Tesla Vehicles Navigate San Francisco Outage While Waymo Fleet Halts
A widespread power outage in San Francisco disabled traffic lights across large sections of the city. The blackout affected approximately 130,000 PG&E customers at its peak. Waymo’s autonomous ride-hailing service suspended operations citywide. Multiple Waymo vehicles stopped in intersections with hazard lights activated. Social media videos captured these vehicles blocking traffic as drivers maneuvered around them.
Waymo coordinated with city officials during the incident. The company later stated it focused on rider safety and emergency access. Service resumed once power stabilized in most areas.
Tesla vehicles equipped with Full Self-Driving Supervised continued operating without interruption. The system relies on vision-based neural networks trained on billions of real-world driving miles. These include scenarios involving non-functional traffic signals. Videos shared online showed Tesla cars treating darkened intersections as four-way stops and proceeding safely.
The outage originated from a substation issue. PG&E crews restored power to over 110,000 customers overnight. Remaining disruptions concentrated in districts like the Presidio and Richmond.
This event highlighted architectural differences in autonomous driving approaches. Waymo’s system incorporates high-definition maps and lidar sensors. These can depend on consistent environmental signals like active traffic lights. Tesla’s end-to-end AI processes raw camera inputs directly. This enables greater adaptability to unexpected conditions without remote intervention.
Industry observers noted the contrast in resilience during infrastructure failures. Tesla emphasized its training data volume as a key factor. The company has accumulated over 6 billion miles of fleet data for FSD development.
San Francisco hosts testing for multiple robotaxi operators. Regulators continue monitoring performance in varied urban scenarios. The incident underscored challenges for deploying fully driverless services at scale in unpredictable real-world environments.
