From Hype to Headway
For a decade, the robotaxi was the technology that was always “five years away.” The promise — summon a car with no driver, ride across town, pay a fraction of a human-driven taxi’s fare — was tantalizing enough to attract tens of billions in investment from companies including Waymo, Cruise, Uber, Lyft, and a rotating cast of automakers and startups. But the technical challenge — building a system that can navigate every edge case a human driver handles without thinking — proved far harder than the early optimists predicted. The result was a graveyard of ambitious timelines: Tesla’s repeated “full self-driving next year” promises, Cruise’s 2019 claim of a driverless service by year’s end, and countless quieter failures that never made headlines.
By 2025, the picture has changed. Robotaxis are no longer a promise. They’re a service — available in a growing number of cities, carrying real passengers, and generating real revenue. The rollout is uneven, geographically constrained, and nowhere near the scale that the hyped predictions promised. But it’s real. Here’s where driverless cars actually work.
The Waymo Footprint
Waymo, the Alphabet-owned company formerly known as the Google Self-Driving Car Project, is the undisputed leader. As of early 2025, Waymo operates commercial driverless ride-hailing services in Phoenix, San Francisco, Los Angeles, and Austin, with service expansions planned for additional cities. The company passed 100,000 paid weekly rides in 2024 — a milestone that took years longer than optimists predicted but represents genuine commercial scale. Waymo’s total fleet of driverless vehicles numbers in the hundreds, with the Jaguar I-PACE electric SUVs forming the backbone of its service.
Waymo’s safety record is its most important asset. The company’s vehicles have driven tens of millions of fully autonomous miles, and its published safety data (validated by independent researchers including Swiss Re) shows that Waymo vehicles are involved in fewer injury-causing crashes than human drivers in comparable conditions. A 2023 study by Swiss Re analyzed Waymo’s insurance claims data and found an 88% reduction in property damage claims and a 92% reduction in bodily injury claims compared to human-driven vehicles. These numbers, while generated by Waymo-funded research, have held up under independent scrutiny.
The economics are improving but still unproven. Waymo’s cost per mile remains higher than human-driven ride-hailing in most markets, primarily because of the amortized cost of its sensor suite (lidar, radar, and cameras) and the remote assistance infrastructure that monitors its vehicles. But the company has driven down sensor costs dramatically — its custom-built lidar, developed in-house, costs a fraction of the $75,000 units it used in its early vehicles. Waymo’s leadership has said the company is on a path to unit economics that make driverless ride-hailing cheaper than human-driven alternatives, but the timeline for profitability remains unclear.
Cruise’s Collapse and Recovery
Cruise, General Motors’ autonomous vehicle subsidiary, provides the cautionary tale. In October 2023, a Cruise robotaxi in San Francisco struck and dragged a pedestrian who had been hit by a separate human-driven vehicle. The incident — and, more damagingly, Cruise’s initial handling of it, which allegedly minimized the severity to regulators — led to the suspension of Cruise’s operating permits in California, the recall of its entire fleet, the resignation of its CEO, and a 24% workforce reduction. GM subsequently slashed Cruise’s budget by more than $1 billion and announced a shift in strategy away from the Origin shuttle (a purpose-built robotaxi without a steering wheel) toward a more conservative approach.
By 2025, Cruise was slowly rebuilding — resuming supervised testing in select markets with human safety drivers — but the damage to its reputation and its business model was substantial. The Cruise incident became the industry’s defining cautionary tale: a reminder that in the robotaxi business, a single failure of safety, transparency, or judgment can set a company back years. It also hardened the regulatory environment, with cities like San Francisco imposing stricter oversight on autonomous vehicle operations in the aftermath.
Baidu’s Apollo Go and the China Angle
While Waymo and Cruise dominate the Western narrative, China’s Baidu has quietly built the world’s largest robotaxi operation. Apollo Go, Baidu’s autonomous ride-hailing service, operated in over 10 Chinese cities by 2024, with Wuhan as its flagship market. The scale is genuinely impressive: Apollo Go has completed over 6 million cumulative rides, and in Wuhan, the service operates hundreds of driverless vehicles across a growing service area. Baidu’s approach differs from Waymo’s in a key respect: it operates a “mixed fleet” that includes vehicles with and without safety drivers, allowing it to scale more aggressively while managing risk.
China’s regulatory environment has been more permissive than the United States in some respects, with the central government explicitly promoting autonomous driving as a strategic priority and several cities issuing commercial licenses more quickly than their US counterparts. But the same safety challenges apply, and Apollo Go’s rapid expansion has generated local pushback from taxi drivers worried about job displacement — a tension that will shape robotaxi policy globally as the technology scales.
The Accident Data Nobody Wants to Talk About
The robotaxi industry’s safety claims rest on a foundation of data that is, at best, incomplete. Waymo’s published safety record is strong, but it reflects a specific set of conditions: geofenced service areas, favorable weather, and a fleet that avoids the most dangerous driving conditions (heavy snow, extreme fog, unmapped construction zones). The comparison between autonomous vehicles and human drivers is further complicated by the fact that humans drive everywhere, while robotaxis drive only where they’re safe. The relevant comparison isn’t “robotaxis vs. all human drivers” but “robotaxis vs. human drivers in equivalent conditions” — a comparison that’s harder to make cleanly.
The honest assessment is that robotaxis are safer than the average human driver in their operating conditions, but the evidence for broader safety superiority is still accumulating. The technology has made enormous progress since 2015, when the first public trials began. It’s also true that the hardest problems — edge cases, adverse weather, unmapped environments — remain unsolved at scale. The robotaxi rollout of 2025 is a genuine achievement. It’s also a reminder that “driverless” doesn’t mean “problem-free,” and that the path from pilot to ubiquity is longer than the hype cycles suggest.
What the Next Five Years Hold
The most likely trajectory is continued geographic expansion at a measured pace, driven by the economics. Waymo’s ride volume growth, Baidu’s aggressive Chinese expansion, and the entry of new players (including Amazon’s Zoox, which began public testing in 2024, and Tesla’s repeatedly-deferred robotaxi ambitions) suggest that driverless ride-hailing will be available in dozens of cities worldwide by 2030. The technology will improve, the costs will fall, and the safety data will accumulate. But the transition will be slower than the optimists promise and less apocalyptic than the skeptics fear. The robotaxi isn’t going to replace human drivers overnight. It’s going to coexist with them — in more places, at greater scale, every year — until, at some point that’s hard to predict, the economics tip decisively in favor of the machines.
The Jobs Question Nobody Wants to Answer
The robotaxi industry’s most politically sensitive question isn’t about safety — it’s about employment. There are approximately 350,000 taxi and ride-hail drivers in the United States. Add delivery drivers, chauffeurs, and other professional driving roles, and the number approaches a million. If autonomous vehicles eventually replace a significant fraction of these jobs, the displacement will be concentrated in communities where driving is one of the few accessible pathways to middle-class income. The industry’s standard response — that new jobs will be created in fleet management, remote monitoring, and vehicle maintenance — is partially true but doesn’t address the scale or timing of the disruption.
San Francisco and Phoenix, the two cities with the most autonomous vehicle deployments, have seen protests from taxi and ride-hail drivers who view the technology as an existential threat. The labor tension is manageable at current scale (thousands of robotaxis versus hundreds of thousands of human drivers) but would become politically explosive if autonomous vehicles started capturing meaningful market share. The robotaxi industry’s social license to operate may depend less on safety statistics than on how it handles the transition for the workers it displaces. That conversation hasn’t started in earnest. It will.
