2026.

• California’s regulatory decision after December 18.
• Aurora’s second-gen hardware deployment in Q2.
• Torc’s progress toward 2027 commercial launch.
• Whether Kodiak can transition from oilfields to highways.
• How the Teamsters’ contract template spreads.
• And the gap between U.S. and Chinese deployment: does it narrow, or keep widening?

The Verdict

2025 answered one question definitively — the technology works. Autonomous trucks can drive highways, deliver freight, and operate commercially without human intervention. Aurora proved it.

The harder questions remain. Can anyone make money doing this? Will regulators allow expansion? Will labor accept displacement? Can American companies compete with China?

The trucks drove themselves in 2025. Everything else is still up for grabs.