The State of Robotaxi SF in 2026: What's Actually Operating
San Francisco remains the most active robotaxi market in the United States. If you've searched "robotaxi SF" to figure out what's actually available versus what's still in testing, this guide breaks down every commercial driverless service currently operating in the city — with hard numbers on pricing, safety records, coverage areas, and the technology making it all work.
As of early 2026, Waymo operates the largest commercial robotaxi fleet in San Francisco, while Zoox runs a more limited paid service. Cruise, once a dominant player, has been gradually rebuilding its operational footprint following its 2023 suspension and subsequent restructuring under GM's revised autonomy strategy. Below, we separate live, bookable services from testing programs and announced expansions.
Robotaxis operating commercially in SF (Waymo, Zoox) are SAE Level 4 — meaning the vehicle handles all driving tasks within a defined operational design domain (geofenced area, specific conditions) with no human driver present. This is fundamentally different from SAE Level 2 systems like Tesla Autopilot, FSD (Supervised), Ford BlueCruise, or GM Super Cruise, which require continuous driver attention and hands-on or eyes-on engagement. If a vehicle has a steering wheel and a driver's seat that you're expected to occupy, it is not a robotaxi — it's driver-assist. Never conflate the two.
Where Can You Actually Use a Robotaxi in SF Right Now?
Not every block of San Francisco is open for robotaxi service. Each operator maintains a geofenced service area that expands incrementally. Here's what's live as of early 2026:
| Operator | Service Status | SF Coverage Area | Hours | Vehicle |
|---|---|---|---|---|
| Waymo | Commercial (open to public via app) | ~80 sq mi — covers most of SF proper including SoMa, Mission, Marina, Sunset, Richmond, Financial District, and SFO airport runs | 24/7 | Jaguar I-PACE (custom 5th-gen sensor suite), Zeekr RT (rolling deployment) |
| Zoox | Commercial (limited public access, invite + waitlist) | ~12 sq mi — SoMa, Financial District, portions of Mission Bay | Limited hours (typically daytime/evening) | Zoox purpose-built carriage (no steering wheel, bidirectional) |
| Cruise | Testing with safety drivers; limited commercial pilot re-launch | Restricted zones — not broadly open to public rides | Limited | Chevrolet Bolt EV / Origin (testing) |
Key takeaway: If you want to book a driverless ride in SF today with no safety driver behind the wheel, Waymo is your primary option. Zoox is expanding access but remains capacity-constrained. Cruise is not yet a viable consumer robotaxi option in San Francisco.
Robotaxi SF Pricing: How Does It Compare to Uber and Lyft?
Robotaxi pricing has stabilized as operators shift from heavily subsidized promotional fares toward sustainable unit economics. Here's how a typical robotaxi SF ride compares to conventional rideshare:
| Service | 3-Mile Ride (est.) | 7-Mile Ride (est.) | Surge Pricing? | Tipping |
|---|---|---|---|---|
| Waymo | $11–$15 | $22–$32 | Yes (demand-based, generally less volatile than Uber/Lyft surge) | In-app tip option (not expected) |
| UberX | $12–$18 | $24–$38 | Yes (frequent surge during peak hours) | Expected (15–20% typical) |
| Lyft Standard | $11–$17 | $23–$36 | Yes (Prime Time pricing) | Expected (15–20% typical) |
| Zoox | Promotional / free (limited access) | N/A (short-distance service area) | No | No |
Waymo fares are generally competitive with UberX and Lyft Standard for comparable distances — and when you factor in that there's no driver to tip, the total out-of-pocket cost is often lower. During peak demand, Waymo's pricing algorithm tends to be less aggressive than rideshare surge multipliers, though wait times can stretch to 10–15 minutes when demand outpaces fleet availability.
How to book: Download the Waymo app (iOS/Android), create an account, set your pickup and drop-off within the SF service area, and confirm. The car arrives, you unlock via the app, and the ride proceeds with no one in the driver's seat. Zoox access currently requires joining a waitlist or receiving an invitation.
How Safe Are Robotaxis in San Francisco? The Real Data
Safety is the most scrutinized dimension of robotaxi operations — and it should be. California's DMV requires operators to file annual disengagement reports, and the National Highway Traffic Safety Administration (NHTSA) tracks incidents involving automated driving systems. Here's what the data shows:
- Cumulative fully-autonomous miles (nationwide): Over 10 million driverless miles completed as of late 2025
- At-fault collision rate: Waymo's published data shows approximately 85% fewer police-reported injury-causing crashes per million miles compared to human drivers in comparable urban environments (per Waymo's safety reports and third-party analyses)
- Minor contact incidents: The majority of reported contacts involve low-speed (<10 mph) scenarios — stationary objects, other vehicles encroaching into the robotaxi's lane, or cautious stops that result in rear-end contact by following human drivers
- Cruise's 2023 incident: The October 2023 event in which a Cruise vehicle dragged a pedestrian who had been struck by a human-driven car led to California DMV suspending Cruise's permit. This single event reshaped the entire regulatory landscape and underscored that post-collision behavior (not just collision avoidance) is a critical autonomy challenge
The data tells a nuanced story. Robotaxis in SF have demonstrated collision rates meaningfully lower than human baselines for injury-causing events. However, they are not incident-free. The types of incidents differ: robotaxis occasionally exhibit overly cautious behavior (sudden stops, extended pauses at intersections) that can confuse human drivers and cyclists. Rear-end contacts caused by following vehicles are the most common incident type.
For context on exposure: Waymo's SF fleet operates hundreds of vehicles completing tens of thousands of paid rides per week. At this volume, isolated incidents are statistically inevitable — the relevant metric is the rate per million miles, where the data favors autonomous operation in geofenced urban environments.
Under the Hood: Robotaxi Technology Stacks Explained
Understanding what's on the roof — and throughout the vehicle — helps explain both the capabilities and the limitations of today's SF robotaxis.
| Component | Waymo (5th-Gen Driver) | Zoox (Purpose-Built) | Tesla FSD (Level 2 — not a robotaxi) |
|---|---|---|---|
| Primary perception | LiDAR (custom-built, roof-mounted) + cameras + radar | LiDAR + cameras + radar (multi-sensor, redundant) | Camera-only (8 cameras, no LiDAR, no radar in current HW4) |
| Mapping | HD pre-mapped routes, continuously updated | HD pre-mapped routes | No HD maps (real-time perception-based) |
| Redundancy | Dual-compute, redundant braking/steering, backup power | Quad-redundant systems (steering, braking, compute, power) | Single compute path; driver is the redundancy |
| Geofencing | Yes — strict operational design domain | Yes — strict operational design domain | No — operates anywhere (with driver supervision) |
| SAE Level | Level 4 (driverless in ODD) | Level 4 (driverless in ODD) | Level 2 (driver required) |
The LiDAR vs. camera-only debate remains one of the most consequential technology splits in autonomy. Waymo and Zoox rely on LiDAR as a core sensor — it provides precise 3D spatial data independent of lighting conditions, critical for nighttime urban driving and detecting objects at range. Tesla's approach eschews LiDAR entirely, betting on neural-network-based vision processing. Notably, Tesla's FSD remains a Level 2 driver-assist system requiring human supervision; it is not a robotaxi platform and is not approved for driverless commercial operation anywhere.
HD mapping is another defining characteristic. Both Waymo and Zoox pre-map every street in their service area at centimeter-level precision. This means expansion to new neighborhoods requires extensive mapping drives before service can go live — a bottleneck that limits how quickly coverage areas grow.
Current Limitations: Where Robotaxis in SF Still Struggle
Despite impressive capabilities, robotaxis in San Francisco face well-documented limitations. Understanding these is essential for setting expectations before your first ride.
- Heavy fog and rain: While Waymo's 5th-gen sensors handle moderate rain, dense fog (common in SF's western neighborhoods during summer) can degrade LiDAR and camera performance. Rides may be rerouted or, in rare cases, the vehicle will pull over safely and end the trip.
- Construction zones: Unexpected lane closures, temporary barriers, and flaggers directing traffic remain challenging edge cases. Robotaxis may take conservative routes to avoid active construction or pause until the path is clear.
- Unprotected left turns at complex intersections: Multi-lane intersections with simultaneous pedestrian crossings, oncoming traffic, and bike lanes require nuanced gap-finding. Robotaxis tend to be overly cautious here, sometimes waiting through multiple light cycles.
- Geofence boundaries: You cannot book a ride from SF to Oakland, Marin, or the Peninsula. Pickup and drop-off must both fall within the operator's approved service area.
- Emergency vehicle interaction: While systems detect and yield to emergency vehicles, complex scenarios (multiple emergency vehicles, accident scenes blocking the route) can cause the vehicle to stop and wait for remote assistance.
- No trunk access: Waymo's Jaguar I-PACE does not provide rider access to the trunk. If you're carrying large luggage or groceries beyond what fits in the cabin, plan accordingly.
- Accessibility: Current Waymo vehicles require the rider to enter a standard SUV-height cabin. Wheelchair-accessible robotaxi options remain extremely limited in SF.
Who Regulates Robotaxis in San Francisco?
Robotaxi operations in SF sit at the intersection of state and local regulation — and the approval process has become considerably more rigorous since the Cruise incident of late 2023.
California DMV: Issues autonomous vehicle testing and deployment permits. Operators must submit disengagement reports, maintain insurance, and demonstrate safety performance. After the Cruise suspension, the DMV tightened its review process for deployment permits.
California Public Utilities Commission (CPUC): Regulates commercial passenger service — the "taxicab" side. Operators must obtain a TCP (Transportation Charter Party) carrier permit or equivalent authorization to charge fares for driverless rides. The CPUC's Autonomous Vehicle Program outlines the phased approval framework.
San Francisco Municipal Transportation Agency (SFMTA): Manages local street operations, parking, and curb access. SFMTA has advocated for tighter operational constraints including fleet-size caps, speed limitations in school zones, and data-sharing requirements.
NHTSA (Federal): Sets vehicle safety standards and investigates defects. NHTSA's Standing General Order requires manufacturers and operators to report crashes involving Level 2 and above automated systems within one day of learning about them.
The regulatory landscape has shifted from permissive expansion to cautious, data-driven approval. Operators expanding to new cities now face longer approval timelines, more granular data-sharing requirements, and greater local-government input into fleet sizes and operating conditions.
Robotaxi SF FAQ: Common Questions Answered
Can I use a robotaxi to get to SFO airport from downtown SF?
Yes — Waymo's SF service area includes trips to and from San Francisco International Airport (SFO). Expect a fare in the $35–$55 range from downtown depending on demand and exact routing. Oakland Airport (OAK) is outside the current geofence.
Is this truly self-driving, or is there a remote human controlling the car?
The vehicle drives itself — there is no remote human "joystick" operator steering the car in real-time. However, remote assistance is available: if the vehicle encounters an ambiguous situation (e.g., a complex construction zone), it can pause and request guidance from a remote fleet operations team, which may confirm a route or approve a maneuver. The remote team does not directly drive the vehicle.
What happens if the robotaxi gets into an accident while I'm a passenger?
The vehicle is insured by the operator (Waymo, Zoox). In the event of a collision, the vehicle's systems will execute a controlled stop, contact emergency services if needed, and the rider can reach support via an in-vehicle button. The operator's insurance covers passenger injuries — you are not liable as a rider.
How do robotaxis handle SF's steep hills?
Waymo's fleet handles SF's hills competently — the Jaguar I-PACE's electric powertrain delivers immediate torque for hill starts, and the sensor suite is calibrated for the city's terrain. That said, the most extreme grades (Filbert Street, portions of Nob Hill) may fall outside preferred routing, and the system will choose navigable streets.
Are robotaxis available in other cities besides San Francisco?
Waymo operates commercial robotaxi service in Phoenix (including Tempe, Mesa, Chandler), Los Angeles, and has been expanding in Austin. Zoox has testing and limited operations in Las Vegas and Foster City, CA, alongside SF. Each city's service area, fleet size, and hours vary independently.
Will robotaxis replace Uber and Lyft drivers?
Not in the near term. Robotaxi fleets are capital-intensive (each vehicle carries $100,000+ in sensor and compute hardware), geofenced, and limited in operational conditions. They complement rideshare networks rather than replacing them. Uber and Lyft are likely to integrate robotaxi options into their platforms — Uber has already announced partnerships for autonomous ride fulfillment in select markets.
