Waymo says it has begun testing its autonomous robotaxi service with employee volunteers in San Francisco, its first expansion beyond the Phoenix metro area
Kyle Wiggers / VentureBeat :
Context & Ripple Effects
In February 2021, Waymo's robotaxi footprint was a single metro: the service ran only in the [[a:none|Phoenix area]], where it had been open to the public since late 2020, and San Francisco testing was limited to employee volunteers rather than paying riders. That employee-only beachhead was the template for everything after — Waymo used it again on Phoenix freeways in early 2024 before opening faster rides more broadly.
The pattern held in San Francisco: staff testing led to driverless rides without safety drivers by spring 2022, then Waymo dropped the city's waitlist entirely in mid-2024, and by that August co-CEO Tekedra Mawakana was citing 100K+ paid rides weekly across SF, LA, and Phoenix. The 2021 expansion is the origin point of a playbook — test with employees first, expand city by city — that later carried Waymo to freeway rides in three metros and next-gen Ojai vehicles.
First-order effects
- Waymo's operations team now has to manage edge cases in an unfamiliar city — the relationships show this means paying workers roughly $20–$24 via the Honk app to close doors and $60–$80 for towing when its cars get stuck, costs that scale with every new geography.
Second-order effects
- San Francisco's dense urban grid becomes Waymo's proving ground for scenarios Phoenix never posed, and the relationships show the strain: during one SF power outage the system inundated human supervisors with confirmation prompts, exposing how thinly remote-assistance staffing stretches as fleets grow — Waymo has trained about 25,000 humans to assist roughly 3,000 robotaxis.
Third-order effects
- The employee-first-then-public rollout sequence is becoming the industry's regulatory de facto standard for entering new cities, and the bottleneck shifts from driving capability to the human support layer — remote assistance throughput, not autonomy software, is emerging as the constraint on how fast any robotaxi operator can open a new market.
The trend: Robotaxi operators are scaling city by city through employee-tested rollouts, and the binding constraint is shifting from autonomous-driving capability to the human remote-assistance infrastructure behind each fleet.