170 rides, 54 cars, no monitor: a hobbyist tracker marks Tesla's Austin handover

The best public record of whether Tesla's Austin robotaxis still carry a human is a crowdsourced site run by one person, and it now says they do not: all 170 Austin Robotaxi rides logged by Robotaxi Tracker over the past two weeks were unsupervised, across 54 distinct cars, creator Ethan McKanna told The Verge ^[1][2]. Elon Musk announced that Austin Robotaxis were operating without onboard safety monitors seven months ago, and the city's service only now appears to match the claim ^[17].

Be precise about what the sample is. One hundred seventy rides over fourteen days works out to roughly twelve logged rides a day in the whole market ^[15], and about three logged rides per car over the entire two weeks ^[16], which tells you nothing useful about utilisation because the tracker sees only the rides its users report. It draws primarily on app users automatically logging their trips, plus other crowdsourced and public information ^[6]. The load-bearing number is not 170, it is 54: a census of plates is harder to fool than a ride count, and McKanna says plates previously marked as supervised "have been confirmed to be offering unsupervised rides" ^[19], with "Austin has seemingly stopped supervised rides for a couple weeks now" ^[18]. Users in the Austin area reported the same thing on Reddit and X ^[7].

The instrument also moved while the thing it measures moved. As recently as last week the same tracker showed just 28 active unsupervised vehicles across six Texas and Florida markets ^[4], so McKanna has now counted nearly twice as many unsupervised Teslas in Austin alone as his site previously showed everywhere ^[5]. Some of that is real and some is calibration: McKanna, who interned with Tesla's Robotaxi team this summer, said some of his metrics had been lagging, and after revising the data sources he found more cars running unsupervised than the site had reflected ^[11]. Anyone quoting the 28-to-54 jump as growth is quoting a mixture of fleet change and measurement change. Outside Austin, he counts roughly 30 driverless Teslas in Dallas and Houston over the past week ^[3].

The reason a third-party plate census carries this much weight is that Tesla does not publish detailed fleet data and typically does not answer press inquiries; it did not respond to The Verge for this story either ^[8]. That is a choice with a cost: the company has outsourced the verification of its most consequential capability claim to a hobbyist spreadsheet.

The timing is not incidental. Tesla is preparing to launch the Cybercab in Austin as soon as this month, and unlike the Model Ys now doing the driving, Cybercabs have no steering wheel or pedals, so removing the onboard monitor is a precondition rather than a milestone ^[9]. A JPMorgan note says Tesla is deliberately holding back on adding Model Ys to the fleet because it expects to scale Cybercab in the near term ^[14]. On the most recent earnings call Musk said the program would scale "very, very rapidly" but gave no vehicle or city target, a departure from the specific forecasts he has repeatedly missed ^[12]. Scale is where the gap still sits: Waymo runs more than 300 driverless vehicles in Austin alone and about 4,000 across ten other cities ^[10], more than five times the Austin car count the tracker logged ^[13].

Watch whether the unanimity holds as ride volume rises, whether supervised plates reappear after the first Cybercab weeks, and whether Tesla starts publishing fleet numbers itself once they are flattering.