Waymo to update software after major power outage

Waymo will update its self-driving software after a major power outage in San Francisco caused its autonomous taxis to stop in traffic, the company said.

The incident occurred on December 20, 2025 when a fire at a PG&E substation disrupted electricity for about 130,000 customers.

During the outage, the Waymo Driver system faced a surge in requests for safety confirmation at darkened intersections, which delayed the company’s ability to address stalled vehicles.

Videos on social media showed multiple Waymo cars stopped in the streets with hazard lights, leading to traffic congestion.

🔗 Source: Bloomberg

🧠 Food for thought

Implications, context, and why it matters.

California may review Waymo permits after outage

The California Public Utilities Commission (CPUC) issues autonomous vehicle permits 1 and said it is ‘looking into’ the December 20 blackout that stalled Waymo cars 2.
The review matters since the CPUC regulates robotaxi services and waived a $200 million PG&E penalty tied to 2020 fires 3.
A gig workers union for app-based drivers wants regulators to suspend Waymo’s permits until safety is proven and run an independent probe 4.
Members of the San Francisco Board of Supervisors called for public hearings on traffic from disabled Waymo cars 3, adding political pressure on regulators.

Vendors can build outage-aware fleet tools for AVs

The outage exposed a gap in Waymo’s safety checks at darkened intersections. Vendors could offer software that pulls utility outage feeds plus traffic signal status in real time.
Buyers include autonomous vehicle (AV) operators such as Zoox, an Amazon-owned robotaxi developer 1, and ride-hailing firms in many cities. Waymo is moving into Austin and Atlanta 2.
The Outage Data Initiative Nationwide (ODIN) standardizes utility outage data and offers near real-time county data 5. California maintains outage datasets 6 vendors can use to build fleet orchestration tools, software that tracks vehicle status plus location to route or dispatch efficiently.
These systems could help AV operators reroute or send vehicles to depots before stalls. That answers Bryan Reimer, a Massachusetts Institute of Technology researcher focused on human–automation interaction, who calls for a mix of human and machine intelligence in emergencies 7.