Austin's robotaxi fleet is growing only on paper. Since the Cybercab launch in Austin on September 3, Tesla’s numbers have consistently looked better on paper. According to Texas DMV registration data, Tesla robotaxis rose from 314 vehicles at the end of August to 420 on September 3, and further to 479 by September 22. Separately, Cybercabs alone increased from 45 at launch to 69 today. Tesla’s official “Front Line of Exaggeration” account posted again today: “Cybercab #67 joined the Austin fleet.” But the problem lies in the numbers that follow. Community trackers don’t track registered VINs—they track actual vehicles that have picked up passengers. According to Robotaxi Tracker, Austin’s 7-day active fleet peaked above 100 right after launch, then dropped to around 18 by September 19, and fell further to just 8 by the 22nd. Other receipt trackers like fsddb show similar figures. While 147 vehicles in Austin have been photographed with license plates, only 14 have been active in the past 30 days, with just 97 recorded passenger pickups. Registered: 479. Active in 30 days: 14. Active in 7 days: 8. Three different clocks measuring the same city—and they’re wildly out of sync. There was a reason for the initial spike: invite-only rides, influencers, and viral videos. During that week, reports surfaced of people seeing “20 cars in two hours.” But once that week passed, the videos dried up. It’s not that all the cars disappeared—Austin’s traffic cameras still capture Cybercabs daily. In the past 24 hours alone, they were spotted 90 times, with the most recent sighting just minutes ago. Robotaxi Radar’s data also shows a plateau around mid-September, followed by a resurgence in unoccupied Cybercab sightings on the 21st. That’s what makes it more troubling. It’s not that you don’t see gold two-seaters on the road. It’s that even when you do, they’re not carrying passengers. Empty vehicle movement, charging, waiting, and small-scale dispatches overlap—cameras stay active, but pickup logs go silent. What matters in robotaxis isn’t inventory in parking lots—it’s supply that responds to ride requests. It’s too clean a pattern to dismiss as measurement error. The spikes align precisely with the Cybercab launch week; once the event ended, the 7-day active count dropped back into single digits. In June 2025, when service officially began, Austin’s active fleet was roughly at the same level. Geofencing expanded, unmanned operations increased, and Cybercabs were added to the fleet—yet this week, the number of vehicles confirmed to have carried passengers remains low. Why increase registrations but reduce dispatches? The most plausible explanation isn’t low demand—it’s risk management. Tesla’s robotaxis operate without LiDAR—relying solely on cameras and neural networks. This is Musk’s long-promoted “vision-only” approach. Supervised FSD ends when a human intervenes. The Cybercab has no steering wheel or pedals. There’s no one inside to take over—only remote intervention remains. NHTSA cited precisely this issue in its engineering analysis: delayed warnings when cameras fail under glare, dust, or adverse weather. Remote monitoring works for a handful of vehicles—but becomes a bottleneck when scaling to hundreds simultaneously. Some of the robotaxi incidents reported in Austin during 2025–2026 occurred when autonomous driving halted and remote operators took over—only to collide with curbs or construction barriers. The pattern of a spike to over 100 active vehicles during launch week, followed by a return to single digits, mirrors this limit—not because cars were scrapped, but because only the number manageable via remote oversight was retained. Operational restrictions are equally telling. Cybercabs avoid highways and certain railroad crossings because the system lacks sufficient confidence in complex or high-speed scenarios. That’s why a 10-minute route in North Austin now takes 70 minutes. Even with 69 vehicles registered, if routes are blocked, utilization plummets. Internally, Tesla has tied this pace to software readiness. Musk stated earlier this year that deploying large-scale robotaxis before completing FSD v15 architecture would be inappropriate. AI lead Ashok Elluswamy noted that only about 40% of v15 improvements had been integrated into fleet deployment by Q2. The September launch did not occur with fully finished software.The company claims there have been no notable incidents over one million autonomous miles. While this number may serve as a basis for confidence, it could also reflect records accumulated under geofenced areas and limited fleet deployments. It’s more accurate to interpret this not as a reduction in scale due to lack of accidents, but as a deliberate limitation on scale to prevent accidents from occurring in the first place. Registrations have increased by nearly 150% in just one month, but the actual supply confirmed through rides has moved in the opposite direction. Publicly disclosed restrictions, NHTSA investigations, delays with v15, remote intervention incidents, and avoidance of highways all point in the same direction. Tesla is indeed increasing registrations, and it’s true they continue to remove Cybercaps from production lines. But to claim that Austin Robotaxis are scaling up, the number of vehicles actively operating on roads this week must also be growing. Right now, approvals and production are ahead of actual deployment, and the high cost of failure in vision-only systems is holding back active usage. Only the paperwork is growing thicker. Investing without careful consideration could lead to serious losses!!