Pony.ai Is Saying Goodbye to "Burning Cash to Scale"

Robotaxi revenue grew 691%, passenger fares rose 849%, over 4,000 units locked for deployment overseas…

On August 18, Pony.ai released an "impressive report card". Data shows, its total revenue for Q2 2026 was 246 million yuan, a year-on-year increase of 68.8%; cumulative total revenue for the first half of the year was 478 million yuan, a year-on-year increase of 98.9%, approaching a doubling.

Among them, Robotaxi business revenue was 81.92 million yuan, a year-on-year increase of 691.2%, still the main growth source. Passenger fare revenue surged 849.3% year-on-year, setting a record high single-quarter increase. CEO Peng Jun emphasized: "These rapid growth rates indicate that Robotaxi continues to serve as our growth engine."

In the subsequent earnings call, Pony.ai management not only broke down the financial data but also detailed the expansion logic of the co-building model and how technology supports acceleration in the second half of the year.

Three Business Lines, Three Growth Rates

First, look at revenue.

In the first half of 2026, Pony.ai's total revenue was 478 million yuan (approximately $70.5 million), a year-on-year increase of 98.9%. Among them, Q2 single-quarter revenue was $36.2 million, a year-on-year increase of 68.8%, with acceleration compared to Q1. Regarding capital reserves, as of June 30, the company held cash equivalents and financial investment investments totaling 9.435 billion yuan (approximately $1.39 billion), although down from $1.44 billion at the end of Q1, it remains in the first echelon of the autonomous driving industry, well-stocked with ammunition.

Pony.ai Three Major Business Revenues

Looking closely at the revenue of the three business lines, their rhythms are completely different.

Robotaxi: $12.1 million, a year-on-year increase of 691.2%. Among them, passenger-paid revenue increased 8.5 times. This number is different from vehicles sold to partners; it represents real users willing to pay for the ride. The rapid growth of fares indicates that Robotaxi's commercial closed loop is moving from "technology validation" to "user payment".

Robotruck: $13.3 million, up by 40% compared to last year. This part is mainly driven by freight services in cooperation with Sinotrans. The fourth-generation autonomous driving heavy truck has been mass-produced as scheduled and commercial deployment with China Merchants Port has started at Shenzhen Mawan Port.

Intelligent Solutions: $10.8 million, a year-on-year increase of 3.9%. This segment was the double-digit growth engine in the same period last year but slowed significantly this quarter. The company attributes this to fluctuations in the delivery rhythm of the autonomous driving domain controller (ADC).

The three lines show a significant growth scissors difference, meaning Pony.ai's growth drivers are further concentrating on Robotaxi.

Then look at the cost side.

Operating expenses in the second quarter increased by 11.4% compared to the same period last year, far below the revenue growth of 68.8%; R&D expenses increased by 14.7% year-on-year, also significantly lower than revenue growth. CTO Lou Tiancheng attributed this change to the upgrade of the technology architecture: "PonyWorld 2.0 allows the company to deploy fleets in multiple countries and cities simultaneously with fewer R&D resources, without the need to increase engineering resource investments proportionally."

Pony.ai Major Financial Data

The loss numbers are also moving in a good direction.

Calculated according to US Generally Accepted Accounting Principles (GAAP), Pony.ai's net loss for Q2 2026 was $45.4 million, narrowing by nearly 15% compared to the same period last year. The operating loss rate narrowed from 285.6% a year ago to 181.5%, and the net loss rate narrowed from 248.3% to 125.2%, each improving by over 100 percentage points in one year.

Regarding cash flow, operating cash flow net outflow for this quarter was $44 million, an expansion compared to $25.4 million in the same period last year. The company explained this was mainly due to normal working capital fluctuations, including quarterly settlements of accounts payable and inventory investments brought by early stocking for fleet expansion in the second half of the year.

CFO Wang Haojun stated in the earnings call: "We will continue to maintain a prudent cash management rhythm, and the financial situation remains robust".

Co-Building Model Accelerates Robotaxi Expansion

Specifically, where does Robotaxi's 691% growth come from?

The answer lies in the synchronized expansion in two markets: penetration into high-value scenarios domestically, and the deployment of thousand-level units leveraging the co-building model overseas.

Domestically, operations are shifting from "territory testing" to "high-value scenario penetration".

Pony.ai 7th Generation Robotaxi Turns Single-Unit Profit Positive in Shenzhen

To date, Pony.ai's Robotaxi service in Guangzhou has expanded from Nansha District to four districts: Haizhu, Tianhe, Huangpu, and Panyu. The operating area in the city center has increased by over 300 square kilometers since the beginning of the year, covering a population of over 7 million. In Shenzhen, the service network extends to three major transportation hubs: Shenzhen Bao'an International Airport, Shenzhen Bay Port, and Shekou Cruise Port, covering high-value travel scenarios such as daily commuting, cross-border travel, and airport shuttle.

In terms of user scale, as of August, Pony.ai App registered users in China have broken 1.5 million, a significant increase from 1 million in March.

Peng Jun emphasized that scale expansion in first-tier cities has formed a positive loop."Expanding fleet size shortens user waiting time and improves user retention, directly translating into growth in daily revenue per vehicle. At the same time, costs are amortized at scale, and unit economics continue to improve."

But the domestic market is just one part of the story. The bigger breakthrough is in the overseas market.

On August 14, Pony.ai announced an expansion of its strategic cooperation with Uber, planning to deploy over 2,000 Robotaxi vehicles in five European cities. This is one of the largest Robotaxi deployment plans in Europe to date.

Why would Uber further choose Pony.ai?

Peng Jun answered: "Uber has been looking for autonomous driving partners with mature and reliable technology capable of large-scale application, while also requiring a cost structure that brings attractive economics. These are precisely the two things we can provide."

Pony.ai CEO Peng Jun (right) and Uber CEO Dara Khosrowshahi

He further explained that Pony.ai has achieved large-scale Robotaxi commercial operations in all first-tier cities in China, and the unit economics model in Guangzhou and Shenzhen has turned positive; at the same time, the autonomous driving kit BOM cost of the 7th generation model has decreased by 70% compared to the previous generation, and the total vehicle cost is far lower than Waymo.

"For Uber, cooperating with multiple autonomous driving enterprises is part of its strategy to become a 'global leading autonomous driving commercialization platform'", and in Peng Jun's view, Pony.ai, with mass-producible whole vehicles, a validated unit economics model, and operational experience honed in the European market, has become the most attractive technology provider in this ecosystem.

The core supporting the above overseas layout is the "co-built fleet" model.This model breaks down Robotaxi scaled operations into three roles: the technology provider (Pony.ai provides Level 4 "virtual drivers" and operational experience), the platform provider (Uber/Bolt provides user reach and mixed capacity networks), and the operations provider (local partners are responsible for fleet daily management and maintenance).

Wang Haojun explained the financial logic behind it: "We first deliver the vehicles to partners and confirm a vehicle sales revenue; once the cars are on the road, we can continue to receive a share from every order. The former is current income, the latter is the core source of future high-profit, recurring revenue."

Pony.ai Accesses ComfortDelGro Zig App

This model allows Pony.ai to expand its fleet without spending a large amount of money to buy cars. Among the newly added vehicles, nearly half have been landed through the co-building model. As of June 30, the global Robotaxi fleet has reached 1,975 vehicles, and locked overseas deployment commitments have exceeded 4,000 vehicles, with partners including Uber, Bolt, and Singapore's ComfortDelGro, among others.

Underlying Logic Supporting Business Acceleration

After understanding "financial data" and "business model", the last question is: why can Pony.ai expand rapidly in both domestic and international markets simultaneously?

Lou Tiancheng's answer is: technical efficiency. PonyWorld 2.0 has greatly reduced the marginal cost of cross-city expansion. This is the most fundamental difference from competitors.

He explained this barrier using "probability distribution": "Traffic participants' behavior patterns differ across different cities, and even different intersections. The probability of pedestrians suddenly crossing the road varies by city. The accuracy of world models lies in matching the probability distribution of the real world; 99% is not enough, 1% is not enough, it must be precise to the probability itself."

The core capability of PonyWorld 2.0 is self-evolution. AI automatically diagnoses hidden problems, generates targeted solutions, and verifies deployment feasibility. Lou Tiancheng used the example of entering a new city: "In the past, it required thousands of engineers to manually review, analyze, upgrade, retrain, and deploy validation. Now the system automatically completes these tasks, and only a few people are needed to run it."

This means entering multiple new markets simultaneously without increasing R&D resources.

In terms of operational efficiency, Lou Tiancheng also disclosed a figure: it takes only 3 employees to maintain daily operations for every 100 Robotaxi vehicles.

"The operating model of traditional taxis has always been 1:1; 100 vehicles require 100 drivers to handle all matters from charging to daily maintenance. Our vehicles can charge and park autonomously after returning to the depot, without human intervention." In Lou Tiancheng's view, this efficiency advantage directly reduces the management costs of local operations providers in the co-building model and is also the commercial foundation for partners willing to join the co-building network.

Based on the accumulation of technical efficiency and operational efficiency above, Pony.ai has given its own outlook for the three major businesses in the second half of the year:

Regarding Robotaxi, the annual goal is to achieve more than 3.5 times growth compared to 2025. Starting from Q3, the European deployment with Uber will gradually land, and vehicle delivery revenue under the co-building model will be gradually recognized.

Regarding Robotruck, the fourth-generation unmanned heavy truck has entered the mass production stage, and mixed convoy commercial operations (unmanned trucks and human-driven trucks cooperating) are being carried out with China Merchants Port at Mawan Port. The company plans to land nearly 1,000 unmanned heavy trucks in the next 2-3 years. In addition, the L4 unmanned light truck released in April this year has entered the testing phase, with the goal of landing 100,000 vehicles before 2030.

Pony.ai Released Unmanned Light Truck in April This Year

Regarding Intelligent Solutions, revenue for the first half of the year was approximately $26.3 million, with a slight year-on-year increase. The company stated that delivery fluctuations were short-term factors, and recovery is expected in the second half of the year.

Pony.ai's Q2 financial report and earnings call content point to a clear judgment: the autonomous driving industry has entered a stage where "whose business model runs faster" matters more than "whose technology is more dazzling".

In terms of technology, PonyWorld 2.0 reduces the marginal cost of cross-city expansion; in terms of business model, the co-built fleet model leverages asset-light to trigger thousand-level scale; in terms of finance, operating leverage is beginning to appear, and the loss rate is continuing to narrow.

Because, in Peng Jun's view, "Pony.ai has operational momentum, global opportunities, and financial resources to execute its four-year goal and support sustainable growth".