Pony ai cashes in on chip gamble while promising robotaxis it still can't run at a profit
Pony ai wants you to believe it just cracked the code on profitable driverless rides. The headline it fed investors Thursday says otherwise: every cent of its first “profitable” quarter came from a Chinese semiconductor lottery ticket, not from moving a single passenger without a human at the wheel.
The company booked $75.5 million in net income for the final three months of last year, trimming annual losses to $76.8 million. Revenue grew 20 % to $90 million. Wall Street cheered. Then the footnotes landed. The windfall traces back to an early stake in GPU designer Moore Threads, whose post-IPO pop delivered a 425 % paper gain. Strip that away and Pony’s core robotaxi unit is still bleeding, only now more quietly.
The uber alliance is a marketing patch, not a business model
Undeterred, Pony and partner Uber announced plans to launch unmanned trips in “more than 10 markets” this year. The wording is deliberate: launch, not scale, and certainly not monetize. Translation—limited geo-fenced demos designed to keep venture taps open while regulators squint at safety files.
Inside China, Pony says it finally hit break-even per vehicle in Guangzhou and Shenzhen after rolling out its seventh-generation robotaxi stack. That metric excludes R&D, headquarters burn and the cost of re-mapping every new city down to the curb paint. Export that model abroad and the math collapses. Mapping Zagreb, a city of 800 k, is already a multi-month crawl; try London or Riyadh where rain, dialects and traffic culture shred training datasets.
Meanwhile, the global scoreboard is crowded. Waymo operates paid driverless rides in 10 U.S. cities and carries real paying customers, not press crews. WeRide and Apollo Go, both Baidu tentacles, inked deals with Uber and Lyft for the Gulf and the U.K. Pony’s edge—if it ever had one—was timing in the Middle Kingdom. That advantage evaporated the moment Beijing stopped handing out unlimited test permits.
The Zagreb play is instructive. Pony, Uber and local outfit Verne promise a fixed-fare robo-taxi service, a European first. What they don’t promise is density: the fleet will start at a couple dozen vehicles, enough for ribbon-cutting selfies, too thin for network effects. Luxembourg trials follow the same playbook—small, rich, regulators who say yes. It’s expansion theater, not logistics.
CEO James Peng insists China’s municipal profits will “replicate overseas.” He leaves out the subsidies, data subsidies and state-plated 5G corridors that make Guangzhou possible. Replicate that in Austin or Zurich and the balance sheet remembers how to scream.
Investors, still high on the Moore Threads jackpot, pumped the stock after earnings. They should read the cash-flow statement. Operating activities consumed $120 million last year even after the one-time gain. At that cadence the newfound paper wealth covers barely 18 months of runway, less if global rollouts accelerate.
Pony isn’t the first self-driving unicorn to swap product milestones for financial sleight-of-hand. It is, however, the first to declare victory while admitting its cars still can’t earn their own keep. The clock starts now: either the robotaxis start generating actual fares, or the chip windfall becomes the company’s only profitable product—and that is a business model no amount of mapping can fix.