Why China’s Autonomous Vehicle Push Matters — Part 2

While China is making strides with its autonomous vehicle (AV) development, the U.S. is also making investments.

Here’s where the United States stands as of May 2025.

• Waymo: Waymo is the only U.S. company running a fully driverless, paid robotaxi service at scale, carrying roughly 250,000 driver-only passenger trips weekly in Phoenix, San Francisco, Los Angeles, and Austin, with launches in Atlanta and Miami slated for later this year. (See Figure 1.) It operates a fleet of ~700 autonomous cars (Jaguar I-Pace EVs and others). In 2023 to 2024, Waymo rapidly expanded. By the end of 2024, it had provided 5+ million rides total, including 4 million driverless rides in 2024 alone. Notably, Waymo cars have collectively driven 50 million+ miles fully autonomously (with no human driver) on public roads as of December 2024. This is a testament to its AI stack maturity. Waymo has announced plans to extend service to Miami and Washington, D.C. and even hinted at eventually selling its robo-driving system to consumers, though no timeline has been given.
• General Motors: The next‑largest U.S. robotaxi program has disappeared. General Motors has shut down Cruise’s ride-hail business and is folding the team into GM’s driver‑assist R&D. Cruise launched in San Francisco in the 2021 to 2022 timeframe, running a fleet of Chevy Bolt BEVs without drivers at night. It logged more than 5 million driverless miles by late 2023 across San Francisco, Austin, and Phoenix. A series of incidents (including a collision where a Cruise taxi dragged a pedestrian) led the California DMV to suspend Cruise’s permits in October 2023. Facing federal investigations, GM halted funding and effectively shut down Cruise’s robotaxi operations. This marked a dramatic setback — after investing $10+ billion, GM abandoned its 2030 robotaxi revenue targets. As of Q1 2025, Cruise remains paused; GM is prioritizing its supervised driving technology (Super Cruise) instead.
• Tesla: Tesla is deploying semi-autonomous technology directly to consumers. Its “Full Self-Driving” (FSD) beta software is in hundreds of thousands of Tesla cars across the U.S., enabling Level 2 automated driving on city streets and highways. Drivers must remain alert and ready to intervene, as the system is not yet truly driverless. Tesla’s approach relies on cameras and neural networks. This pure vision strategy contrasts with virtually all other AV companies, which use LiDAR/radar for redundancy. Tesla indicated on its Q1 2025 earnings call that it aims to launch a pilot robotaxi service in Austin, Texas by mid-2025.

Figure 1: Waymo is the only U.S. company running a fully driverless, paid robotaxi service at scale. Source: techcrunch.com.

The picture is brighter for heavy trucks. Aurora ran its first actual driverless Class 8 loads between Dallas and Houston on May 1, 2025, and is expanding to El Paso and Phoenix. (See Figure 2.) Kodiak delivered the first customer-owned RoboTrucks in January, completing 100 driver-out hauls for Atlas Energy in West Texas. Gatik’s middle-mile box trucks are in paid service with retailers in Texas, Arkansas, Arizona, and Ontario (Canada).

Figure 2: Volvo Autonomous Solutions has partnered with Aurora. Source: foxnews.com. Image: Volvo Autonomous Solutions.

U.S. hardware supply is improving, but it is not yet on par with Chinese levels. Luminar has commenced serial LiDAR production for Volvo’s EX90, and NVIDIA’s new DRIVE Thor software on chips (SoCs) will be shipped inside Aurora’s next-generation trucks starting in 2027, while Qualcomm’s Snapdragon Ride Flex has secured Hyundai Mobis as a launch customer. Washington is emphasizing security. In March, the Commerce Department finalized the Connected Vehicle Rule, which bans Chinese software and certain chips from U.S. cars beginning in 2026, and the National Highway Traffic Safety Administration’s (NHTSA) AV STEP program (December 2024) finally offers companies a voluntary path to deploy vehicles without a steering wheel. Yet, Congress still has not passed a binding national AV law, leaving developers to juggle 50 sets of state regulations.

So, What Should We Do?

A decisive U.S. response will require action on eight fronts — each one urgent, none of which is optional.

1. Pass a federal law to replace today’s 50-state patchwork. Today, every autonomous vehicle program must navigate California’s disengagement reports, Arizona’s hands-off rules, and everything in between. NHTSA’s new AV STEP framework serves as a helpful voluntary bridge, but it cannot override state statutes or grant blanket approval for steering-wheel-free cars. Congress needs a 2025 statute that preempts conflicting state rules, incorporates AV STEP’s data reporting into law, and establishes national cybersecurity and over-the-air update requirements. Until that occurs, U.S. developers will continue to incur compliance costs that Chinese fleets never face under Beijing’s unified GB 4449x regime.
2. Re-shore the autonomy supply chain before turning off the tap. The Commerce Department’s Connected Vehicles Rule — effective March 2025 — blocks Chinese software and certain chips from Level 3-plus vehicles sold or tested in the U.S. That rule is a prudent security policy, but it also exposes how dependent U.S. pilot projects have become on low-cost Chinese LiDAR and AI silicon. Washington should utilize the open tranches of the CHIPS Act to fund fabrication facilities that produce automotive-grade lasers, radar MMICs, and compute accelerators — devices designed to speed up specific types of computing tasks — domestically, and pair the grants with a tax credit for Tier 1 suppliers that source at least 75% of their sensors domestically by 2028.
3. Close the data gap with a national AV Data Trust. Waymo is now booking more than 250,000 fully autonomous paid rides every week across four U.S. metro areas, however, that still lags the vast amount of road data generated by Chinese fleets each month. A federally managed, anonymized repository — open to companies that contribute incident and disengagement logs — would enable smaller U.S. players to train on a nationwide dataset and accelerate verification research for regulators and academia alike. Advanced Research Projects Agency–Infrastructure (ARPA-I), the new advanced transportation research agency, already has the legal charter to host such shared infrastructure.
4. Build real “automation corridors” instead of ad‑hoc test routes. Texas just watched Aurora launch true driver-out freight runs between Dallas and Houston, but every company still relies on piecemeal state or local investments in 5G, truck parking, and roadside sensing. The next surface transport bill should fund at least three multi-state corridors — think I-10, I-40, and I-95 — with C-V2X units, edge-compute cabinets, and dedicated freight lanes so robo-trucks can scale revenue service quickly without negotiating every new jurisdiction from scratch.
5. Make safety transparency mandatory, not marketing. The voluntary tier of AV STEP asks companies to disclose incident metrics, but it remains just that — voluntary. Federal approval for any Level 4 deployment should be contingent on releasing standardized disengagement and crash data (per 10,000 miles) to the public. Uniform disclosure builds public trust faster than company blogs and prevents the information vacuums that helped sink past U.S. programs such as Cruise.
6. Fund moon‑shot R&D where China isn’t looking. Beijing’s strengths lie in its dense cities and controlled expressways. The U.S. can leverage its geography by tasking the ARPA-I with a $300 million, three-year program focused on all-weather, rural, and low-bandwidth autonomy. Success would open markets — ag, mining, military logistics — that Chinese players, bound to V2X-blanketed corridors, find difficult to navigate.
7. Secure talent and grow the industrial workforce. A dedicated “critical mobility technologies” visa track, coupled with a 20% refundable payroll tax credit for AV apprenticeships at OEMs and suppliers, would keep AI researchers and maintenance technicians onshore as demand for autonomy expertise spikes. Without fresh human capital, new fabrication facilities and R&D grants will remain unused.
8. Fight the standards war now, not later. China’s cybersecurity rules already are infiltrating Belt and Road markets, (a global infrastructure development plan initiated by China in 2013), threatening to confine emerging economies to a Beijing-defined tech stack. Washington should spearhead an Organization for Economic Cooperation and Development-back Cyber Secure Vehicle Protocol that aligns with — but is not dictated by — Chinese specifications and then mandate its adoption in future U.S. trade deals. The goal is to establish interoperable, trust-anchored rules before the GB series becomes the only game in town.

Time is short. China’s AV revenue pool is forecast to grow tenfold this decade, and each quarter the U.S. delay turns the gap into a moat of cost, data, and influence that will be brutally expensive to breach.

Let’s Seize This Moment

America thrives on challenges. We’re a nation of dreamers, builders, and innovators. China may be ahead today, but the race isn’t over. With our talent, resources, and unwavering spirit, we cannot only catch up but lead the way.

Imagine a future where our roads are safer, our cities are smarter, and our technology sets the global standard. If we don’t act, our roads will be filled with foreign-made AI vehicles, our data will be in the hands of a government that doesn’t share our values, and our children will grow up in a world where America is a follower, not a leader.

The AI vehicle revolution is here and is a race we can’t afford to lose. We’re teetering on the edge of irrelevance — our economy, security, and freedom hang in the balance. America, the time for complacency is over.