Outlook: where robotaxis are heading
Our read of where this market goes, near and far. The direction is broadly agreed; the timing and size are genuinely uncertain, so we give a range and name the assumptions that swing it.
Read this as analysis, not a crystal ball
This is our neutral read, not a prediction and not investment advice. The market figures below are named analysts' estimates, not facts, and they disagree by more than an order of magnitude. Treat the direction as high-confidence and any specific number or date as low-confidence.
The one-line thesis
The direction is close to inevitable: robotaxis get cheaper per mile than human ride-hail, fleets scale, and personal car ownership slowly erodes. The magnitude and timing are wildly uncertain. Credible 2035 market estimates run from about $105B to roughly $1.5 trillion, a 14x spread; the two big banks (Goldman Sachs, Morgan Stanley) cluster around $300B to $415B, which is the most defensible central case. So the honest forecast is a range with named assumptions, not a single number.
Where it actually is now
Strip away the hype and mid-2026 is lopsided. Waymo is the only one at real scale, around 500,000 paid rides a week and raising fresh capital near a $126B valuation. Tesla is tiny but cheap, a driverless fleet in the low dozens on hardware that costs a fraction of Waymo's. China runs a parallel, cheaper race, with Baidu's Apollo Go doing similar volumes on roughly $28,000 vehicles. Everyone else is pre-revenue or nascent. This is not a mature field of rivals: it's one company at scale, one cheap-hardware wildcard, one cheaper country, and a long tail.
The economics that decide everything
The consensus is that robotaxis undercut human ride-hail around 2028 to 2030, because you delete the driver's roughly $0.50 to $1.50 a mile. Goldman models total cost per mile falling below $1.00 by 2035. The catch nobody has proven: those economics need very high utilization, on the order of 200 to 400 revenue miles per vehicle per day, and no fleet has sustained that at scale. Empty miles (10 to 30% of the total), plus charging, cleaning, and depot overhead, are a real drag that skeptics say keeps fares above the bull-case levels this decade. The tailwind is that vehicle cost is deflating fast, so cheaper cars fix a lot.
Nobody is profitable yet, so balance sheets decide it
Every scaled Western player is burning cash at the corporate level. Waymo sits inside an Alphabet unit that lost about $2.1B in a quarter; Tesla told investors robotaxi is not material in 2026; Pony.ai and WeRide post steady quarterly losses funded by their IPO war chests. So the real question is who can afford to lose money the longest while unit economics catch up. That reorders the field: Uber looks like the clearest financial winner precisely because it barely touches a robotaxi, staying profitable and letting partners carry the capital cost. Waymo, Tesla, and Zoox have deep-pocketed parents. The Chinese pure-plays have finite runway and are the most likely consolidation targets. One wrinkle: Uber's partners (Rivian, Lucid) are themselves losing money, a real risk to its fleet plans.
The vision-versus-LiDAR fork
Tesla bets on cameras only; everyone else uses LiDAR for redundancy. This is the single largest swing factor, and it is unresolved. If vision-only generalizes unsupervised, Tesla's manufacturing scale makes it structurally the lowest-cost player and it could win on price. If it does not (weather, edge cases), it eats a safety and regulatory penalty that erases the cost edge. Bull-Tesla and bull-Waymo forecasts assume opposite technical outcomes and cannot both be right. As of mid-2026 the LiDAR camp has the deployment and the safety record; the vision camp has the cost thesis but not yet the unsupervised-at-scale proof.
The timeline, by confidence
Near term (2026 to 2027), higher confidence: Waymo extends its lead and pushes toward a million rides a week, adding US metros and starting London and Tokyo; Tesla scales slowly rather than explosively; Zoox goes paid; China keeps compounding; and weak players get absorbed rather than new ones appearing. Regulation stays lumpy, with permissive states expanding while some blue cities and unions push back. Mid term (2028 to 2030): unit economics cross human ride-hail in leading cities, the sensor bet resolves, and a carmaker likely buys an AV stack as survivors consolidate. Long term (2030s), lower confidence: erosion of private car ownership (Morgan Stanley's 'Peak Car by 2035'), pressure on roughly 4.4 million US driving jobs (the biggest political flashpoint), and insurance shifting from drivers to fleet operators, backed by Waymo's strong safety record.
The safety tailwind and the political headwind
The strongest structural argument for robotaxis is safety data: an independent Swiss Re study of 25 million driverless Waymo miles found 88% fewer property-damage claims and 92% fewer injury claims than human drivers, the basis for cheaper AV insurance and pro-AV rules. The strongest headwind is politics and labor: New York shelved its AV plan in 2026 under union pressure, and Waymo faces organized opposition in several metros. So US rollout is lumpy and path-dependent, city by city, rather than a smooth national curve, even as the safety case strengthens.
Three honest scenarios
Base case (most defensible)
Waymo-led, methodical scaling; the economic crossover lands around 2028 to 2030; the market reaches roughly $300B to $415B by 2035; Tesla is a strong number two or better if vision works; China wins on volume and units, the US on revenue per ride.
Bull case
Vision-only generalizes and utilization hits its targets, so costs collapse and adoption goes non-linear into trillion-dollar territory well before 2035.
Bear case
Utilization disappoints, a high-profile safety failure hardens regulation, and unit economics stay underwater outside a few dense cities, leaving a slow, subsidized, city-by-city grind through the 2030s.