Waymo's robotaxi fleet is expanding aggressively, but a San Francisco power outage last December exposed a fundamental problem: self-driving cars still struggle with situations that human drivers handle instinctively. The incident, where lines of Waymo vehicles simply stopped when traffic signals went dark, underscores a gap between the company's impressive growth metrics and the messy reality of deploying autonomous vehicles at scale.

How Fast Is Waymo Actually Growing?

The numbers look impressive on paper. Waymo now operates across more than 1,400 square miles spanning 11 U.S. cities, a 27% increase from its previous coverage area, and has completed 20 million trips since launching its robotaxi service. The company is targeting around 1 million rides per week by the end of 2026, roughly four times its current volume. To achieve this scale, Waymo is cutting costs by switching to cheaper vehicles made by China's Geely, which require fewer expensive cameras and sensors than the Jaguar I-PACE models it previously used. A $16 billion funding round is bankrolling new factories and international launches in London and Tokyo.

Waymo isn't the only player racing to scale. WeRide's global robotaxi fleet reached approximately 1,300 vehicles as of April 2026, while Tesla has begun limited ride-hailing services in Austin and the San Francisco Bay Area. Commercial fleet operators are testing autonomous driver kits in everything from compact sedans to Mercedes Sprinter vans used for airport shuttles and last-mile delivery routes.

Why Do Edge Cases Still Pose Such a Problem?

The San Francisco power outage revealed something uncomfortable: when conditions deviate from what the AI has been trained to handle, the vehicles default to stopping entirely. Beyond traffic lights, earlier rollouts produced even more awkward moments. Autonomous vehicles were spotted driving into emergency response scenes because they didn't recognize fire trucks or fire hoses. These aren't minor glitches; they're evidence that the technology is still in a middle stage where ambition outpaces what the streets, laws, and public are ready to handle.

Supporters of autonomous vehicles point to compelling safety data. Monitored deployments have shown reductions of 86% in property damage and roughly 90% in injury claims, with AI systems cutting down on human error factors like distraction and poor reaction time. The broader argument is compelling: more than 40,000 people die each year on U.S. roads, most in crashes caused by human error. Yet those edge cases remain a persistent vulnerability that requires human oversight.

What's Holding Back Public Adoption?

Trust remains the biggest barrier. A February 2025 AAA survey found that only 13% of U.S. drivers said they would trust riding in an autonomous vehicle, up slightly from 9% in 2024. About 6 in 10 drivers report feeling uneasy about getting into a self-driving car. This hesitation isn't irrational; it reflects legitimate concerns about whether the technology can handle the unpredictable nature of real-world driving.

How Are Regulators Trying to Keep Up?

Washington is finally catching up, though the process remains messy. The U.S. Department of Transportation plans to propose three new rules in spring 2026 to update outdated safety standards for vehicles with automated driving systems. These updates focus specifically on vehicles built without steering wheels or pedals, and one rule would modernize requirements for how transmissions shift and how vehicles start and stop.

On Capitol Hill, Representatives Bob Latta and Debbie Dingell released the SELF DRIVE Act of 2026, which would be the first federal statute dedicated to the safety of autonomous vehicles. The current cap on the number of vehicles a manufacturer may deploy under a testing permit is 2,500 per year; companion legislation would raise that cap to 90,000, a 36-fold increase. Critics worry the bills lean too heavily on self-certification and vague safety language. Supporters argue the current patchwork of 34 different state laws, some permissive and some restrictive, is slowing American firms while Chinese competitors expand.

Steps to Understanding the Autonomous Vehicle Landscape

• Track Deployment Scale: Monitor how many square miles and cities each company operates in, as this indicates real-world readiness beyond controlled testing environments.
• Watch for Edge Case Incidents: Pay attention to failures in unusual conditions, such as power outages or emergency scenes, as these reveal gaps in AI training and decision-making.
• Follow Regulatory Changes: Keep up with federal and state legislation, including permit caps and safety standards, as these will determine how quickly companies can scale operations.
• Assess Public Trust Metrics: Review surveys on consumer willingness to ride in autonomous vehicles, as adoption rates depend on both technology maturity and public confidence.

The autonomous vehicle industry has crossed a real threshold. Robotaxis are picking up paying passengers, logging serious mileage, and forcing lawmakers to rewrite rulebooks built around the assumption of a human at the wheel. But scaling from a few hundred square miles in sunny cities to nationwide, all-weather, all-conditions transportation is a fundamentally different challenge. The next few years won't decide whether autonomy works. They'll decide how fast society is willing to trust it, and whether the technology can move beyond its current limitations faster than regulators and the public can adapt to it.