Uber rival inDrive eyes 2027 robotaxi tests in Central Asia with Pony.ai and Nvidia

inDrive, the ride-hailing app built around passengers and drivers haggling over fares, has signed a memorandum of understanding with Pony.ai and Nvidia to explore robotaxi testing in Kazakhstan. The plan, per inDrive's CTO, is to start in the capital, Astana, in the first half of 2027 with a safety operator on board, with fully driverless rides as the eventual goal.

This matters because the robotaxi story has mostly been told in San Francisco, Phoenix, Beijing and Wuhan. Here's a budget-first platform in a cold-weather, emerging market saying it wants in. If autonomy only works where fares are high, it stays a niche. If it can compete on cost in places like Astana, it becomes a global labor story.

We think this is a smart, honest bet, and we want to be clear about what it is: a memorandum of understanding, not a fleet. inDrive calls its approach "capital-light by design." Pony.ai brings its Gen-7 vehicles, Nvidia brings the Hyperion platform and Halos safety system, and inDrive brings the app and local operations. That's the same modular playbook Uber is running in London and the Middle East. The ride-hailing apps are becoming distributors for robotaxi makers, not builders.

The interesting part is the economics. Nvidia's automotive chief argued that a ready-made sensor-and-compute foundation lowers development costs, and that matters most in emerging markets, where rides are already cheap and autonomy has to compete on price. We buy the logic. But it cuts both ways. In a market where a human driver costs relatively little, the case for an expensive robotaxi is thinner, and the first people it threatens are drivers who often depend on these platforms for income. If the cost curve really bends, that pressure arrives faster in exactly the places with the thinnest safety nets.

Then there's the weather. Astana has months of snow and ice, and inDrive's CEO is openly pitching it as a winter proving ground. We like that candor; most robotaxi programs grew up in sunshine. Snow-covered lane markings, ice and low visibility are where autonomy has the most to prove. A credible winter record would be genuinely useful data for the whole industry. A bad one, early and public, could set the local program back years.

I'm also watching inDrive's track record on follow-through. Its December agreement with Saudi Arabia's AI Driver, using WeRide vehicles in Riyadh, targeted a pilot in the first half of 2026, and it's still unclear whether testing has started. MoUs are cheap. Kazakhstan's government has its own 2027 autonomous taxi plans and rules for testing on some public roads, which helps. Regulators that want this to happen are the single biggest variable inDrive itself names.

Our read: useful technology, smart partner choices, right instinct to test in hard conditions. The proof is a safety-operator pilot that actually runs on schedule, with transparent incident reporting, and a plan for the drivers who built inDrive's business in the first place.

Pony.ai says it aims to have more than 3,500 robotaxis in over 20 cities worldwide by the end of 2026, up from 1,975 vehicles at the end of June. Earlier this month Uber announced plans to test Pony.ai robotaxis in London, where it already offers supervised rides with Wayve. inDrive operates in more than 1,200 cities across 47 countries and has expanded into delivery, intercity travel and financial services. The announcement coincided with Nvidia CEO Jensen Huang's first visit to Kazakhstan.