As autonomous vehicles start to grace our streets and inch closer to scalable profitability, does this herald a future where human drivers become obsolete?

In the realm of autonomous driving, Pony.ai has consistently made headlines with its groundbreaking achievements. After achieving its first monthly operational profit per vehicle in Guangzhou in November 2025, the company announced in February of the following year that it had replicated this feat in Shenzhen.

Consequently, Pony.ai became the first Chinese company to attain per-vehicle profitability for its Robotaxi services in two major cities. This milestone is widely regarded by industry experts and observers alike as a pivotal turning point, marking the transition of autonomous driving from a "money-draining" experimental phase to a "money-making" commercialization phase.

However, behind these impressive operational figures, the realities faced by Pony.ai and the broader driverless industry remain intricate. While per-vehicle profitability has been achieved, the path to overall corporate profitability is still long and arduous. Furthermore, amid escalating industry competition, the trust deficit surrounding driverless technology among users has yet to be fully bridged.

Profitability Signal: A Preliminary Triumph of Scale Effects

Reflecting on Pony.ai's journey, its "per-vehicle profitability" is underpinned by a robust technical foundation and strategic ecological partnerships.

Founded in 2016, the company was established in Beijing's Economic and Technological Development Zone by Peng Jun, the former chief architect of Baidu's autonomous driving division, and Lou Tiancheng, a former Baidu engineer. At the time, the autonomous driving market was rife with uncertainty, and the technology faced numerous hurdles.

Nonetheless, Peng Jun possessed a keen market insight and a precise understanding of industry trends and capital movements, while Lou Tiancheng brought deep technical expertise to the table. Together, they led a team of technical elites from tech giants like Google and Baidu, tackling challenges head-on with a shared vision of "making autonomous driving accessible."

Two years later, Pony.ai launched China's first regularly operating autonomous driving fleet in Guangzhou, introducing its Robotaxi service, "PonyPilot," which allowed passengers to book rides via a WeChat mini-program. In the ensuing years, it sequentially obtained permits for driverless testing in Beijing and Guangzhou, as well as licenses for autonomous truck testing.

By 2024, Pony.ai had secured permits for driverless mobility services in all four first-tier cities—Beijing, Shanghai, Guangzhou, and Shenzhen—signaling the normalization of its Robotaxi operations nationwide. Simultaneously, it forged strategic partnerships with GAC Aion, BAIC New Energy, and others. That same year, Pony.ai successfully listed on Nasdaq, becoming the "world's first Robotaxi stock."

Last year, Pony.ai's seventh-generation Robotaxi officially commenced operations in Guangzhou, with each vehicle averaging 23 daily orders and achieving positive per-vehicle operational profitability. Guo Yu, Pony.ai's market communications manager, stated, "This profitability level already covers vehicle depreciation, charging, and other costs."

As of February 28, 2026, the seventh-generation Robotaxi's profitability performance in Shenzhen showed an average daily net revenue of 338 yuan per vehicle, with an average of 23 daily orders.

This breakthrough was made possible by Pony.ai's concerted efforts to reduce technological costs. Through technological iteration, the cost of the autonomous driving suite for the seventh-generation Robotaxi dropped by 70% compared to the previous generation, with costs for onboard computing units and LiDAR decreasing by 80% and 68%, respectively. The company plans to further reduce costs by 20% for models slated for mass production in 2026.

Cost reductions have made per-vehicle profitability and even sustainable commercialization feasible. Additionally, Pony.ai is transitioning from a mere technology provider to building an open commercial ecosystem.

Recently, Pony.ai announced its integration with the "Tencent Mobility Services" mini-program. In Guangzhou's operational area, users can hail autonomous vehicles through this mini-program, experiencing safe and convenient travel services. The company has also deepened cooperation with OnTime, deploying over a hundred new vehicles, and partnered with Moore Threads to accelerate the large-scale deployment of autonomous driving in China.

These collaborations have effectively increased Pony.ai's order density and user reach efficiency, marking a crucial step toward scalable operations.

Financial Reality: Per-Vehicle Profitability ≠ Corporate Profitability

While "per-vehicle operational profitability" is a landmark achievement, it does not equate to overall corporate profitability for Pony.ai.

In terms of revenue alone, as of the first three quarters of 2025, Pony.ai's total revenue reached 181 million yuan, up 72% year-on-year, with revenue growth achieved for three consecutive quarters. Revenue from the Robotaxi business reached 47.7 million yuan in the third quarter, up 89.5% year-on-year, with passenger fare revenue surging by over 200%.

However, during the same period, its net loss for the third quarter of 2025 widened to $61.6 million, compared to $42.1 million in the third quarter of 2024.

The increase in net losses stems from high R&D investment and substantial management expenses.

Pony.ai plans to mass-produce over 1,000 fourth-generation Robotruks in 2026 and accelerate the deployment and commercialization of the seventh-generation Robotaxi, aiming to achieve mass production of over 3,000 Robotaxis by 2026.

This means the company must continue to increase upfront investments in R&D and market expansion, all unavoidable operational costs before achieving true profitability. From this perspective, the company's path to overall profitability remains lengthy.

Thus, the current "per-vehicle profitability" can be understood as a positive shift in the unit economic model (UE), meaning revenue exceeds direct costs at the level of a single Robotaxi, achieving profitability. It proves that positive cash flow can be generated per service after deducting vehicle depreciation and operational costs.

While the company's business model has passed this critical validation, its offline operational "profitability" currently covers only partial areas of Guangzhou and Shenzhen, two first-tier cities. Achieving comprehensive coverage of first- and second-tier cities will take considerable time. Therefore, to extend profitability to cover R&D, management, and market expansion costs company-wide, Pony.ai still has a long journey ahead.

Industry Concerns: Trust Crisis and Disorderly Competition

Beyond the challenges of achieving overall profitability, Pony.ai also faces severe internal industry challenges.

Today, the autonomous driving sector is increasingly competitive, with Pony.ai's rivals including self-developed tech companies like WeRide and Apollo Go, as well as automakers such as Tesla, Huawei, Li Auto, and XPeng Motors.

The "clash" between Pony.ai and WeRide in the second half of 2025 was one of the most talked-about events in China's autonomous driving industry.

At the time, both companies were in the critical phase of listing on the Hong Kong Stock Exchange. During Pony.ai's closed-door roadshow to investors on the Hong Kong Stock Exchange, it cited operational data claiming that competitor WeRide "only operated Robotaxis in Beijing" and had "zero orders."

In response, WeRide's CFO, Li Xuan, retorted that WeRide "has already launched commercial Robotaxi operations in Beijing, Guangzhou, and multiple locations in the Middle East, with a fleet of over 700 vehicles accumulating more than 2,200 days of public operation records, all clearly verifiable in the prospectus." Regarding order volume, she cited second-quarter 2025 data, stating that Robotaxi service revenue alone reached 45.9 million yuan, up 836.7% year-on-year, definitively not "zero transactions."

Additionally, Pony.ai hinted that WeRide's technology was a "two-stage architecture," inferior to Pony.ai's "end-to-end L4 system." Li Xuan sharply countered, stating that the company's "end-to-end solution, developed in collaboration with Bosch and Chery, has already entered mass production," while accusing Pony.ai's L2+ project of "never reaching mass production."

Notably, this was not the first direct confrontation between the two sides. In June of the previous year, Pony.ai's co-founder, Lou Tiancheng, claimed that only Waymo, Pony.ai, and Baidu were "at the table" in the industry, with other companies lagging by more than two and a half years. Subsequently, Li Xuan listed Pony.ai's past accident records on her social media, questioning its technical safety and mocking it for "claiming to be a global company after signing a bunch of MOUs."

Of course, the closed-door roadshow materials also implicated Baidu's Apollo Go, showing its operational cities as Beijing and Shanghai. As a result, Baidu's investor relations director, Lin Juan, stated on social media, "We disdain to respond to certain competitors' baseless smears in the capital market. However, as such false claims escalate, we have taken legal action to protect our rights."

These public disputes not only damaged corporate reputations but also reflected the industry's anxious mindset in vying for market share and capital favor.

While the driverless industry is booming, with autonomous driving technology moving from laboratories to the streets, another crisis lurks beneath the intense internal competition and "clashes"—user trust.

One major reason for the public's low trust and acceptance of driverless technology is safety concerns. After all, no one is willing to risk their life by fully entrusting the steering wheel to AI. Even a single severe traffic accident can erase trust accumulated over millions of kilometers of safe driving.

In July of the previous year, U.S. short-seller Grizzly Research published a report questioning Pony.ai's technical capabilities, data fabrication, and other issues, while mentioning safety incidents. Although the company denied the claims, such doubts are unlikely to disappear entirely in the public eye anytime soon.

Epilogue

Pony.ai's "per-vehicle profitability" represents a commercial turning point driven by technological breakthroughs, proving the commercial viability of L4 autonomous driving to the market and offering a glimpse of commercialization for China's autonomous driving industry.

However, Pony.ai still faces enormous challenges in transitioning from technical success to commercial success. With necessary R&D investments and profitability goals already set, cost reduction and efficiency improvement must remain a priority. Additionally, as a leading company in the fiercely competitive industry, Pony.ai should take the initiative to establish a healthy industry order.

Winning broader societal trust will be the "essential question" Pony.ai and all autonomous driving players must answer next.