Robotaxis Q1 2026: Real Trip Counts, Safety Rates, and Whether Anyone Makes Money

The first quarter of 2026 delivered what the autonomous-vehicle industry has been promising for a decade: real commercial robotaxi rides, in multiple cities, with paying passengers who did not touch a steering wheel. Waymo—Alphabet’s self-driving unit—reported its fifth commercial market opening. Tesla moved a supervised Cybercab pilot from announcement to street. Zoox, Amazon’s bidirectional AV subsidiary, began its first fare-paying operations. The milestone numbers are finally arriving. So are the hard questions about safety, cost-per-mile, and whether any of these companies can make money before their backers lose patience.

What follows is a city-by-city accounting of where autonomous rides actually exist today, what the safety data shows, and a frank look at the unit economics—with a clear label on every figure that is an estimate rather than a disclosed number.

Waymo: Five Cities, One Dominant Market Leader

Waymo One has operated in Phoenix since October 2020, making it the longest-running commercial driverless service in the United States. San Francisco came online for fully driverless paid rides in August 2023 after the California Public Utilities Commission issued its permit. Los Angeles—initially covering West Hollywood and Santa Monica—followed in 2024, alongside an Uber partnership for integrated booking. Austin opened for rider-facing Waymo service in mid-2025, and Atlanta launched commercial service in Q4 2025 covering Midtown and Buckhead. Miami remains in a testing-only phase.

Trip volume numbers have been disclosed sporadically via Alphabet earnings calls. The 50,000 weekly paid rides milestone was confirmed on Alphabet’s Q1 2024 earnings call in May 2024. By late 2024, Waymo crossed 150,000 weekly paid trips—a figure Alphabet disclosed at its investor day. No official per-city breakdown has ever been released.

Q1 2026 Estimated Trip Volumes by City

• Phoenix (commercial since October 2020): Approximately 85,000–105,000 trips per week. Phoenix remains Waymo’s highest-volume market, with service covering Chandler, Tempe, Mesa, and central Phoenix.
• San Francisco (fully driverless since August 2023): Approximately 40,000–55,000 trips per week. After Cruise’s October 2023 permit suspension, Waymo absorbed significant diverted demand and expanded its San Francisco geofence.
• Los Angeles (commercial since 2024): Approximately 50,000–70,000 trips per week across West Hollywood, Santa Monica, Beverly Hills, and West LA.
• Austin (opened mid-2025): Approximately 12,000–22,000 trips per week within a geofenced area covering downtown, the Domain district, and the University of Texas corridor.
• Atlanta (opened Q4 2025): Approximately 6,000–12,000 trips per week in Midtown and Buckhead.

The vehicle in commercial service is the fifth-generation Waymo system mounted on a Jaguar I-PACE, carrying 29 cameras, 5 lidar units, 6 radar units, and an audio sensor array. Waymo announced a partnership with Zeekr—a premium EV brand under China’s Geely Group—in 2024 to co-develop a sixth-generation purpose-built AV platform. Testing of Zeekr-based vehicles was underway in late 2025; no commercial deployment timeline has been confirmed for that generation.

Tesla Robotaxi: The Cybercab Moves From Stage to Street

Elon Musk unveiled the Tesla Cybercab on October 10, 2024, at an event called “We, Robot” held at Warner Bros. Studios in Burbank, California. The Cybercab is a two-seat vehicle with no steering wheel, no pedals, butterfly doors, and inductive wireless charging. Musk projected production starting in 2026 and commercial service beginning in 2025. The vehicle uses Tesla’s Hardware 4 FSD computer—a custom 7-nm chip installed in Model 3 and Model Y vehicles from mid-2023—paired with a camera-and-radar sensor suite that deliberately excludes lidar.

Tesla’s Full Self-Driving v12, introduced in October 2023, replaced a modular rules-based pipeline with an end-to-end neural network trained on video. FSD v12.3 shipped in February 2024. FSD v13 rolled out through Q4 2024 and Q1 2025, with Tesla describing it as capable of unsupervised driving in limited geofenced areas. Tesla reported approximately 500,000 active FSD subscribers as of early 2024 at a subscription price of $99 per month following a mid-2024 price reduction from $199.

Tesla’s structural advantage, if FSD scales commercially, is sheer vehicle volume. The company produced 1.77 million vehicles in 2023 (Tesla 2023 Annual Report). A software-enabled upgrade converting even a fraction of existing Model 3 and Model Y fleets to robotaxi service would instantly exceed Waymo’s hardware-deployed scale. The bet is that camera-only perception, trained at Tesla’s fleet data scale, can achieve Waymo-comparable urban safety—a proposition that remains unproven in commercial, unsupervised operation as of this writing.

Zoox: Amazon’s Bidirectional Bet on Las Vegas

Amazon acquired Zoox in June 2020 for approximately $1.3 billion. Unlike Waymo’s retrofit approach, Zoox built a purpose-designed vehicle from the ground up. The Zoox AV is fully symmetrical—it drives in either direction with equal capability—and seats four passengers in a face-to-face configuration. There is no steering wheel, no pedals, and no external visual indicators distinguishing front from back. The sensor stack includes 12 cameras, 10 radar units, and a multi-lidar array. CEO Aicha Evans, who joined in January 2021, has described the vehicle as designed for “cities that weren’t built for cars.”

Zoox has accumulated testing miles in Foster City (its California headquarters), San Francisco, Las Vegas, and Seattle. Nevada’s regulatory framework—which permits autonomous commercial service with prior state approval and does not require a human safety driver—made Las Vegas a natural first commercial market. The Las Vegas Strip and convention district offer a dense, relatively constrained geography well-suited to a geofenced launch.

Safety: What the Incident Data Actually Shows

The most rigorous public dataset on AV safety is NHTSA’s Standing General Order reporting (SGOs 2021-01 and 2022-01), which requires manufacturers to report collisions involving automated systems within 30 days. Waymo, operating the most commercial AV miles in the United States, appears most frequently in SGO filings. Raw incident counts without miles-per-incident context are meaningless; incident rate per million miles is the relevant metric.

Waymo published a study in December 2023 covering 7.14 million rider-only miles in Phoenix and San Francisco, co-authored with Swiss Re’s actuarial team. Key findings: Waymo’s system experienced 6.7 times fewer injury-causing crashes per million miles and 2.3 times fewer property-damage-only crashes compared to a human-driver baseline derived from insurance industry data for the same geographies and time periods. The full 7.14 million mile dataset recorded zero fatalities.

The zero-fatality figure requires careful interpretation. The U.S. national fatality rate is approximately 1.37 deaths per 100 million vehicle miles (NHTSA 2022 data). A proportional extrapolation predicts roughly 0.10 fatalities in a 7 million mile sample—meaning zero observed fatalities is consistent with human-level or better performance, but does not definitively prove superior safety. RAND Corporation’s 2016 analysis estimated 11 billion test miles would be needed for 95% statistical confidence that an AV is safer than a human driver; geofenced urban operation has lowered that bar, but the sample-size constraint remains real.

The defining AV safety incident of recent years occurred on October 2, 2023: a driverless Cruise vehicle struck a pedestrian in San Francisco and subsequently dragged her approximately 20 feet before stopping. The California DMV suspended Cruise’s operating permit; Cruise withdrew from all markets nationwide. The incident did not involve Waymo or Zoox. It did demonstrate that AV safety failures, when they occur, produce outsized regulatory consequences—a lesson that every remaining operator has internalized into its development priorities.

Unit Economics: The Business Nobody Has Solved Yet

Waymo’s fifth-generation vehicle is a Jaguar I-PACE with a 2024 MSRP of approximately $68,000. The sensor suite—historically the dominant cost driver—has declined from an estimated $150,000–200,000 in early generations to approximately $50,000–80,000 in fifth-gen deployment, per Morgan Stanley analyst estimates from 2024. Total per-vehicle cost: approximately $120,000–150,000.

Amortized over 300,000 commercial miles, hardware depreciation alone runs $0.40–0.50 per mile. Fleet management, remote vehicle assistance staff, compute and data infrastructure, maintenance, cleaning, and insurance add an estimated $0.70–1.10 per mile. Estimated total operating cost: $1.10–1.60 per mile. Waymo fares in Phoenix and San Francisco average approximately $10–20 per trip covering 3–6 miles, implying roughly $2.50–4.50 per mile in gross revenue. That positive contribution margin does not cover corporate overhead: Alphabet’s Other Bets segment, which includes Waymo, posted a combined operating loss of $1.14 billion in Q3 2024 alone. Fleet utilization above 50% is likely required for segment profitability; no AV operator has confirmed reaching that threshold.

Tesla’s Cybercab, if it reaches its stated $30,000 production price target, restructures these economics entirely. Remove lidar, reduce the sensor suite to cameras and radar at consumer hardware cost, and per-vehicle expense drops to near-conventional-car levels. A fleet of 100,000 Cybercabs at $30,000 each represents a $3 billion capital outlay—versus what Waymo would require for equivalent trip capacity at its current vehicle cost. The math is compelling on paper; the safety validation at commercial scale is the unresolved variable that determines whether the math is real.

What to Watch in Q2 2026

Three dynamics will define the competitive landscape through the rest of 2026: whether Tesla confirms a public fare-paying service with disclosed safety reporting; whether Waymo crosses 300,000 weekly trips as its Atlanta and Austin markets mature; and whether Zoox publishes its first ridership figures from Las Vegas. The race is no longer hypothetical. The question has shifted from “can AVs work on public roads” to “can the unit economics work fast enough to justify the capital being deployed.” The answer will be determined by trip counts, safety records, and per-mile margins—not by press events in parking lots.