Pony.ai is scaling faster than expected, but the gap between its revenue trajectory and its cost base is also widening. By Stewart Burnett
Chinese robotaxi player Pony.ai has raised its 2026 robotaxi fleet target by 500 vehicles to 3,500 after reporting first-quarter revenue of US$34.3m, a 145% year-on-year increase that comfortably beat the US$21.7m average analyst estimate. Core robotaxi services revenue rose almost fivefold to US$8.6m in the period, and full-year robotaxi revenue is now forecast to exceed 3.5 times the company’s 2025 level of US$16.6m, up from a prior projection of threefold growth.
The company’s stock value in the US also rose 3.6% on the results, but remains down approximately 36% year-to-date. While a portion of this can be attributed to broader uncertainty about Chinese automotive software in the US market, the gap also reflects a lack of confidence that strong growth rates will necessarily translate into profitability for Pony.ai.
In fact, Pony.ai appears further from profitability than it did when it was operating on smaller fleet sizes. Its operating loss widened 4% to US$58.3m, indicating that aggressive deployment is expanding the cost base at a slightly faster pace than revenue is being built. At the time of writing, the company operates more than 1,700 Gen-7 vehicles with a presence in more than 20 cities spread across nine countries. It should be noted multiple of these are in exploratory or testing stages. Among its robotaxi manufacturing partners it counts Toyota and the state-owned automakers BAIC and GAC.
There is some cause for celebration despite the widening losses: earlier in 2026, the company shared that it had achieved unit-economics breakeven in both Shenzhen and Guangzhou. key Chinese markets — the more meaningful operational milestone beneath the headline figures. The company is aiming to bring Gen-7 bill-of-materials costs below CN¥230,000 (US$34,000) by mid-2027, the threshold it considers necessary for high-density commercial deployment to be economically sustainable at scale.
Global expansion serves at the core of Pony.ai’s growth ambitions. A partnership with Dubai’s Roads and Transport Authority is targeting 1,000 robotaxis in the MENA region by 2028; trials are also running across Singapore, South Korea and the UAE. An SAE Level 4 autonomous light truck launched this year extends the same underlying platform into freight and logistics, broadening the company’s addressable market without requiring a parallel hardware programme.
Still, the first-quarter results land against a complicated regulatory backdrop in China. The government paused approvals for new autonomous vehicle licences following a widespread outage that caused Baidu’s Apollo Go robotaxis to abruptly stop on streets in Wuhan in late March. This, in turn, triggered a national safety review of how operators and local authorities manage autonomous system failures.
Pony.ai’s Chief Executive James Peng confirmed that the company has completed all required evaluations without operational disruption and that expansion into additional cities is continuing. While companies like Apollo Go reckon with these new hurdles, Pony.ai has pulled off a clean passage through the review that, under the circumstances, carries weight as a competitive credential.
Internationally, Pony.ai moved beyond trials in April with the launch of Europe’s first commercial robotaxi service in Zagreb, in a tripartite arrangement with Verne and Uber, covering approximately 90 square kilometres of the Croatian capital using BAIC Arcfox vehicles. Verne is targeting expansion to 11 cities across Europe and MENA, with a further 30 under consideration. Peng singled out the UK as “a very interesting market” while stopping short of confirming any specific plans.
The markets Pony.ai is entering are becoming sharply contested as robotaxi momentum picks up. Apollo Go is targeting Berlin and London through a deal with Lyft, which recently acquired FreeNow in Europe; scores of autonomy players count Uber as their platform partner; and Waymo is conducting mapping and testing across London boroughs ahead of a planned commercial service later this year.