Pony AI

wrote a post ·

Pony AI Q1 Earnings Call: Robotaxi Revenue Surges 395% Year-over-Year; 2026 Fleet Target Raised to Over 3,500 Vehicles

On May 26, Pony AI ( $Pony AI (PONY.US)$ $PONY-W (02026.HK)$ ) released its Q1 2026 financial results. In the first quarter, Pony AI reported total revenue of RMB 236 million (USD 34.25 million), up 145% year-over-year. Robotaxi business revenue reached RMB 59.12 million (USD 8.57 million), soaring 395% year-over-year, while passenger fare revenue jumped 456% year-over-year, marking a record-high quarterly Robotaxi revenue. Gross profit totaled RMB 38.36 million, an increase of 140.1% year-over-year. As of March 31, 2026, the company held RMB 9.902 billion (USD 1.4 billion) in cash and cash equivalents, plus short- and long-term investment products.

Conference Call Highlights:

Robotaxi operations continue to serve as the company's core growth engine. As of May 2026, the global Robotaxi fleet has exceeded 1,700 vehicles; registered users in China have more than tripled year-over-year; and average weekly paid orders in May 2026 doubled compared to January, significantly outpacing industry-average growth rates—demonstrating clear improvements in fleet supply capacity, user travel demand, and overall operational efficiency.

Buoyed by strong commercial performance in Q1 2026, the company has raised its full-year operational targets: Robotaxi revenue is now expected to exceed 3.5 times that of 2025, up from the previous target of 3 times; the year-end 2026 Robotaxi fleet size goal has been increased from 3,000 to over 3,500 vehicles; and service coverage will expand to more than 20 cities globally.

The dual-engine strategy continues to drive expansion, with a focus on both domestic Chinese and overseas markets. The Chinese market remains the primary growth engine, with Pony AI deepening its presence in high-value, high-complexity core operational zones in Tier-1 cities. In Guangzhou, operations have expanded from Nansha and Panyu districts into the central Haizhu district, covering key travel destinations such as Canton Tower, Pazhou CBD, and the Canton Fair Complex. In Shenzhen, fleet deployment and operational density are being further intensified in core commercial and commuting areas like Nanshan and Bao’an. Additionally, Pony AI offers airport shuttle services in Beijing, Shenzhen, and Guangzhou.

Although Q1 is traditionally a slow season for ride-hailing, key metrics—including Robotaxi fleet size, user base, and paid orders—have grown month-over-month since the beginning of the year, achieving counter-cyclical momentum. The company continues to penetrate high-value, complex urban areas in Tier-1 cities. Despite Robotaxi pricing remaining higher than entry-level ride-hailing services, user demand remains robust, making it the preferred mobility choice.

Overseas expansion is progressing steadily, with the pace of global operations accelerating. Pony AI currently operates in nine countries worldwide and has launched commercial public ride-hailing services in four key international markets: Croatia, Qatar, Singapore, and South Korea. Notably, the company partnered locally to launch Europe’s first commercial Robotaxi service in Zagreb, the capital of Croatia, and continues to deepen its footprint in the Middle East by expanding operations in Dubai and Qatar.

The 'co-built fleet model' has begun generating diverse and substantial revenue streams. Leveraging its leading autonomous driving technology, operational systems, and business model, Pony AI has attracted numerous high-quality domestic and international partners to jointly build and deploy Robotaxi fleets. This innovative approach not only enhances capital efficiency in scaling the fleet but also establishes a diversified revenue structure.

Pony AI’s per-vehicle Robotaxi operating costs are globally competitive. The company expects the bill-of-materials (BOM) cost for Robotaxi vehicles in China to fall below RMB 230,000 by 2027, further refining its profitability model. As the fleet scales, the dual levers of declining operating costs and optimized BOM costs will significantly boost Robotaxi margins.

Pony AI’s three core technical capabilities are: an advanced training paradigm, a fail-operational redundant architecture, and safe, efficient fleet management. Large-scale deployment of L4 autonomous driving cannot rely solely on accumulating conventional driving data or expanding large-model parameters—it must be grounded in reinforcement learning and world-model-based technical approaches to meet intelligent decision-making demands in complex traffic conditions and edge-case scenarios.

Ongoing technological refinement and a robust safety framework ensure sustainable commercial deployment. Pony AI’s Robotaxi features a fully integrated, multi-layer hardware-software redundant architecture that enables fail-operational capability, preventing traffic congestion or rear-end collisions. The vehicles do not depend on high-definition maps or GNSS signals and can dynamically adjust driving decisions based on real-time road conditions. Additionally, a three-tier safety operations defense system—comprising prevention, detection, and response—technologically safeguards large-scale operations.

Robotruck and Intelligent Solutions businesses (formerly 'Technology Licensing and Applications') have also delivered solid growth. In Q1 2026, Robotruck service revenue rose 31% year-over-year to RMB 70.33 million, while Intelligent Solutions revenue surged 246% year-over-year to RMB 107 million. This growth was primarily driven by a sharp increase in sales of autonomous driving domain controllers, which exceeded six times the volume of the same period last year.

Q&A Session:

Jefferies, Lei Xiaoyi:

We’ve observed a significant number of new policy developments in the Robotaxi sector, both in China and overseas. We’d appreciate your perspective on this evolving regulatory landscape. More importantly, how do you see these changes impacting Pony AI’s future business development or competitive positioning? Thank you.

Dr. Peng Jun:

To my knowledge, current policy discussions—both domestically and internationally—are primarily centered around the safe operation of Robotaxis. As everyone knows, safety is the cornerstone of the autonomous driving industry. Therefore, I believe that discussions around safety, the resulting standardized safety measures, and even higher safety benchmarks will benefit the industry’s long-term, stable growth. At Pony AI, we have years of experience successfully operating large-scale fleets and have maintained strong collaborative relationships with regulators to foster a healthier and more transparent industry environment. Particularly in China, we’ve built deep trust with regulators and will continue working closely with them to ensure autonomous driving can safely serve public transportation needs.

Returning to safety itself—as Tiancheng mentioned—we’ve established a comprehensive, end-to-end safety management system covering both the autonomous vehicles themselves and fleet operations. Each of our vehicles features a fully redundant architecture with fail-operational capability, meaning that even in the event of an extreme system failure, the vehicle can still safely pull over. Additionally, our fleet management system can detect and respond to any unexpected situations on the road, functioning like a citywide safety net that prevents traffic congestion and handles real-time changes in road conditions. This forms the critical foundation for our ability to achieve safe, large-scale operations.

Our highly mature and robust safety systems, combined with our strong safety track record, give us confidence to rapidly scale our business. I believe the current policy discussions and updates will not have any direct short-term impact on our operations. On the contrary—as I mentioned earlier—we have raised our full-year 2026 business targets and are continuing to advance the deployment of our seventh-generation (Gen-7) vehicles. We are making solid progress toward our goals in fleet size, revenue, and operational footprint expansion.

As I just noted, there is currently no immediate short-term impact from these policy shifts. I believe that, in the medium to long term, the trend toward standardized regulations will particularly benefit companies like ours that have already established leadership in the industry. This also underscores the complexity of scaling Robotaxi operations in high-density urban environments—a core scenario where we continuously validate our capabilities. Ultimately, these higher standards will drive market consolidation, eliminate players lacking the requisite capabilities, and further raise the barrier to entry for new participants, thereby supporting the industry’s long-term growth.

Bank of America Securities, Lee Myung-hoon:

Given that you’ve raised your year-end Robotaxi fleet target to 3,500 units and also increased your revenue outlook, could you elaborate on the key drivers behind these upward revisions?

Dr. Wang Haojun:

This upward revision of guidance undoubtedly reflects our strong confidence in the robust business momentum we’re seeing, particularly driven by our first-quarter performance. Frankly speaking, the actual pace of progress has exceeded our expectations, as evidenced across multiple core areas of our Robotaxi business. For instance, we’re witnessing accelerating growth in our domestic operations. We observe sustained increases in revenue, paid order volume, and user base across all Tier-1 cities in China. This demonstrates our ability to consistently deliver high-quality service in Shenzhen and Guangzhou, attracting an increasing number of repeat customers—thanks to our stable and reliable performance even during peak hours and in complex traffic scenarios, which ultimately translates into higher revenues.

Another key milestone is that we’ve achieved positive unit economics per vehicle in Guangzhou and Shenzhen. This success case validates the potential for broader scalability. As a result, we’re seeing strong interest from numerous prospective partners—both domestically and internationally—in joining our 'co-built fleet' business model. This approach not only enables more efficient capital utilization but also allows us to deploy more vehicles across different markets. Given all these developments and encouraging progress, we are confident in raising our Robotaxi revenue growth target to 3.5x by 2025 and setting our year-end fleet size target at 3,500 vehicles.

Wei Huang, Deutsche Bank:

Regarding your Robotruck business: Your company just unveiled an L4 autonomous light-duty truck at the Beijing Auto Show. Could you elaborate on the strategic rationale and expansion plans behind this launch?

Dr. Peng Jun:

As many of you know, since our inception, Pony AI’s vision has always been 'to make autonomous driving accessible.' For us, 'accessible' carries two meanings: first, expanding our footprint across both domestic and international markets; and second, extending our autonomous driving technology to diverse vehicle platforms, covering a wide range of use cases including passenger transport and freight logistics. Therefore, the launch of our L4 autonomous light-duty truck aligns closely with this vision and our long-term ambitions.

In the logistics sector, the value chain spans long-haul trunk transportation, urban distribution, and last-mile delivery. We have already established a dedicated Robotruck division focused on trunk-line logistics. While we are not directly involved in last-mile delivery, we have become a leading supplier of Autonomous Driving Controllers (ADCs). The recent launch of our L4 light-duty truck is specifically aimed at filling a critical gap in our comprehensive logistics strategy. The L4 light-truck platform shares nearly identical software architecture with our Robotaxi fleet and leverages our existing operational infrastructure—including remote assistance, ground support networks, and even cleaning and charging facilities. This unified architecture generates powerful synergies: it can reduce operating costs for our autonomous light trucks by half compared to human-driven fleets, and because we share substantial backend support resources, it also lowers the day-to-day operational expenses of our Robotaxi services.

In terms of current progress, development of the L4 light-duty truck is advancing steadily. For example, we co-developed this L4 battery-electric light truck with CATL and are now establishing solid collaboration agreements with several leading logistics companies for its future deployment. Additionally, we have begun discussions with regulators regarding licensing and fleet management. We expect to begin scaling operations of this autonomous light truck in early next year.

Ting Song, Goldman Sachs:

My question is technical: regarding VLA (Vision-Language-Action) models in the autonomous driving domain, could you please share Pony AI’s strategy and future technological roadmap? We’ve noticed that recently some supply chain participants have started removing the language component from their models. Do you still consider the 'language' component necessary at this stage? Thank you.

Dr. Tiancheng Lou:

First, the essence of driving lies in understanding the intentions of other road users and responding appropriately. By incorporating an 'intention layer' into our world model training, we can generate various combinations of intentions. We evaluate the probabilities of all other road participants’ behaviors. This design ensures that our on-vehicle model always selects the safest trajectory and has contingency plans for any scenario—even extremely low-probability edge cases. We believe language is not essential to driving. Moreover, language models are too computationally expensive for on-vehicle deployment. Instead, we believe 'intention' is the true core of driving. When humans drive, they think about other vehicles’ intentions—not natural language. Crucially, this intention data is difficult to obtain through simple road testing alone; we must generate it via a world model.

We believe large language models or language layers provide no practical benefit for on-vehicle inference, whereas world models and synthetic data generation are indispensable for training. In fact, autonomous driving and large language models perform fundamentally different tasks. A large language model agent—such as a code-writing tool—does not require ultra-low latency, nor does it need to be perfect on the first attempt. It operates in a low-cost environment where it can experiment, fail, and fix errors within a sandbox.

Driving, however, leaves no room for error—any mistake results in an accident. Therefore, our tolerance for AI hallucinations is zero. To address this, we’ve built a virtual driving environment within our world model. This allows the system to experiment and make mistakes during training—not on real roads. During real-world on-vehicle inference, our model doesn’t simply select the single most probable trajectory. Instead, it chooses actions that guarantee safety under any probability scenario.

Ming Cong, Citi:

With your upward revision of the full-year target, how do you plan to balance sustained high growth with increasing strategic investments?

Dr. Haojun Wang:

Yes, we delivered outstanding results in Q1, demonstrating that our Robotaxi commercialization strategy—the dual-engine strategy—is effectively translating into accelerated revenue growth. As you’ve seen, our revenue growth has outpaced expense growth, significantly narrowing our operating loss margin this quarter.

Given this strong momentum, we are confident in raising our full-year business targets to achieve a higher growth trajectory—which we believe is critical for any growth-stage company. At the same time, to maintain our leadership position in the industry, we must strategically increase investments in certain areas. For example, through R&D efforts and deeper collaboration with OEMs, we are on track to reduce the total BOM cost per vehicle in the domestic market to below RMB 230,000 by mid-next year. We believe these investments will yield returns in future deployments and attract more partners to our co-investment fleet business model. Thus, I view this as a balancing act between spending/investment and growth trajectory—we clearly prioritize growth trajectory above all, but for these upfront expenditures, we remain firmly value-driven and committed to disciplined execution.

He Pianpian, Huatai Securities:

I’d like to focus on the overseas expansion strategy. Given that we’ve already seen some commercial progress in international markets recently, I’d appreciate more insight into the company’s overall roadmap for global fleet expansion. Specifically, when evaluating different markets such as the Middle East, Europe, and Asia, what factors determine the company’s regional prioritization? Thank you.

Dr. Peng Jun:

As I mentioned in my previous response, our company vision is ‘making autonomous driving accessible to everyone,’ and global expansion has always been a key component of our strategy. Our dual-engine strategy is accelerating this global rollout. With an increasing number of countries enacting regulations supportive of autonomous driving and more international partners seeking collaboration with us, these two factors together are creating significant growth opportunities overseas. In fact, several international markets have already started contributing meaningful revenue in the first quarter.

We are seizing the current window of opportunity because our extensive experience in technology development and commercial operations in China’s tier-1 cities has equipped us to handle the most complex urban scenarios. Moreover, we’ve already achieved positive per-vehicle profitability in Shenzhen and Guangzhou. This technological validation, combined with our cost advantages and continuously opening regulatory environments overseas, forms a strong foundation for accelerating our global expansion.

Regarding our international footprint—as you noted—we’re rapidly expanding across the key regions mentioned. We currently operate in nine countries and have launched public Robotaxi services in four overseas markets: Croatia, Qatar, Singapore, and South Korea. In Europe, we partnered with Uber and Verne to launch the region’s first commercial Robotaxi service in Zagreb. In the Middle East, we are rolling out paid services in Doha and initiating fully driverless operations in Dubai. In Asia, we’ve deployed public Robotaxi services in Singapore and are conducting intensive testing in Seoul, South Korea.

Going forward, we will continue working closely with local regulators and trusted partners to accelerate commercialization. We will double down on investment and spare no effort to achieve our goal of expanding operations to over 20 cities worldwide by the end of this year.

Sean Hsiao, Macquarie:

The earnings report mentioned that part of the capital expenditure in the first quarter was allocated to inventory stocking of autonomous driving kits (ADKs). Could you clarify whether any recent component cost increases have materially impacted overall investment costs? Additionally, Leo previously mentioned that we remain on track to meet our bill-of-materials (BOM) cost reduction targets, aiming to lower BOM costs to around RMB 230,000 next year. Could you elaborate on which specific areas or components we are primarily focusing on to achieve this target? Thank you.

Dr. Wang Haojun:

Regarding BOM cost reduction, we have consistently pursued a holistic, end-to-end and systematic cost-down strategy. In other words, we optimize comprehensively across every component of the vehicle and autonomous driving-related hardware to continuously lower overall BOM costs. Therefore, I believe several key factors will primarily drive further BOM cost reductions in the future.

First, the continued expansion of vehicle scale. As our fleet size keeps growing—especially with more 'co-built fleet' partners joining—we can place larger-volume procurement orders with our supply chain. This significantly strengthens our bargaining power with suppliers, thereby reducing costs.

Second, our seventh-generation vehicles have officially hit the road and have already accumulated millions of kilometers of real-world driving data. This data helps us more clearly identify parts of the system that can be further optimized, simplified, or even re-architected. Based on these real-world insights, we can continue iterative R&D efforts to further reduce BOM costs.

Of course, the supply chain still faces certain uncertainties. However, Pony AI has effectively managed these fluctuations for many years. For example, memory chips experienced some supply tightness this year, but we proactively secured supply for critical memory components as early as last year. This demonstrates our strong capabilities in supply chain management. Therefore, we are confident in achieving our targeted BOM cost by mid-next year.

45K Views