Chinese autonomous driving company Pony.ai is building an overseas robotaxi pipeline of more than 4,000 vehicles as it accelerates international expansion, seeking to turn advances in China’s highly competitive autonomous driving market into a global commercial opportunity.

The company said on Tuesday that the vehicles have already been contracted, although deployment schedules will depend on regulatory approvals, operating permits and other local requirements. Pony.ai did not disclose how many of the vehicles are already operating outside China or provide a timeline for completing the deployments.

The expansion comes as Chinese robotaxi companies look abroad for growth. Competition in their home market is intensifying, while regulatory and technological progress is opening opportunities in the Middle East, Europe and other Asian markets.

Pony.ai is targeting a global robotaxi fleet of 3,500 vehicles by the end of the year.

Europe Emerges As A Major Battleground

Pony.ai last week announced plans to deploy more than 2,000 robotaxis across Europe through an expanded partnership with Uber, marking one of its most significant overseas expansion initiatives.

The broader pipeline of more than 4,000 vehicles indicates that the Uber deployment is only part of the company’s international strategy.

Chief Executive James Peng said during an earnings call that all of the vehicles in the overseas pipeline had already been contracted, but their rollout would depend on securing the necessary permits and meeting operational requirements in individual markets.

The approach highlights one of the biggest constraints facing autonomous driving companies: having the technology is only one part of commercialisation. Companies must also navigate different rules governing autonomous vehicles, passenger safety, insurance, mapping, data and liability across jurisdictions.

Pony.ai’s international ambitions are being supported by rapid growth in its robotaxi business. The company reported second-quarter revenue of $36.2 million, an increase of 68.8% from a year earlier. Its net loss narrowed 14.9% to $45.4 million.

The strongest growth came from robotaxis. Robotaxi revenue surged 691.2% year-on-year and accounted for about one-third of total company revenue for the first time.

Pony.ai said it remains on track to generate more than 3.5 times its 2025 robotaxi revenue this year.

“We will continue to advance our full-year plans and are confident in our ability to exceed our full-year robotaxi services revenue target,” Peng said.

The numbers point to an important shift for the company. Robotaxis are moving from a technology demonstration and pilot project toward a meaningful commercial business, although Pony.ai remains loss-making as it spends heavily on autonomous-driving technology and fleet deployment.

Pony.ai is not alone in pursuing overseas markets.

Rival WeRide is also examining expansion opportunities in Australia, South Korea, Japan and Southeast Asia. The international push reflects a broader strategy among Chinese autonomous-driving companies to diversify their revenue sources as competition at home intensifies.

China has become one of the world’s most active markets for autonomous driving, giving local companies access to large passenger and vehicle datasets, dense urban environments and an increasingly mature electric-vehicle ecosystem.

That domestic scale can help companies improve their technology and lower operating costs. But it also means several companies are competing for partnerships, passengers and regulatory approvals in the same market.

International expansion offers a way to increase the addressable market and establish commercial relationships before competitors secure key territories.

Robotrucks Provide Another Growth Avenue

Pony.ai is also expanding beyond passenger transportation. Revenue from its robotruck business increased 40% year-on-year in the second quarter, as the company develops autonomous heavy-duty and light-duty vehicles for commercial applications.

The company aims to deploy between 500 and 1,000 autonomous heavy-duty trucks in China within two to three years. It also plans to scale its driverless light-duty truck fleet to 100,000 vehicles by 2030, according to its robotruck chief, He Xing.

The robotruck market could ultimately provide a different commercial model from robotaxis. Freight operators can potentially benefit from autonomous vehicles through higher utilization, lower labour costs and more predictable logistics operations, making the economics potentially attractive even before fully autonomous passenger transport becomes widespread.

Pony.ai’s 4,000-vehicle pipeline should therefore be viewed as a measure of contracted demand and expansion potential rather than an indication that thousands of robotaxis will immediately begin carrying passengers.

But each deployment must clear local regulatory and operational hurdles, and autonomous driving rules differ significantly between countries.

For Pony.ai and its rivals, the next phase of competition will consequently involve more than developing capable autonomous systems. Securing government approvals, local partners, fleet infrastructure and commercial contracts will be equally important.

If Pony.ai can convert a substantial portion of its overseas pipeline into operating fleets, the company would gain an important foothold in markets outside China while accelerating the transition of robotaxis from experimental services into a global transportation business.

Perplexity’s year-long giveaway of its premium AI service to millions of Airtel customers is emerging as an important early test of whether AI companies can turn free access into durable usage and, eventually, paying customers in India.

The experiment generated a huge increase in Perplexity’s reach, according to TechCrunch. But as the first free subscriptions expire, the data presents a more complicated picture: downloads have collapsed, while user activity and revenue remain well above pre-promotion levels.

That distinction could be important for an AI industry increasingly using free trials, discounted plans and telecom partnerships to accelerate adoption in price-sensitive markets.

Perplexity began the experiment in July 2025 through a partnership with Airtel, India’s second-largest telecom operator. Airtel’s roughly 360 million customers were offered a 12-month subscription to Perplexity Pro, which normally costs about $200 a year.

New redemptions ended on January 16, but customers retained access for a year from the date they activated the offer. The earliest participants therefore began reaching the end of their free subscriptions in July and faced renewal charges unless they canceled.

The promotion immediately transformed Perplexity’s position in India.

According to Sensor Tower data shared with TechCrunch, the company’s app recorded 5.9 million downloads in India in July 2025, a 625% increase from June. The figure was also greater than the 5.4 million downloads Perplexity had accumulated during the entire first half of 2025.

The growth continued well beyond the initial launch.

Perplexity recorded an estimated 56 million downloads in India during the seven months in which new customers could claim the Airtel offer, more than nine times the number recorded during the preceding seven-month period.

Monthly active users more than doubled to 8.9 million in July and eventually reached about 22 million in October.

The Free Users Did Not Simply Disappear

The more revealing part of the experiment began after the promotional window closed. Perplexity’s Indian downloads fell sharply once new customers could no longer claim the free subscription. Sensor Tower estimates that the company recorded 3.3 million downloads between February and July, a decline of more than 90% from the previous six months.

Yet usage remained considerably higher than before the Airtel partnership.

Monthly active users stood at nearly 14 million in July, down 37% from the October peak but still more than five times the roughly 2.6 million monthly users Perplexity averaged during the first half of 2025.

“While the time-sensitive nature of this promotion would naturally lead to a decline in adoption after the offer period, ongoing usage has remained resilient,” said Abe Yousef, senior insights analyst at Sensor Tower.

Perplexity, he said, now has “significantly more users” in India than it did before the promotion.

That suggests the giveaway may have done something more valuable than simply generate a temporary download spike. It appears to have introduced millions of Indian consumers to Perplexity and created a larger installed user base that persisted after the promotion ended.

The bigger question is whether that user base can be monetized.

Revenue Is Rising As Downloads Fall

Early revenue data offers an encouraging signal for Perplexity. Sensor Tower estimates that the company’s Indian in-app purchase and subscription revenue between February and mid-August increased about 60% compared with the period when new users were still able to claim the Airtel offer.

The increase has continued as the first free subscriptions have begun expiring.

Between July 18 and August 12, Perplexity’s average daily in-app purchase revenue in India was 9% higher than during the preceding 30 days and 27% above the average recorded during the first half of 2026. That creates an important divergence: Perplexity is acquiring far fewer new users, but the users already on the platform appear to be generating more revenue.

However, the figures do not establish that former Airtel users are converting to paid Perplexity Pro subscribers. The free subscriptions were configured to renew automatically, meaning users who did not want to continue had to cancel before their renewal dates. Some of the increase in revenue could therefore come from customers who simply failed to cancel.

Sensor Tower cannot distinguish between those users and customers who deliberately decided to pay after their free year ended. It also cannot isolate former Airtel subscribers from Perplexity’s other paying customers in India.

That makes the coming months more important than the initial revenue increase.

Data from another app intelligence company, Appfigures, suggests the Airtel partnership produced growth that was specific to Perplexity rather than simply reflecting a broader surge in demand for AI applications.

“I compared Perplexity’s downloads to ChatGPT and Claude to ensure it wasn’t more appetite for AI, and it wasn’t,” said Ariel Michaeli, co-founder and CEO of Appfigures.

Before the promotion, Perplexity averaged about 11,200 daily downloads in India during the week before its launch. That increased roughly twentyfold to nearly 223,000 downloads per day during the first week of the promotion.

The pace continued increasing, reaching approximately 305,000 daily downloads between mid-September and mid-October.

Downloads of OpenAI’s ChatGPT and Anthropic’s Claude remained broadly stable over the same period, according to Appfigures.

Appfigures also estimates that Perplexity’s monthly net mobile revenue in India increased from approximately $34,000 in January 2025 to $70,000 in December and $156,000 in July 2026.

The company generated an estimated $878,000 in India during the first seven months of 2026, 16% more than it generated during the whole of 2025.

Michaeli cautioned, however, that the increase cannot automatically be attributed to Airtel users converting into paying customers. The enormous visibility created by the promotion could also have attracted paying customers who never participated in the giveaway.

India Has Become The AI Industry’s Growth Laboratory

The significance of Perplexity’s experiment extends beyond one company. India has become the world’s largest market for generative AI app downloads, helped by its huge population, more than a billion internet users, relatively inexpensive mobile data, and a smartphone market of more than 700 million users.

But converting that enormous audience into revenue is considerably harder. Consumers in India are generally more price-sensitive than users in wealthier markets, making the country attractive for companies willing to subsidize access in exchange for scale and long-term user acquisition.

Perplexity’s Airtel experiment has since been followed by even greater promotional efforts.

OpenAI made its lower-priced ChatGPT Go plan available free for a year in India in August 2025. Google subsequently partnered with Reliance Jio to provide eligible customers with free access to its AI Pro subscription for 18 months.

The emerging model is therefore becoming increasingly clear: telecom operators provide distribution at enormous scale, while AI companies absorb much of the initial cost of acquiring users.

The bet is that once consumers incorporate an AI assistant into their daily routines, some will eventually become willing to pay for continued access.

That is a very different proposition from simply maximizing downloads.

The Real Test Starts Now

Perplexity’s Airtel deal appears to have succeeded at the first stage of the experiment. It dramatically increased awareness, downloads, and active users and appears to have left the company with a substantially larger Indian user base than before.

The second stage is considerably harder.

The company must determine how many of those users are willing to pay when they lose access to the free premium service. That conversion rate will determine whether the billions of dollars potentially spent by AI companies subsidizing users in India represent customer acquisition investments or simply expensive giveaways.

There is also a retention question. Users who became accustomed to Perplexity Pro for a year may continue using the service, but they have alternatives. ChatGPT, Claude, Gemini, and other AI services are competing aggressively for the same users, often with their own free tiers and promotional offers.

That makes habit formation crucial. If consumers use Perplexity because it has become part of their regular search and information workflow, a meaningful share may be prepared to pay. If they primarily used it because Airtel made the premium product free, conversion could be much weaker.

The Airtel cohort will provide one of the clearest early indicators of whether AI companies can successfully move from subsidized distribution to sustainable consumer revenue.

For Perplexity, the most encouraging signal so far is not the 56 million downloads generated during the promotion. It is that nearly 14 million people were still actively using the service in July, even after downloads had collapsed.

But the decisive metric will be the number of those users who continue paying after their free year ends.

The experiment has therefore moved from an acquisition story to a monetization test. Over the coming months, as successive groups of Airtel customers reach their renewal dates, the industry will get a much clearer picture of whether India’s enormous appetite for AI can finally translate into the recurring revenue that AI companies need.