Market researchers Berg Insight say the global robotaxi fleet could swell from roughly 6,500 vehicles at the end of 2025 to 1.53 million by 2036 — more than 1.5 million driverless cabs on the road within a decade, if their forecast holds.
The Gothenburg-based firm’s first dedicated study of the sector, The Global Robotaxi Market – 1st Edition, puts fleet growth at a compound annual rate of 64.4 per cent. Passenger fare revenue from commercial robotaxi services is projected to jump even faster: from US$260 million in 2025 to US$158.7 billion in 2036, a 79.2 per cent CAGR — figures also reported by IoT Business News and ITWeb.
That is a research house forecast, not a booking diary. Reaching those numbers would mean turning today’s city-by-city pilots into repeatable fleets that clear safety rules, keep vehicles busy enough to pay for themselves, and win public trust.
Where services are now
Driverless robotaxi rides are still concentrated in the United States, China and the Middle East. Berg Insight notes newer launches in Europe, South Korea and Singapore, and expects more markets to follow.
For UK and European drivers, that timeline already feels closer than a distant slide deck. TaxiFeed has tracked union pressure for a levy-backed transition fund, London rows over Uber–Wayve trials, and plans such as Lucid and Bolt’s 25,000-vehicle Europe pitch. A 1.5 million global fleet by 2036 would dwarf those early deployments — if utilisation and regulation allow it.
Who sits in the middle of the stack
Berg Insight argues the automated-driving provider is the real integration layer: not a single sensor supplier, but the firm that binds perception software, sensors, compute and vehicle controls into a Level 4 system.
Senior analyst Martin Cederqvist said the automated driving system and the data used to train and validate it are the main proprietary assets. Providers named in coverage of the report include Avride, Baidu’s Apollo Go, May Mobility, Mobileye, Momenta, Motional, Pony.ai, Tesla, Waymo, Wayve, WeRide and Zoox — some also running mapping, simulation, fleet data and remote assistance.
Mobility apps sit on top of that stack. Uber, Lyft, Bolt, DiDi, CaoCao Mobility and T3 Mobility can bring bookings, payments, pricing and existing demand — which matters because a technically capable car without enough paid trips still loses money. IoT analyst Erica Rickard said platforms can use brands, user bases and travel-pattern data to match supply and demand and lift utilisation.
What has to go right for 1.5 million
Berg Insight’s own caveats read like a checklist of why drivers should treat mega-forecasts with care. Geographic scale still depends on how much of a system can be reused from one city to the next; end-to-end AI and large driving models may cut local engineering, but local data needs still vary. Purpose-built or factory-integrated robotaxis could lower costs only once volumes are large enough. Carmakers are taking more selective roles; some AV firms may launch without a big mobility partner; and driverless services still lean on human remote support.
Cederqvist’s closer, reported by ITWeb, is blunt: firms that combine AI and vehicle design, build partnerships, meet safety and regulatory tests, and invent commercially viable models will be best placed to scale. Everyone else is still running expensive experiments.
For working taxi and private hire drivers, the headline number is less important than the conditions behind it — regulation, public acceptance, and whether robotaxi fares can cover full operating cost without undercutting licensed trade. Those fights are already under way, long before any global fleet nears 1.5 million.