Pony AI Inc. American Depositary Shares (PONY) Earnings

Management highlights

• Global and Domestic Robo-taxi Expansion - Domestic: The robo-taxi fleet expanded to 2,000 vehicles, on track to reach 3,500 vehicles by end of 2026. Total registered domestic users surpassed 1.5 million. Operational coverage expanded by over 300 square kilometers in Guangzhou, covering 7 million residents, and was validated in complex scenarios in Shenzhen including peak hours, extreme weather and high holiday demand. - International: Over 4,000 vehicle commitments have been secured with Uber and other overseas partners, including 2,000 robo-taxis across 5 European cities with Uber. Commercial service is now live for the general public in Singapore via ComfortDelGro's Zig app, and deployment progresses in Luxembourg with Bolt and Stellantis. - Joint Deployment Model (JDM): The asset-light JDM relies on partners to fund fleets, enabling faster scaling, lower unit costs, and higher capital efficiency. The model delivered strong Q2 revenue growth and is a repeatable engine for global expansion. • Robo-truck Progress - Gen 4 robot trucks have entered mass production and commercial operations. Fully driverless robot trucks were launched for commercial operation at Shenzhen's Marwan Port, leveraging cross-segment technology synergies from robo-taxi and long-haul robo-truck operations. - Revenue grew over 40% YoY, and management expects this momentum to strengthen in H2 2026. • Technology Innovation and Efficiency - The upgraded Pony World 2.0 AI-powered closed-loop R&D framework automatically identifies and resolves local driving scenario issues, drastically reducing engineering labor required to enter new markets. This enables simultaneous launch in multiple distinct markets with minimal engineering input, putting the 20-city end-of-year target on track. - Technological improvements have also driven operational efficiency: robot taxis can autonomously navigate to open charging spots in standard public parking lots without human intervention, greatly improving vehicle-to-staff ratios and lowering unit operating costs. • Financial Performance - Operating expense growth (9.6% YoY) was far outpaced by revenue growth (69% YoY), leading to significantly narrowing operating losses. GAAP operating loss margin improved 100+ percentage points YoY to negative 181.5%, and net loss narrowed 14.9% YoY to $45.4 million, demonstrating emerging operating leverage and economies of scale. The company held $1.39 billion in total cash and liquid investments as of end-Q2, maintaining a robust balance sheet.

Guidance

• The company remains on track to reach 3,500 robo-taxi vehicles by the end of 2026 and exceed its full-year target of 20 cities of operation by end-2026. • Management upgraded its full-year 2026 robo-taxi revenue outlook, now expecting robo-taxi revenue to exceed 3.5 times 2025's level. • The 4,000+ vehicle commitments from international partners are expected to act as a multi-year growth catalyst from 2026 onwards. • Management expects robo-truck growth momentum to persist and strengthen in H2 2026. • Capital discipline will be maintained, with the JDM framework enabling scalable expansion without proportional increases in capital intensity.

Segment performance

Pony AI reported total Q2 2026 revenue of $36.2 million, a 69% year-over-year (YoY) increase. 1. Robo-taxi segment: Revenue hit a record $12.1 million, growing 691% YoY, accounting for 33.4% of total revenue. Fare charging revenue grew 849% YoY, driven by fleet expansion into high-value downtown areas and growing momentum from the joint deployment model. 2. Robo-truck segment: Revenue grew 40% YoY to $13.3 million, representing 36.7% of total revenue. Growth was driven by increased logistics transportation revenue and successful commercial deployment at Shenzhen's Marwan Port. 3. Intelligent Solutions segment: Revenue was $10.8 million, a 4% YoY increase, accounting for 29.8% of total revenue. Growth moderated due to delivery fluctuations for domain controllers.

Risks & headwinds

No explicit material new risks or operational failures were discussed by management during the call.

Analyst Q&A

• Q: Why did Uber select Pony AI for its European autonomous driving rollout, given Uber partners with multiple autonomous driving firms globally? /

  A: Uber prioritizes partners with proven scalable reliable technology and competitive cost structures. Pony AI has demonstrated its capability by launching commercial operations across all Chinese tier-1 cities, achieved positive unit economics in Guangzhou and Shenzhen, and successfully launched Europe's first city-center commercial robo-taxi service in Zagreb with Uber. Pony AI also has the industry's most competitive total cost per mile, and shares aligned strategic culture with Uber focused on starting in high-value markets. As Uber's largest autonomous driving partner in Europe, there is significant room for further fleet expansion as performance validates at scale.

• Q: How does Pony AI's joint deployment asset-light model work, and what is Pony's strategy for the model going forward? /

  A: The model creates a win-win ecosystem where Pony provides proven AI driver technology, local mobility platforms bring user demand, and local fleet operators handle management and maintenance. For Pony, this is an asset-light structure that generates recurring, high-margin revenue from either revenue sharing or technology licensing, without requiring proportional capital investment for fleet expansion. The model leverages existing local ecosystems rather than disrupting them, and the 4,000+ vehicle commitments from global partners will drive multi-year growth starting in 2026.

• Q: What makes Pony World 2.0's self-evolution unique, how does it improve R&D efficiency, and would an open-source world model erode Pony's competitive advantage? /

  A: Unlike general-purpose open-source world models that only generate simulation data, Pony World 2.0 is purpose-built for autonomous driving, focusing on capturing the exact real-world probability distribution of traffic participant behaviors (varying by city and region) that is required for safe deployment. The old human-led R&D workflow for new markets required dozens of engineers per city, but Pony World 2.0 automates most of the process, with humans only performing final validation. This drastically reduces engineering input per new market, enabling entry into multiple new markets simultaneously without increasing R&D headcount. This scalable, data-driven moat cannot be replicated by open-source general models.

• Q: What operational efficiency improvements allow Pony to scale its robo-taxi fleet, and how do these gains support expansion? /

  A: The core operational efficiency metric for fully driverless fleets is the vehicle-to-staff ratio. Traditional human-staffed taxis require a 1:1 ratio, but Pony's technology enables fully autonomous self-parking, charging, and depot navigation with zero human assistance. This means only 3 staff are needed to manage 100 robotaxis, drastically cutting per-vehicle operating costs and improving unit economics. This low-cost, scalable operating model has been standardized, making it more attractive for ecosystem partners to join the joint deployment model, supporting faster expansion into new markets.

• Q: What is the outlook for Pony's new L4 light truck business initiative? /

  A: The L4 light truck business aligns with Pony's vision of autonomous mobility everywhere, and benefits from strong synergies with existing segments: it shares nearly 100% of robo-taxi technology and operational infrastructure, and shares the same customer base as Pony's heavy robot truck business, cutting development and operating costs. It extends Pony's logistics portfolio from long-haul to urban delivery, opening a large new market with over 8 million active light trucks on Chinese roads. The world's first automotive-grade fully redundant L4 light truck is co-developed with CATL, and Pony has already secured partnerships and deployment plans with leading logistics firms SF Express and China Post Technology. Additional partner pipelines are expected to drive scaling through the remainder of 2026.